Tobacco Documents Online

BOOK 53 TAB# 1-53 LOW TAR AND DISEASE 2063628OOO t

TABLE OF CONTENTS 2063628001

Tab # Author's Name TABLE OF CONTENTS Title 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Lee, P.N. Stellman, S.D. Peto, R. ' Koziowski, L.T. Peto, R. Benowitz, N. Winkelstein, W. Peto, R. Hoffmann, D. Horm, J.W. Schmidt, F. Holland, W.W. Int Agency Res Cancer Work Group Eval Carcino Peto, R. Kozlowski, L.T. Bross, T.D. Weisberg, R.L. Parkin, D. Brown, C.C. Frogatt, P. Stephen, A. Benowitz, N.L. Froggatt, P. Hecht, S.S. Darby, S.C. Ebi-Kryston, K.L. Jarvis, M.J. Kozlowski, L.T. Shiffmann, S.M. Harris, J.E. Speizer, F.E. Kozlowski, L.T. Lung Cancer and the 'Safer' Cigarette Tobacco (A Major International Health Hazard (Cigarette Yield and Cancer Risk (Evidence from Case-Control and Prospective Studies))) Tobacco-Related Diseases Less Hazardous Tobacco Use as a Treatment for the 'Smoking and Health' Problem Lung Cancer (Causes and Prevention (Chapter 1 Keynote Address (The Control of Lung Cancer) Is There a Future for Lower-Tar-Yield Cigarettes Some Ecological Studies of Lung Cancer and Ischaemic Heart Disease Mortality in the United States The Value of Preventive Medicine (Control of Tobacco-Related Disease) Tobacco (A Major International Health Hazard (V Smoking. Current Research Issues...)) Falling Rates of Lung Cancer in Men in the United States Smoking (The Position in the Federal German Republic) Low-Tar Cigarettes Put to the Test Tobacco Smoking (Tobacco Smoking (Epidemiology Studies of Cancer in Human (1 Introduction (2 Cancer of the Lung))))) Tobacco ( A Major International Health Hazard (IV Health Effects of Low-Tar, Low-Nicotine...)) Less Hazardous Smoking and the Pursuit of Satisfaction Crimes of Official Science (A Casebook) Smoking and Health 1987 ( A World Report (Trends in Cigarette Consumption in the USA)) Surveillance in Health and Disease (Part 2 Practical Applications of Surveillance (13 Surveillance of Cancer)) Projections of Lung Cancer Mortality in the United States... Nicotine, Smoking and the Low Tar Program Nicotine, Smoking and the Low Tar Programme (111 Smoking Yields and Consumption (8 Estimating the Extent of Compensatory Smoking)) Health and Public Policy Implications of the 'Low Yield' Cigarette Determinants of Policy on Smoking and Health The Relevance of Tobacco-Specific Nitrosamines to Human Cancer Nicotine, Smoking, and the Low Tar Programme (11 Smoking Habits and Related Mortality in the UK...)) Predicting 15 Year Chronic Bronchitis Mortality in the Whitehall Study Comment on the Hunter Committee's Second Report Have Tar and Nicotine Yields of Cigarettes Changed A Safe Cigarette (Session 5 Behavioral and Economic Issues (Diminished Smoking, Withdrawal Symptoms, and Cessation))) A Safe Cigarette (Session 5: Behavioral and Economic Issues (Public Policy Issues in the Promotion of Less Hazardous...) Epidemiology of Respiratory Diseases (Task Force Report (Smoking)) Smokers, Non-Smokers, and Low-Tar Smoke

33 34 35 36 37 38 39 4O 41 42 43 44 45 46 47 48 49 5O 51 52 53 Holbrook, J.H. Holbrook, J.H. Kannel, W.B. Jaffe, J.H. Hirayama, T. Richmond, J.B. Hoffmann, D. Benowitz, N.L. Luoto, J. Lenfant, C. Rickert, WoS. The Changing Cigarette National Conference on Smoking and Health (Developing a Blueprint for Action) Update on the Role of Cigarette Smoking in Coronary Artery Disease Low Tar Cigarettes Flunk the Test Epidemiological Aspects of Lung Cancer in the Orient Ending the Cigarette Pandemic Human Carcinogenesis (VIII Laboratory Epidemiology Studies (33 Tobacco Carcinogenesis (Metabolic Studies in Humans))) Smokers of Low-Yield Cigarettes Do Not Consume Less Nicotine Reducing the Health Consequences of Smoking - A Progress Report Are 'Low-Yield' Cigarettes Really Safer A Comparison of the Yields of Tar, Nicotine, and Carbon Monoxide of 36 Brands of Canadian Cigarettes Tested Under Three Conditions Rickert, W.S. Wynder, E.L. Wald, N.J. Devesa, S.S. Loeb, L.A. Samet, J.M. Wald, N.J. Tomatis, L. Diamond, L. Gillis, C.R. Yields of Tar, Nicotine, and Carbon Monoxide in the Sidestream Smoke from 15 Brands of Canadian Cigarettes Demographic Aspects of the Low-Yield Cigarette (Considerations in the Evaluation of Health Risk) Relative Intakes of Tar, Nicotine, and Carbon Monoxide from Cigarettes of Different Yields Lung Cancer (Causes and Prevention (Chapter 3 Trends in Lung Cancer Incidence and Mortality in the U.S.) Smoking and Lung Cancer (An Overview) Less Hazardous Cigarettes and Disease of the Lung Cancer Risks and Prevention (3: Smoking) Tobacco (A Major International Health Hazard (Foreword)) Augmentation of Elastase-lnduced Emphysema by Cigarette Smoke (Effects of Reducing Tar and Nicotine Content) Cigarette Smoking and Male Lung Cancer in an Area of Very High Incidence (11 Report of a General Population Cohort...)

2063628004

CONTROVERSY Lung cancer and the 'safer' cigarette PETER N LEE MA Consultant in Statistics and Honorary Research Fellow, Divison of Epidemiology, Institute of Cancer Research, London years ago, when the evidence relating cigarette smoking to lung ~cer first started to appear, nearly all British smokers Smoked unripped (plain) Narettes with a tar yield of Today, more than of cigarettes smoked have and average tar yields are 15rag. Even the tar yields cigarettes have declined ~y so that hardly any garettes nowadays are above le 'middle tar' (17 to 22rag) ~nge while 'low tar' (0 to 10mg) lter cigarettes have captured rare than 15% of the market in ~last ten years. Similar trends ~e OCcurred in most developed .a developing countries (Lee, ~3). Although this suggests a ~uCtion in 1~ r'~neer c~uld l~u~(the tar ~ cffrcinouenic) it is ~h Considering a ~mber of .~ ~i,Rant~N points "sk of lung ~.~N~~ ~-$i~t, since the n dent on ~r is strongly depen ~--~d2tion of smoking (Doll and ~~, 1981), past exposure to ~~ettes is relevant, so that the ~~i~re of the effect of lower tar cigarettes is likely to take many years to emerge; early findings are likely to under- estimate the true benefit. Second, it does not necessarily follow that halving the dose of tar received by the smoker from each cigarette has the same effect on risk as halving the number of cigarettes smoked per day. Only for the. latter dose-response relationship is good evidence available. Third, changes in cigarette de- sign have resulted in chang6s in tar 'quality' as well as tar quanti- ty. Studies by Wynder and Hoff- man (1979) have shown that the mouse skin carcinogenicity of tar per unit dose has been steadily reducing. Fourth, reductions in risk will only be expected if smokers do not increase the number of cigarettes they smoke to 'com- pensate' for the reductions in nicotine yield that tended, at least until about ten years ago (Lee, 1976), to occur con- comitantly with the reductions in tar. Although consumption of .cigarettes per smoker has in- creased nationally, some studies suggest that changes in con- sumption are in fact essentially independent of changes in the tar/nicotine yield of the brand smoked (Garfinkel, 1979; Wald et al., 1980). Given that people currently smoking low tar cigarettes consume ten to 15% fewer cigarettes than middle tar smokers (Lee, 1983), the rise in consumption seems di~e to other causes such as lighter smokers giving up, or changes in price. Smokers may possibly compensate by altering the way they smoke their cigarettes, so the amount of tar they receive may bear little relation to the published yields, which are based on machine smoking under stan- dard conditions. A number of studies have investigated the relationship between machine nicotine yield and nicotine up- take. None of these studies is of the large, long-term 'within- smoker' type needed to provide reliable conclusions, and there is considerable variation in their findings, but all are consistent .with the theory of 'partial corn- September 1983 Vol. 227 1459 0 0

TOPICAL FUCIDIN sodium fusidate fusidic acid rapidly effective ih'sscTlbing Information Fucidin H Gel orgamsms. Fucidin hypersensmvn~, Fucidm Cream 0043/0065. Fucidin 2% fusichc acid with 1% hydrocorusone Avoid extensive use of hydrocomsone m Ointment 0043/5005. B~sic N.ILS. P~ice acetate, Fucudin Cream 2% fumdic acicL pregnancy and mf~tts. Do no~ use/n or 10 g Fucidm H Gel £1.70, Fucidin Cream Fucid./n Oinn'nent 2% sodium fnsidat e, ne~ eyes. Adverse reacti~ms Occa~onal £ 1,55. Fucidin Ointment £ 1,41, Contra-indication~/precaution~ hvDersenmu~ary reacnons, Product Longw~ck Road, Princes

CONTROVERSY that is a person smok- a reduced nicotine brand ~educes his nicotine intake, t~0ugh not proportionately by so large a factor. Russell et al. 11980) claimed support for the mnflicting viewpoint of corn- but re-analy- by Kozlowski et al. I1982) showed a clear trend to- wards reduced blood nicotine Nels in those smoking lower arettes. ~, now at the epidemiol- 0gy, four prospective and five ~ective studies have speci- investigated the relation- between lung cancer and smoked. Of the Separate comparisons made studies and sexes, 19 smokers of lower tar or clgarettes to have a lower lung cancer than smokers ~er tar or plain cigarettes, eXception being a non-sig- increase based on a re- small sample. Overall, reduction in mortal- 25 to 30%. It should be that in all these studies COmparison was between of the old high tar plain and smokers of filter of the middle or mid- tar group, typical of some vears ago. bach et a'l. (1979) com- legree of lung changes dying in 1955-60 dying in 1970-77 and the results show a marked differ- ence. If the lesions he studied are directly related to lung cancer, one cigarette of the past seems to be equivalent {o perhaps three or four modern cigarettes and life- time exposure to lower tar cigarettes may involve markedly tess risk than lifetime exposure to higher tar cigarettes. The changes in national mor- tality rates seem superficially rather unpromising, with male rates fairly static and female rates rising quickly. However, study of trends in total lung cancer rates gives a deceptive picture because in old men and in old and middle-aged women the average number of years smoking is sub- stantially greater now than for men and women of comparable age in earlier years. Since risk is related to the fourth or fifth power of duration of smoking, this increased duration will over- whelm any reduction in risk per cigarette smoked. To gain a more valid picture of potential benefits of tar re- ductions it is necessary to look at young and middle-aged men and young women, where changes in duration of smokir~g will not confound the picture. Over the last 20 years risk in men has fallen at all ages below 60 and has been halved in those aged 45 or less, and risk in women has fallen at ages below 45, and has been halved in the 30 to 34 age group. Although it is possible that reductions in air pollution fol- lowing the Clean Air Act are partly responsible, the magni- tude of the fall is too large to-be wholly due to this. Since average cigarette consumption per head has changed relatively little over the period, the trends are all consistent with tar reduction being the major factor. The fall in the lung cancer rate could be greater in years to come with cigarettes having even lower tar levels. References Auerbach O, Hammond EC, Garfinkel L. (1979): 'Changes in bronchial epithelium in relation to cigarette smoking 1955-1960 vs. 1970-1977". N Engl J Med. 300, 381-386. Doll R, Peto R. ( 1981 ): 'The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today', J Natl Cancer lnsr 66, 1191-1308. Garfinkel L. (1979): "Changes in the cigarette consumption of smokers in relation to changes in tar'nicotine content of cigarettes smoked', Am J Pub Hlth. 69, 1274-1276. Kozlowski LT, Frecker RC, Lei H. (1982): "Nicotine yields of cigarettes, plasma nico- tine in smokers and public health'. Prev Med, I1,240-244. Lee PN. (1976): Statistics of smoking in the United Kingdom. Tobacco Research Coun- cil Research Paper I, 7th ed. Lee PN. (1983): Lung cancer incidence and o'pe of cigarette smoked. Int Lung Cancer Update Conference, New Orleans. Russell MAH, Jarvis M, Iyer R, Feyerabend C. (1980): 'Relation of nicotine yield of cigarettes in blood nicotine level of smok- ers', Br Med J. 280, 972-976. Wald N J, Idle M, Boreham J, Bailey A. (1980): 'Inhaling habits among smokers of different types of cigarette', Thorax, 35, 925-928. Wynder EL, Hoffmann D. (1979): 'Tobacco and health: a societal challenge', New Engl J Med, 30t1, 854-903. 0 The Practitioner 1461 ~0 o 0~ o~ o o

2063628008

WORLD HEALTH ORGANIZATION INTERNATIONAL AGENCY FOR RESEARCH ON CANCER A TOBACCO: MAJOR INTERNATIONAL HEALTH HAZARD Proceedings of an International Meeting organized by the IARC and co-sponsored by the All-Union Cancer Research Centre of the Academy of Medical Sciences of the USSR, Moscow, USSR held in Moscow, 4-6 June 1985 EDITORS D. G. ZARIDZE R. PETO IARC Scientific Publications No. 74 INTERNATIONAL AGENCY FOR RESEARCH ON CANCER LYON 1986

CIGARETTE YIELD AND CANCER RISK: EVIDENCE FROM CASE-CONTROL AND PROSPECTIVE STUDIES S.D. STELLMAN American Cancer Society, Inc. 4 West 35th Street, New York, NY 10001, USA INTRODUCTION The belief that cancer risk can be reduced by lowering the tar yield of cigarettes has been developed from three basic observations: (1) many cancers exhibit a dose-response with respect to the number of cigarettes smoked per day, as shown in Figure 1 (Wynder & Stellman, 1977); (2) cancer risk decreases with number of years of smoking cessation (Fig. 2); (3) tumours can be produced quantitatively in animals using cigarette combustion products (Wynder & Hoffmann, 1967). Although quantitative relationships between cigarette smoking and cancer risk had been developed in both case-control and prospective studies in the 1950s and even earlier, epidemiological confirmation of a specific relationship with cigarette tar yield was not achieved consistently until the late 1960s. Since that time, differences in relative risk have been observed for at least four cancer sites: lung, larynx, oral cavity, and bladder. In this paper we review the data which have led to these conclusions, and discuss some of the similarities and differences in the studies. LUNG CANCER Case-control studies Three series of case-control studies have estimated the relative risk for developing lung cancer in relation to cigarette yield: Bross and Gibson (1968), the series begun by Wynder in the 1960s and continuing into the present (Wynder et al., 1970; Wynder et al., 1976; Wynder & Goldsmith, 1977; Wynder & Stetlman, 1977; Mushinski & Stellman, 1978; Wynder & Stellman, 1979; Wynder et al., 1984), and a cooperative European study begun in 1976 under the auspices of the US National Cancer Institute, covering five countries: the results have been presented as a whole (Lubin et al., 1984a,b) and the Austrian - 197- 03

198 STELLMAN Fig. 1. Relative risk for cancers of the lung (Kreyberg types ~ and II), oral cavity, larynx, oesophagus, and bladder for male current smokers, according to number of cigarettes smoked per day. N, number of cases in case-control • study (from Wynder & Stellman, 1977) lOOI LUNG CANCER (Z) 251 LUNG CANCER N : 486 CASES 1 20 ~ 604 ~ 20 5 ~ O' 0 ~ ORAL CAWTY LARYNX o N: 388 N~27C ~ 20 SMOKER NON- ~-~3 II-ZO 2~-30 3~-~,O 4~* SMOKER OESOR~GUS N:I2~ BLADDER N= 384 NON- I-I0 ll-20 21-30 31-40 41+ NON- I-IO II-20 21-30 31-40 41+ SMOKERSMOKER NO OF CIGARETTES SMOKED PER DAY component has also been published separately (Kurtze & Vutuc, 1980; Vutuc & Kunze, 1982a,b, 1983). Results of these case-control studies are summarized in Table 1, in which comparisons are made between smokers of filter versus nonfilter cigarettes. The relative risk of lung cancer in nonfilter as comoared to filter cigarette smokers as a referent rnn~e~ from 1 3 to 2.3. This must be understood in the context of an individual's lifetime exposure to cigarette tar. The average age of lung cancer diagnosis in the USA is now about 58 years. Most

ClGARElq-E YIELD AND CANCER RISK 199 Fig. 2. Relative risk for cancers of the lung (Kreyberg types I and ll), oral cavity, larynx, oesophagus, and bladder for male former cigarette smokers, according to number of years since cessation of smoking. N, number of cases in case-control study (from Wynder & Stellman, 1977) 16! LUNG CANCER LUNG CANCER(I) CASES=687 CONTROLS=6534 YEARS OF SMOKING CESSATION smokers in this cohort began smoking at a time when there were very few filter cigarettes on the market, and the tar yield of nonfilter cigarettes was over 30 rag. Data from the new Ameriean Cancer Society study (Stellman & Garfinkel, 1986) suggest that a wave of switching from nonfilter to filter cigarettes occurred in the mid-1960s immediately after the appearance of the Surgeon General's report in 1964, which received widespread publicity. Figure 3 shows the proportion of a smoker's lifetime which would have been spent with filter cigarettes, assuming smokers switched from nonfilter cigarettes at about that time, and assuming average ages of beginning to smoke characteristic of this population. It is obvious that recent lung cancer cases received a great deal of their tar exposure in their early smoking years from nonfilter, or from the early high-tar filter cigarettes, irrespective of the types of cigarette they smoke today.

200 STELLMAN Table 1. Relative risks for lung cancer reported from case-control studies, in relation to filter usage' Study Sex Comparison Relative risk Bross & Gibson (1968) Males Fto NSR 3.8 NFto NSR 6.5 NF to F 1.7 Wynder etal.(1970) ~ Males F to NSR 23.6 NFto NSR 38.3 NF to F 1.6 Wynder & Stetlman (1979) Males NF to LTF 1.3 Females NF to LTF 1.4 Lubin et al. (1984 a, b) Males Mixed F and NF to F 2.1 NFto F 2.I Females Mixed Fand NFto F 2.3 NF to F 2.3 • Abbreviations: F, filter cigarette smokers; NSR, nonsmokers; NF, nonfilter cigarette smokers; LTF, long-term fiiter cigarette smokers ~Cases were Kreyberg type I only In three of these case-control series, results have also been presented in terms of specific tar yields. These findings, shown in Table 2, demonstrate that, even allowing for substan- tial differences in schemes for estimating smokers' tar dosage, dose-response relationships are easily discernible. Follow-up studies There have been three important follow-up studies of lung cancer in relation to cigarette smoking in which cigarette yield has been studied in detail The American Cancer Society enrolled over one million men and women aged 40 years and over, in 25 states, in a prospective study in 1959. Follow-ups were conducted annually through 1966, and again in t971 and 1972. Analyses of lung cancer death rates in relation to smoking habits were originally published by Hammond (1966). Hammond et al. (1976, 1977) later presented evidence from this study showing that the lung cancer mortality rates for smokers of 'low tar-nicotine' cigarettes, compared to rates in smokers of 'high tar-nicotine' cigarettes, were reduced by about 20% in men and by about 40% in women. These estimates were made using a matched group analysis which permitted adjustment for many variables at once, including age, race, number of cigarettes smoked per dav. a~e smokin~ began, urban/rural residence, education, iob exoosure to chemicals, X-rays, or other toxicants, history of prior illness, and calendar period (Ham- mond, 1985). Hammond's results are shown in Table 3. For the present review we have re-calculated the standard mortality ratios (SMR) according to quantity smoked daily by current smokers, and by tar yield of cigarette at baseline, for lung cancer in men during 1960-1966, the six years when annual follow-up was done. Calculations were also restricted to this period to minimize effects of changes in

CIGARE'TqE YIELD AND CANCER RISK 201 Fig. 3. Filter cigarette usage as a percentage of total smoking experience, by birth cohort (from Wynder & Stellman, 1979) 70, 60. 50- 20- 10- ~FI LTER i INON-FILTER 25% 5o% 58% 5O% 75% YEAR OF BIRTH smoking habits. In addition, during the first six years of the study, additional confirmation was sought whenever cancer was mentioned on the death certificate, so that the cause of death was based upon 'best evidence'. "Results of this new calculation are shown in Figure 4. There were 967 deaths from lung cancer during this period. For statistical convenience, the reference population is the ,4,° ......e,~,~'v, ~-,~-'~, ~L~CL~ U~ m=diura ~ar-nicoane ctgare~tes, wt~o smoked 20 cigarettes per day. For all other tar-nicotine and quantity categories of smokers, as well as for exsmokers and nonsmokers, expected numbers of deaths were computed by multiply- ing age-calendar-year-specific lung cancer death rates in the reference population by the person-years of exposure to risk of dying in the target group, and summing over age- calendar-year strata. The SMR is the number of observed divided by e:epected deaths. Data were renormalized to give lifetime nonsmokers an SMR of 1.0. 0 O~ O~ 0

202 STELLMAN Table 2. Relative risk for lung cancer according to tar exposure indices proposed by various authors" Reference Sex Relative risk Mushinski & Stellman (1978) Current tar level (rag/day) 0 1- 200- 400- 600- 800- 199 399 599 799 999 1000-1200-1400+ 1199 1399 Kreyberg I Males1.0 5.1 7.4 12.2 20.1 24.8 34.2 30.6 29.9 Females 1.0 7.9 9.6 18.9 28.5 14.8 Kunze & Vutuc (1980); Lifetime tar score Vutuc&Kunze(1982b) Below 501- 1001- 2001- 3001+ 500 1000 2000 3000 Kreyberg t Males2.0 2.6 5.3 7.2 8.3 Females 1.5 4.2 4.8 4.9 6.8 Kreyberg II Males - 1.8 1.8 3.5 3.9 Females - 1.1 3.1 - 2.3 Lubin etal, (1984a) Mean cigarette tar content (mg) ~ (15.6) (18.5) (20.6) (23.6) (25.2) (28.8) Lung cancer Males 1.0 1.2 1.7 1.3 1.3 1.4 Females 1.0 1,9 1.3 1.1 1.5 - Nonsmokers and referent; see Table 5 for definitions of tar exposure indices Categories were combined from wRhin-country 10, 25, 50, 75, and 90th percentiles. Mean tar values (given in brackets) are within each such category Table 3. Standardized mortality ratio for lung cancer among one million men and women followed up for twelve years, relative to lifetime nonsmokers, according to tar-nicotine yield of usual cigarettes, adjusted for age, calendar year, and many other variables (see text) = Standardized mortality ratio 'Low T/N' 'Medium TIN' Males 0.81 0.95 1.00 Females 0.60 0.79 1.00 'Adjusted' deaths: 235.2 285.5 318.4 mFrom Hammond et aL (1976}

CIGARE-i-rE YIELD AND CANCER RISK 203 Fig. 4. Standardized mortality ratios for lung cancer in males, among nonsmokers, exsmokers, and current smokers of low-, medium-, and high-tar/nicotine (T/N) cigarettes (defined by Hammond etaL, 1976). The group was enrolled in 1959, and followed up through 1966. 2O ~X-~,~ i....,.- e . ..... MEDIUM SMOKER S~,......'"" • ........... LOW ~ ./NON-SMOKERS l I L I t t 0 I0 ZO ~0 4.0 4.5 CIGARETTES SMOKED PER DAY At each tar-nicotine level, the SMR increased with quantity smoked, in an approxi- mately linear dose-response relationship. For current smokers of at least 20 cigarettes per day, at each value of daily quantity smoked, the SMR for the 'high tar-nicotine' cigarette smokers exceeded that for the 'medium' group, which in turn exceeded that for the 'low' group. Lifetime non-smokers had lung cancer death rates well below any of the current smokers, irrespective of cigarette yield for the latter. Two other studies are worthy of mention. Rimington (1981) observed 104 lung cancer cases in a follow-up study of 10 414 male volunteers for a mass radiography screening in England. Subjects were enrolled in 1970--1971, and followed for 69 to 81 months. The relative risk for nonfilter versus filter cigarette smokers was reported as 1.54. The inci- dence was computed by dividing the numbers of cases by numbers enrolled, without considering person-years at risk. Adjustment was made for age and for quantity smoked. .In the Whitehall study (Higenbottam et al., 1982), smoking data were available for 17 475 of 18 403 male civil servants aged 40-64 years who were enrolled during 196%1969 and followed for at least ten vears. Ten-~'o~ ~,-~t~ ~-~*~,~ ~.4~,_,_~ted f~r age a=~ cm~Ic3-r-cnt grade, were computed for current smo'kers within categories of inhalation, quantity and tar-yield. There were 108 deaths due to lung cancer among inhalers, and 35 among noninhalers, with tar- and quantity-specific rates for both groups shown in Table 4. Among inhalers, the data show a distinct dose-response at the two lowest consumption levels (1-9 and 10-19 cigarettes per day), although not at the highest, and among noninhalers there is a possible dose-response at the two highest levels (10-19 and 20 or more cigarettes per day). t li;. II tiii!! OTM O~ O~ O

2O4 STELLMAN Table 4. Ten-year lung cancermortality rates (and numberofdeaths) among 17 475 male British civil servants in the Whitehall study, according to quantity smoked, tar yield, and inhalationm No. cigarettes smoked per day Tar yield (rag) 1~3 2~2 ~+ Inhalers 1-9 0.39 (2) 0.53 (1) 1.62 (7) 10-19 1.46(19) 1.55 (8) 2.61 (20) 20+ 2.23 (35) 2.00 (13) 1.79 (3) Noninhalers 1---9 1.08 (4) 0.00 (0) 0.93 (1) 1 0-19 1.25 (5) 1.28 (2) 4.18 (5) 20+ 1.71 (7) 5.81 (9) 5.85 (2) • From Higenbottam et aL (1982) ..2_ co ci: ca 1o ca re ca st: at CANCERS OTHER THA~N LUNG Studies of cigarette yield and cancer have focused mainly on lung cancer, for the obvious reason that, having the greatest incidence and mortality rate of tobacco-related cancers, the numbers of cases available for study are greater than for other sites. Several studies, however, have examined the possible influence of cigarette yield on other cancers. In the American Health Foundation case-control studies, interviewers were instructed to see patients with cancers of the tung, mouth, oesophagus, larynx and bladder. Wynder and Stellman (1979) published relative risks for cancer of the larynx based on 286 male and 64 female cases. After adjusting for age, duration of smoking, number of cigarettes per day, and alcohol consumption, the risk of larynx cancer in nonfilter versus long-term filter cigarette smokers (at least ten years on filters) was 1.49 for men and 3.97 for women (both significant). The relative risk was greater for nonfilter than for filter cigarette smokers at every quantity level. Lee and Garfinkel (1981) reported new analyses of data from the American Cancer Society follow-up study of 1959-1972, in which the relative mortality for smokers of low tar/nicotine cigarettes (as defined by Hammond et aL, 1976) was consistently lower in both men and women than for high tar/nicotine cigarettes for cancer of the buccal cavity and pharynx, oesophagus, larynx, bladder and pancreas. The adjustment procedure, based upon simultaneous matching for nine separate variables, rendered the numbers of effec- tive ('adjusted') cases very small. The mortality rati0g were statistically significant only for cancers of the oesophagus and bladder in women, and for none of the sites in men. Wynder et al. (1976) gave relative risks for cancer of the oral cavity in a case-control study of 593 men and 280 women and matched controls: for nonfilter cigarette smokers versus nonsmokers, 7.8; for long-term filter cigarette smokers versus nonsmokers, 5.7; and for nonfilter versus long-term filter cigarette smokers, 1.4. Adjustment was made for age, but not for alcohol consumption. Significance levels were not given.

CIGARER-E YIELD AND CANCER RISK 205 In a Canadian, population-based, case-control study of 480 male and 152 female case- control pairs, Howe et aI. (1980) reported a reduced risk associated with the use of filter cigarettes compared to nonfilter cigarettes. A recent Italian study of 512 male bladder cancer cases and 596 controls gave a relative risk of 3.0 for nonfilter versus filter cigarette smokers (Vineis e~ al., 1984). On the other hand, there was no difference for men between long-term filter and nonfilter cigarette smokers in the relative risk for bladder cancer in a case-control study by Wynder and Goldsmith (1977), which involved 574 cases and an equal number of matched controls. DISCUSSION There are many methodological issues which must be dealt with in the assessment of the relationship between cigarette yield and cancer outcomes. These fall roughly into four categories: questions of dosage, outcome, other etiological factors and confounding. The strengths and weaknesses of the studies described may be examined largely through attention to these four items. Dosage In any study of cigarette type and disease, dosage is the most important - and in some ways the most difficult - variable to estimate. There are many reasons for this. In the first place, the average tar content of cigarettes has fallen considerably during the past 30 years, even within the same brand. Secondly, some smokers switch brands fre- quently, particularly in response to promotion of the new brands or in response to 'health' publicity. Thirdly, most smokers try to quit at some time in their lives; some are successful, others quit and begin again repeatedly. The actual lifetime dosage of persons in the latter category is quite difficult to determine. Finally, even in well-conducted interviews, sub- jects sometimes recall their smoking history imperfectly, especially regarding duration of smoking specific brands. Many different ways of expressing cigarette dosage have been used, ranging from simple classification as filter versus nonfilter, to elaborate algorithms designed to account for 'complete' year-by-year smoking histories. Cumulative dosage measures have the advan- tage of taking into account the subject's entire history, including early smoking, which may have contributed disproportionately to lifetime tar exposure, since the cigarettes first smoked by persons now in the cancer age group had tar contents two to three times those of current cigarettes. It has the disadvantage of making cumulative scores 'pile up' at the be~innin~ of a smoker's life. during the ve~r~ scores may be insensitive to differences in tar levels between recent brands. Furthermore, cumulative dosage scores, particularly when expressed as 'pack-years', have the disadvan- tage of making two packs per day for 10 years equivalent to one pack per day for20 years, necessitating further adjustment for duration or other parameters. The wide range of tar exposure indices which have been used by various authors is shown in Table 5. These range from categorization of smokers as either filter or nonfitter cigarette smokers (Bross & Gibson, 1968; Wynder & Stellman, 1979), use of the tar rating of the

2O6 STELLMAN Table 5. Tar exposure indices used by various authors Reference Indices Bross & Gibson (1968) 1. Quantity-duration combinations (low, medium, high) 2. Filter versus nonfilter High, 25.8-35.7 mg; medium, 17.6-25.7 mg; low, below 17.6 mg Tar rating of current cigarette Z'(quantity x duration x k) where k= 1, below 15 rag; k = 2, 15-24 mg; k= 3, above 24 mg 1. Lifetime filter versus mixed filter and nonfilter versus lifetime nonfilter 2. W{thin-country quintiles of: Z (tar x cluantity)/.£' (quantity) combined across five countries Hammond etal. (1976, 1977) Mushinski & Stellman (1978) Kunze & Vutuc (1980) Lubin etal. (1984a) current cigarette (Hammond et al., 1976; Mushinski & Stellman, 1978), to fairly elaborate scoring systems presented by Lubin et al.. (1984a), and Kunze and Vutuc (1980). Finally, it has been repeatedly demonstrated and emphasized that people do not smoke identically to machines, and that the tar yields upon which machine analyses are based do not represent the true quantities of particulates or concentrations of vapour phase toxi- cants to which people were actually exposed (Kozlowski et al., 1980; Benowitz et al., 1983). At best, machine-determined yields give relative representations of degree of exposure to cigarette combustion products, such as tar. Since, as has been seen in the preceding sections, the results of studies using different dosage measures are remarkably consistent, we may reasonably conclude that the basic principle that relative risk for lung cancer is in rough proportion to tar yield has been confirmed, despite these many difficulties and the disparities between studies, and that age-specific lung cancer rates may be expected eventually to reflect the falling average tar levels in many Western countries. Outcome In both case-control and follow-up studies, specification of the outcome under investiga- tion is not trivial and may strongly influence interpretation of results. In the series of studies by Wynder and colleagues, and in those by Kunze and Vutuc, lung cancers were classified as Kreyberg Types I or II, the former invariably exhibiting a stronger dose- response to quantity of cigarettes smoked per day. If these observations are correct, it follows that any ameliorative ettect ot lower tar yaelc~ will De oI lesser importance /or adenocarcinoma of the lung than for squamous-cell carcinoma. Other etiological factors Smokin~ is the maior cause of lung cancer in the populations studied, but it is not the only cause., Few of the studies mentioned have made adjustment for exposure to other factors related" to occupation, environment, or nutrition. We have recently shown (Stell-

CIGAREq-FE YIELD AND CANCER RISK 207 man, 1985) that smokers consume foods rich in vitamins A and C much less frequently than nonsmokers. Since vitamin A and similar compounds have been suggested as possible inhibitors of epidermoid cancers, it may in the future be desirable to examine dietary intake along with smoking history. None of the studies reviewed here have done so. Other confounding factors Most of the studies have adjusted for age and sex, but few have examined other potential biases in selection of subjects, differences in social class between cases and controls, etc. These are factors which, especially in hospitalized populations, can strongly affect smoking habits (Wynder et al., 1984). Considering the consistency of results, despite the variety of study designs and populations summarized above, it is not likely that these confounding fachors have played a major role in the studies summarized here. However, it is important to keep them in mind when designing future studies. CONCLUSIONS In three series of case-control studies and three prospective studies conducted in the USA and Europe, the relative risk for lung cancer was found to be consistently lower in both male and female smokers of lower-yield cigarettes. This basic finding continued to hold irrespective of the many different ways in which dosage was expressed, whether qualitatively (filter versus nonfilter) or quantitatively (with explicit tar yields or ranges). Risks for other types of cancer, notably mouth, larynx and bladder, were also found to be lower in smokers of filter cigarettes in a number of North American and European studies. This is all the more remarkable since the designs of studies differed considerably, and the designation of cigarette tar yields for specific cigarettes reflected only crudely true lifetime exposures for individuals. Smokers reaching lung cancer age during the past few years have almost invariably begun smoking nonfilter cigarettes, and many switched to filters during the 1960s, when health warnings gained prominence. It is very likely that as successive cohorts of smokers are exposed to cigarettes of much lower yield for much greater proportions of their lives, the associated risks will decline even further. However, it is to be emphasized that in all studies, risks of smokers of all types of cigarettes, no matter the yields, were significantly higher than those of lifetime nonsmokers. REFERENCES Benowitz. N.L.. Hall. S.M.. Hernin~. R.I.. Jacob. P.. Jones. R.T. & Osman. A.-L. (1983~ Smokers of low-yield cigarettes do not consume less nicotine. New EngI. J. Med., 309, 139-142 Bross, I.D.J. & Gibson, R. (1968) Risks of lung cancer in smokers who switch to filter cigarettes. Am. J. Public Health, 58, 1396-1403 Hammond, E.C. (1966) Smoking in relation to the death rates of one million men and

208 STELLMAN women. In: Haenszel, W., ed., EpidemioIogic Approaches to the Study of Cancer and Other Chronic Diseases (National Cancer Institute Monograph No. 19), Bethesda, MD, US Department of Health, Education. and Welfare, Public Health Service, National Institutes of Health, National Cancer Institute, pp. 127-204 Hammond, E.C. (1985) Matched group analysis method. In: Garfinkel, L., Ochs, O. & Mushinski, M. , eds, Selection, Follow-up, and Analysis in Prospective Studies: A Workshop (National Cancer Institute Monograph, No. 67; NIH Publication No. 85- 2713), Bethesda, MD, US Department of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Institute, pp. 15%160 Hammond, E.C., Gaffinkel, L., Seidman, H. & Lew E.A. (1976) "Tar" and nicotine content of cigarette smoke in relation to death rates. Environ. Res., 12, 263-274 Hammond, E.C., Garfinkel, L., Seidman, H. & Eew E.A. (1977) Some recent findings concerning cigarette smoking. In: Hiatt, H.H., Watson, J.D. & Winsten, J.A., eds, Origins of Human Cancer, Book A, Incidence of Cancer in Humans, Cold Spring Harbor, NY, Cold Spring Harbor.Laboratory, pp. 101-112 Higenbottam, T., Shipley, M.J. & Rose, G. (1982) Cigarettes, lung cancer, and coronary heatX disease: the effects of inhalation and tar yield. J. Epidemiol. Community Health, 36, 113-117 Howe, G.R., Burch, J.D,, Miller, A.B., Cook, G.M., Esteve, J., Morrison, B., Gordon, P., Chambers, L.W., Fodor, G. & Winsor, G.M. (1980) Tobacco use, occupation, coffee, various nutrients, and bladder cancer. Y. natl Cancer Inst., 64, 701-713 Kozlowski. L.T., Frecker, R.C., Khouw, V. & Pope, M.A. (1980) The misuse of "less- hazardous" cigarettes and its detection: hole-blocking of ventilated filters. Am. J. Public Health, 70, 1202-1203 Kunze, M. & Vutuc, C. (1980) Threshold of tar exposure: analysis of smoking history of male lung cancer cases and controls. In: Gori, G. & Bock, F.G., eds, A Safe Cigarette? (Banbury Report No. 3), Cold Spring Harbor, NY, Cold Spring Harbor Laboratory, pp. 29-34 Lee, P.N. & Garfinkel, L. (1981) Mortality and type of cigarette smoked. J. Epidemiol. Commun. Health, 35, 16-22 Lubin, J.H., Blot, W.J., Berrino, F., Flamant, R., Gillis, C.R., Kunze, M., Schmahl, D. & Visco, G. (1984a) Patterns of lung cancer risk according to type of cigarette smoked. Int. J. Cancer, 33, 569-576 Lubin, J.H., Blot, W.J., Berrino, F., Flamant, R., Giliis, C.R., Kunze, M., Schmahl, D. & Visco, G. (1984b) Modifying risk of developing lung cancer by changing habits of cigarette smoking. Br. reed. J., 288, 1953-1956 Mushinski, M.H. & Stellman, S.D. (1978) Impact of new smoking trends on women's occupational health. Prey. Med., 7, 349-365 Rimington, J. (1981) The effect of filters on the incidence of lung cancer in cigarette smokers. Environ. Res.. 24, 162-166 btettman, 5.D. (t985) Chairman's remarks on Session V: Data analysis in cohort studies. In: Garfinkel, L., Ochs, O. & Mushinski, M., eds, Selection, Follow-up, and Analysis in Prospective Studies: A Workshop (National Cancer Institute Monograph, No. 67, NIH Publication No. 85-2713), Bethesda, MD, US Department of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Insti- tute, pp. 145-147

CIGARETFE YIELD AND CANCER RISK 209 Stellman, S.D. & Garfinkel, L. (1986) Smoking habits and tar levels in a new American Cancer Society prospective study of 1.2 million men and women. J. natl Cancer Inst. (in press) Vineis, P., Estbve, J. & Terracini, B. (1984) Bladder cancer and smoking in males: types of cigarettes, age at start, effect of stopping and interaction with occupation. Int. J. Cancer, 34, 165-170 Vutuc, C. & Kunze, M. (1982a) Lung cancer risk in women in relation to tar yields of cigarettes. Prey. Med., 11,713-716 Vutuc, C. & Kunze, M. (1982b) Cigarette tar exposure and occupation in female lung cancer patients. Excer_pta reed. Int. Congr., Ser. 55B, 41-48 Vutuc, C. & Kunze, M. (1983) Tar yields of cigarettes and male lung cancer risk. J. natl Cancer Inst., 71, 435-437 Wynder, E.L. & Goldsmith, R. (1977) The epidemiolog2¢ of bladder cancer. A second look. Cancer, 40, 1246-1268 Wynder, E.L. & Hoffmann, D. (1967) Tobacco and Tobacco Smoke. Studies in Experi- mental Carcinogenesis, New York, Academic Press Wynder, E.L. & SteIlman, S.D. (19"~7) Comparative epidemiology of tobacco-related cancers. Cancer Res., 37, 4608-4622 Wynder, E.L & Stellman, S.D. (1979) Impact of long-term filter cigarette usage on lung and larynx cancer risk: a case-control study. J. natl Cancer Inst., 62, 471-477 Wynder, E.L., Mabuchi, K. & Beattie, E.J. (1970) The epidemiology of lung cancer. Recent trends. J. Am. reed. Assoc., 213, 2221-2228 Wynder, E.L., Mushinski, M. & Stellmart, S.D. (1976) The epidemiology of the less harmful cigarette. In: Wynder, E.L., Hoffmann, D. & Gori, G.B., eds, Modifying the Risk for the Smoker. Vol. I. Proceedings of the Third World Conference on Smoking and Health, New York City, June 2-5, 1975, (DHEW Publication No. (NIH) 76-1221), Bethesda, MD, US Department of Health, Education, and Welfare, Public Health Service, National Institutes of Health, National Cancer Institute, pp. 1-12 Wynder, E.L., Goodman, M.T. & Hoffmann, D. (1984) Demographic aspects of the low- yield cigarette: considerations in the evaluation of health risk. J. natl Cancer Inst., 72, 817-822

2063628023

I 48 New Scientist 24 May 7984 Experimental animals E. Pascoe's attack on Dr Alma is disingenuous (Letters, 12 April. p 54). If his pamphlet was not intended to dissuade young researchers from becoming too concerned about the use of animals in experiments, why choose the very odd wording which he employs? \Vhy the persistent emphasis on whether the researcher finds a particular experiment upsetting to do? Why not. for example, a paragraph simply stating: "Vv'e can reasonably make the assumption that the experience of pain. discomfort or mental anguish is an evil for all sentient creatures. Hence it is obligatory, for the researcher to take all possible measures to reduce any distress likely to result from an experiment. Anaesthetics. analgesics and tranquillisers may be used. or it may be possible to substitute in vitro methods. You should discuss the proposed experiment with your colleagues, and with the Ho~e Office Inspector. and seek advice from a statistician so that numbers of animals can be reduced to the minimum needed to yield a valid result. You should famitiarise yourself with the relevant literature so that you can be confident what you propose to do really does represent a new contribution to ;cience, If your proposed work nvolves surge~', it is desirable to :[iscuss it with colleagues who possess veterinary, qualifications." As it stands, the pamphlet can only give young biologists the impression that. so far as the Physiological Society is concerned, concern about the feelings of experimental animals is merely an irrational prejudice, Rosemary Rodd Scientifid adviser Quaker Concern ./'or Animal ;f'elfare Cambridge As a laboratory technician working with animals Iapplaud Gill Langley's article on the exploitation of laboratoD' animals (Redundancy for the laboratory guinea pig", 3 May, p 12). All animals should be free from our modern concentration camps. To cause suffering to any creature is wrong, and I long for the day that I can talk about my job without feeling guilt. Ashby McGowan Glasgow Sounds bad I object to the Fisher advertisement in New Scientist (I0 May, pp 34-35), The advertisement refers to an imaginery Daphne Heaton- Smvthe as an inane woman who "thi'nks 'Wow" and 'Hurter' are dogs in the local hunt', Her husband "more sensibly" (of course) realises how good the product is. As a woman (who happens to be a scientist and head teacher of a girls" comprehensive school) I find this deeply offensive. Sister Mary B. 0 ?vSaIIey West Didsbury Manchester Pesticide hazards Using alternatives such as cell cultures instead of live animals (3 May, p 12) is one way of reducing animal experimentation: freedom of information is another. Under Britain's Pesticides Safety Precautions Scheme~ however, many manufacturers may wish to sell the same pesticide they must all repeat an identical battery of toxicity tests. Each manufacturer's results are kept secret from the others--and from the public. The object apparently is to stop, say, the 20th manufacturer of a given pesticide getting safety clearance without meeting the same costs as the first. That may appeal to accountants, but from every, other point of view it is wasteful and senseless. Animals are endlessly sacrificed repeating tests that have been done many times before; and people who work with, or are exposed to, pesticides are denied basic information about their hazards (the Ad~4sory Committee on Pesticides won't even say what kinds of tests have been done, let alone discuss the results). The government has announced that it proposed new pesticide legislation. It has the opportunity to introduce a more efficient and accountable system. Publish a manufacturers' results the first time they are done. When a second manufacturer wants to market the same chemical, it could be required to compensate the first for sharing the original data, instead of pointlessly duplicating it. Even better, the money could go to a central research ~'und and be used to investigate new problems. That would prevent much wasteful experimentation: it would increase the amount of original research done on such chemicals: and it would for the first time allow public access to basic information about pesticide hazards. Maurice Frankel The ] 984 Campaign for Freedom of InJbrmation London Dialogues of the deaf The example of the short paper On the role and activities of the Advisor' Board for the Research Council~ (ABRC) and inability of scientists to comprehend it (Forum, 24 April, p 24) does not seem to be unique to Britain. Wilson Dizard writes in his book The Coming Information Age about the US: "As an MIT study put it several years ago: communications technology is flooding policy makers with options they do not understand, among which they must choose, and which will have profound effects on society." What chances does the rest of the society stand to understand the impact o(IT on their lives, if two (or even three) relatively small "cultures" fail to communicate with each other, especially when, as Dizard writes, in the present age of converging technologies and greater social complexity, the balance between economic I THIN~, (VE I..-OGAT',~ TH~ B o~ ~roduct harmony become difficul to maintain"? This Dizard says, is why "we do need a better understanding of the issues raised in the Nora and Minc report to the French government form 1978, facing esser~tially the same situation--a massive technocractic drive, threatening to go out of control unless its potentially dehumanising effects are understood and reined in". Ivor Fodor Darmstadt West Germany Tobacco-related disease In Britain, only two external factors have thus far been identified as really major killers--tobacco smoke, and the (still incompletely characterised) dietary, determinants of blood lipids. Each of these is responsible for oftbe order of 100 000 deaths a year, out of our annual total of 600-odd thousand. At a recent Ciba symposium, where I was asked to lecture on the control of tobacco-related disease, I pointed out two main approaches were possible--reduction of the number of cigarettes consumed, and reduction of the hazard per cigarette--and that both were important. Admittedly, the evidence thus far available suggests that the types of tar-level reductions that have been introduced in Britain during the past quarter of a century appear to produce substantial avoidance of risk only for lung cancer: no GRIMBLEDON DOWN q Bill Tidy

New Scientist 24 May 1984 substantial effect (in either direction) on the.other main slno,Vdng-relatk~d diseases has been demonstrated. Therefore, although further reductions in tar levels should continue to be encouraged (as long as this can be done in prays that do not interfere with ff~rts to reduce overall cigarette consumption), we need to encourase research on how to design cigarettes that also produce lower risks of heart disease and chronic lung disease. Your report of my talk (19 April) inadvertently attributed to me certain views that I neither expressed nor could consistently hold• For example, while I emphasised the importance of continuing to encourage reductions in tar level, the first sentence of your report (under the headline "Unhealthy verdict on tow-tar brands") misleadingly attributes to me the view that low- tar brands could increase the chance of dying from heart and lung disease. Three paragraphs later, it attributes to me exactly the opposite "belief" for chronic obstructive lung disease! Elsewhere. although in fact Englishmen have lung cancer rates a hundred times larger at age 80 than at age 40. I am misquoted as describing 35-44 year-olds as being the group "most susceptible to lung cancer". Ricitard Pcto Radcli~, Infirmary O.~tbr3 " " Parlous view I dtd not say that "'particle a~sicists have no wish to know other branch of science", but nat we have no wish to knock any other branch ILetters. 10 May, p 49). That, I am sure you will agree, is a very different thing. Imperial Colleee ol" Science and London Leprosy The article on leprosy by Debora MacKenzie (3 May. p 30) reiterates the current World Health Organlsation (WHO) view that the prevalence, but not the incidence, of leprosy has declined since the introduction of dapsone. This view is held despite the huge reductions in numbers of patients reporting for treatment (the so- called "case-detection rate"). The argument case-aetecuon ~oes not mean a true reduction in new cases of the disease/that is to say, the incidence), but merely reflects a declining bac "klog of old patients over time. This explanation will certainly account for a reduction in the treatment is first offered, but as a decline continues over many years it becomes increasingly implausible that this merely reflects a backlog effect. For ~mple, in Burma the case- tcct~on rate fell from 5 per I000 1962 to 0.8 per 1000 in 1972, and the rate of decline was as great in the second half of that period as in the first (Int. J. Leprosy, vol 43, p 125). Further evidence for the effectiveness of dapsone in reducing the true incidence of new cases of leprosy comes from a comparison of Uganda and New Guinea. Sulphones were introduced to an area of New Guinea only in late 1967, and the case-detection rate remained at 6 per 1000 each year from 1964 to 1969 (Med J. Austral., vol 1, p 1258). In contrast, the case-detection rate in Uganda fell from 5.5 to 1.7 per 1000 in four years (in the control group of the BCG trial), but dapsone treatment was employed throughout that period (Nature, vol 254, p 168). Finally, the map in the article incorrectly attributes a prevalence rate of more than 20 per 1000 to the Northern Territory of Australia, despite the publication of the correct data in 1977 (Med. J. Austral., vol 2, p 652), showing a fall from 46 new cases in 1967 to just 6 in 1976. C. L. Crawford Chafing Cross Hospital Medical School, London QED In the issue of New Scientist for 10 May (p 3) you carry a note "UN admits failure to halt deserts". In the same issue (p 30) there is an article b\ John Gribbin entitled "The world's beaches are vanishing". Surely the solution is obvious! Rtlth Newmark Bishops Stortford Random leapfrogs There is another explanation for the occurrence of "leapfrog" patterns in the plumage of birds (Monitor. 10 May, p 22) If a species becomes broken up into three fairly isolated w~atever reason, in the ~lumage of the central group would inhibit breeding between it and either of the other groups. The two outer groups would be far apart and so also unlikely to mate. This situation could arise only if the ,.g, ou.p He along a line (which is there were no ancient flight corridors linking the outside pair of sub-species. If one of the outer populations develops different plumage, then two species would result (assuming occasional breeding between the two unchanged groups). This interpretation differs from J. V. Remsen's in predicting leapfrog patterns in factors which may affect mating behaviour such as vocal dial.ects, but not in other characteristics such as bone structure. In plants only those characteristics important to animal pollinators, or which otherwise affect pollination, could be expected to show this pa~ern. Whether the initial change were random or adaptive could only be decided in each case individually, if at all Paul Gailiunas Gosfor~h, Newcastle upon Tyne Competition Robert Brooks (Forum, 5 April, p 45) advises "never cite your enemies in the bibliography. The danger here is that the editor may select referees from your list of references..." In the Journal of the American Medical Association (vol 218, p 886). I indicated that it is necessar7 to quote your competitors. If you would have people locate your publications through Science Citation Index, whenever innocent strangers look up your enemies, they will be certain to learn about your publications. This is a new variation on the uncertainty principle. The closer you get to ignoring your competition the closer you come to oblivion. To achiev~ a total state of oblivion always use clever but ambiguous tries in your papers, use a pseudonym and publish in any one of the numberous obscure languages available. Eugene Garfield Institute for Scientific Information, Philadelphia, Pennsylvania I cannot let Trevor Kitson (Letters, 3 May, p 50) have the I~ last word. Both he and Robert. Brooks's (Forum, 5 April, p 45) make the error of believing that age brings quality. Simply dividing the number ofphblicafio'ns by age a person who, from age 25, " publishes regularly some two papers a year can expect his coefficent to rise from 2/26 to 50/50, a 25-fold improvement. Furthermore, it is accepted that the £wst author actually carried out most of the work, No. 2 kept things running when No. 1 was on holiday, No. 3 made the coffee, and so on right down to the last author who never bothered to read the paper or, if he did, didn't understand it. My revised coefficent of publicatonmanship (CA is therefore- x 1 ~ X age--25 y Where x is the number of publications, and y is the position ~n order of names. R. Lathe Edinburgh We welcome letters from our readers. Short communications stand the best chance of publication. We reserve the right to edit the longer ones. Write to: Letters to the Editor, New Scientist Commonwealth House, 1-19 New Oxford ST, London WCIA 1NG. Warren Spring Laboratory 1959-1984 WSL will be displaying work on Process Technology on June 28 and 29 1984 For an invitatiop these Open Days Telephone 0438-313388 Ext 256 or 219 O~ ~0 O~ 0

0 O~

RESEARCH ADVANCES IN ALCOHOL AND DRUG PROBLEMS Series Editors Reginald G. Smart Howard D. Cappell Frederick B. Glaser Yedy Israel Harold Kalant Robert E. Popham Wolfgang Schmidt Edward M. Sellers A Continuation Order Plan is available for this series. A continuation order will bring delivery of each new volume immediately upon publication. Volumes are billed only upon actual shipment. For further information please contact the pub- lisher. Research Advances in Alcohol and Drug Problems Volume 8 Edited by Reginald G. Smart, Howard D. Cappell, Frederick B. Glaser, Yedy Israel, Harold Kalant, Robert E. Popham, Wolfgang Schmidt, and Edward M. Sellers Addiction Research Foundation and University of Toronlo Toronto, Ontario, Canada PLENUM PRESS • NEW YORK AND LONDON

308 BARRY S. BROWN Vagl~rn, P., and Fossheim, L, 1980, Differential troatment of young abusers: A quasi.experimental study of a "therapeutic coanmunity" in a psychiatric hospitut, J. Drug Issues 10:505. Waal, H., 1980, Unconventional treatment models for young drag abuses in Scandinavia, J. Drug l~ues 10:441. Wesson, D. R., and Smith, D. E., 1979, Treatment of the polydrng abuser, in: Handbook on Drug Abuse, (R, L. DuPont, A. Goldstein, J. O'Donn¢ll, and B. S. Brown, eds.), pp. 151-157, Government Printing Office, Washington, D.C. Wesson, D. R., Smith, D. E., and Lemer, S. E., 1975, Streetwise and nonsUeetwise polydrug typology: Myth or reality, J. Psychedel. Drugs 7:121. Wesso~, D. R., Smith, D. E., Lerner, S. E., and Keuner, V. R., 1974, Treatment of polydrug users in San Francisco, Am. J. Drug Alcohol Abuse 1:159. Wesson, D. R., Grant, I., Carlin, A. S., Adams, K. M., and Harris, C., 1978, Neuropsychological impairment and psychopathology, in: Polydrug Abuse (D. R. Wesson, A. S. Ca.din, K. M. Adams, and G. Beschner, eds.), pp. 263-272, Academic Press, New Wexler, H. K., and De Leon, G., 1977, The therapcntic community: Multivariate prediction of retention, Am. J. Drug Alcohol Abuse 4:145. Wikler, A., 1968, Diagnosis and treatment of drug deport@race of the barbiturate type, Am. J. Psychiatry 125:758. Winer, L. R., Lorio, .I.P., and Scraffo~, I., 1974, Effects of treatment on drug abuser and family, Special Action Office for Drug Abuse Preventiou Report 4 RGO03 (1974). Winn, L, 1982, Kukulu Kumuhana--Final Evaluation Repot, NIDA Grant Report HSI DA 02056 (March, 1982). Wunderlich, R. A., Lozes, L, and Lewis, J., 1974, Recidivisru rates of group therapy participants and other adole.set~ts processed by a juvenile court, Psychother. Theory Res. Pracffce 11:243. gZOgg9890g 11 Less-Hazardous Tobacco Use as a Treatment for the "Smoking and Health" Problem LYNN T. KOZLOWSKI Tobacco is a dirty weed. I like tt. It satisfies no normal need. I like it. It makes you thin, it makes you lean, It takes the hair right off your bean. It's the worst darn stuff I've ever seen. I like it. G. L. Hre.Ml~ff~OER, in Penn State Froth, 1915 1. INTRODUCTION The health care industry cares about tobacco use mainly because it causes death and disability in users and perhaps in their associates (e.g., U.S.D.H.E.W., 1979, U.S.D.H.H.S., 1982). The war against tobacco use is at root a war of messages and recommendations about conduct. Persuasion is important in this particular war on drugs because the product in question is neither illegal nor difficult to obtain. If tobacco kills or injures, then, assuming no redeeming values, the obvious message is to stop or not start using tobacco. Unfortunately, tobacco appears to have some redeeming value, if only to the dependent user who suffers without it (e.g., Schachter et al., 1977; Silverstein, 1982). Whatever the reasons, history shows that tobacco, once introduced to a culture, is never eliminated, even in LYNN T. KOZLOWSKI • Addiction Research Foundation, Toronto, Ontario, Canada. The views expressed in this publication are those of the author and do no/necessarily reflect those of the Addiction Research Foundation. 309

310 LYNN T. KOZLOWSKI LESS-HAZARDOUS TOBACCO USE highly coercive societies (Brooks, 1952). However, change or evolution has taken place in the types of tobacco that are most popular (Kozlowski, 1982a). Once one is forced to assume that tobacco has redeeming values (benefits), then one needs to open negotiations with the enemy to determine if some deal can be struck with those who do not wish or who are unable to give up tobacco use entirely. The goal of the negotiation is to make the best of a health risk by minimizing it, knowing that, for some, it can not be eliminated practically. Partial victory is substituted, where possible, for total defeat. The less-hazardous-tobacco-use message is directed toward those who are unwilling or unable to stop using tobacco completely. Ideally, it acts when the antitobacco message (stop or don't start) fails, and it complements the antitobacco message; in practice, these two messages are competitive and troubled allies. At present, both the less-hazardous and the antitobacco messages indicate goals that we are struggling to find out how to attain. We are still trying to discover how best to prevent, stop, or modify tobacco use. [Incidentally, the protobacco response to the above messages is that we really don't know yet if tobacco is dangerous, so continue to use tobacco if you care to (Friedman, 1975).] This chapter will argue that the use of less-hazardous tobacco, if prohibi- tionistic impulses can be put aside, may have an important role in the treatment of the smoking and health problem. Just as research efforts arc needed to try to improve prevention and cessation techniques, they are needed to try to improve the techniques of less-hazardous tobacco use. (For a review of issues related to the application of less-hazardous-tobacco-use treatments, see Kozlowski, 1984.) The phrase "less-hazardous tobacco use" is meant to be inclusive. Cigarettes, for example, are the most hazardous tobacco products overall; yet even cigarettes can become a less-hazardous use of tobacco, if only a little of a few cigarettes is smoked each day. On the other hand, some less-hazardous tobacco products are less-hazardous in certain respects no matter how they are used: chewing tobacco, for example, carries no risks of fire and essentially no risk of lung disease. For some workers concerned with smoking and health, the mission of this chapter is outrageous. For these individuals (and institutions), no tobacco product can be part of the treatment of the smoking and health problem; complete pre- vention and absolute cessation of all tobacco use are the only acceptable gods. The exclusive goal of exterminating all smoking and tobacco use is so prominent a feature of the contemporary discussion of "Smoking or Health" that it will be necessary to (1) try to account for the predominance of this goal and (2) confront the possible pitfalls of pursuing only this means of reducing the health conse- quences of smoking. To try to avoid some needless arguements, I will define how I am using some key terms, before entering into the debate. Less-hazardous means reduced in risk or not as dangerous; it does not automatically mean safe or without risk. Tobacco use can refer either to (1) the type of product or (2) the nature of its use. This chapter is not mainly about the so-called "less-hazardous cigarette." Low-yield cigarettes will not be referred to 6~0~9890E as less-hazardous or safer cigarettes. Though they may indeed be less hazardous than high-yield cigarettes, this point is still controversial (e.g., Kozlowski et al., 1982b; Gerstein and Levison, 1982). Though low-yield cigarettes are low-yield when placed in the ports of smoking machines, they are not necessarily low- yield when placed in the mouths of smokers. In fact, the lowest-yield cigarettes (1 mg tar, 0. lmg nicotine) can turn into medium- or high-yield cigarettes when a smoking machine assay is adjusted tO simulate better the smoking behavior seen in a human smoker apparently bent on compensating for the reduced yields (Kozlowski et al., 1982c). Although treatment is often a medical term, it is employed here in its more fundamental meaning as a way of dealing with something. The smoking problem and the tobacco problem are more general than the smoking and health problem. Some individuals view any form of tobacco use as a serous waste of time and resources and as an activity to be discou~ged; these views would hold even if tobacco use posed no risk of disease or disability. If one believes that tobacco use, per se, is a problem to be eliminated, then less-hazardous tobacco use presents at least one problem too many. If one believes that tobacco is a problem primarily because of serious effects on health, then the reduction of the toxic consequences of tobacco use is a worthwhile goal. The smoking and health problem focuses on the damage to health caused by cigarettes. In this chapter, addiction or compulsive drug use, per se, is not considered a major health problem, unless the drug-taking behavior causes serious physical, social, psychological, or behavioral disturbance. According to the Diagnostic and Statistical Manual of the American Psychiatric Association (DSM-III) (Amer- ican Psychiatric Association, 1980), smoking becomes an official disorder 005. lx Tobacco Dependence) if serious attempts to stop or reduce tobacco use have been unsuccessful, if tobacco withdrawal occurs during tobacco abstinence, or if the tobacco use continues despite a serious physical condition (e.g., respiratory or cardiovascular disease) that the user knows is exacerbated by tobacco use. No mention is made of "impairment in social or occupational functioning." Tobacco dependence is, in fact, alone among the several substance use disorders described in DSM-III (alcohol, barbiturates, cocaine, opioids, amphetamines, phencyclidine, hallucinogens, cannabis) in that impairment in social or occu- pational functioning is not judged to be an "immediate and direct" result of the use of the substance. 2. BACKGROUND Compassion and Venom According to an Arabian story (Bain, 1896), the Prophet, Mahomet, rescued a snake from freezing by warming tbe snake against his body. The thankless snake bit him, but Mahomet sucked the venom from his wound and spat it upon

312 LYNN T. KOZLOWSKI the ground. On that spot, it was said, grew the first tobacco plant, combining the compassion of the prophet with the venom of the serpent. The quest for less- hazardous tobacco products has been directed toward reducing or eliminating the "venom" of tobacco, while at the same time keeping its "compassion." From a hedonistic perspective, then, less-hazardous tobacco use strives to maintain pleasure and minimize pain. From the perspective of the marketplace, the ideal less-hazardous product sells well, but does not kill the customers. Low-yield cigarettes lack one of the key requirements of a less-hazardous tobacco product, in that they remove the "compassion" along with the "venom." Low-yield cigarettes---as designed, not necessarily as smoked---are little more than placebo cigarettes. Ultra-low-tar cigarettes are ultra-low-smoke cigarettes. As much as 80% of the smoke in each puff of a cigarette yielding 1 mg tar can be diluting air (Kozlowski, 1981b). On the assumption that some of the phar- macologic actions of tobacco are responsible for the "compassion" (nicotine is most often thought to be tbe key ingredien0, an across-the-board reduction in drug delivery hardly qualifies as a practical strategy for producing an acceptable less-hazardous tobacco product. Tobacco use is often understood in a much too simplified way. That harm and benefit, venom and compassion, can reside in the same product is readily appreciated in many areas of applied research, yet researchers in the tobacco area have tended either to identify tobacco as a killer or to deny that claim. Though many individuals die from both the direct and indirect use of automobiles, I can not recall heating the argument that, therefore, all automobile use should be prevented or stopped. Because of the widely appreciated benefits of the automobile, the less-hazardous automobile movement has been more prominent than the antiautomobile movement. Jumping to Exclusions In their classic book on logic, Cohen and Nagel (1962) note: "One of the most fruitful sources of intellectual confusion is the too facile assumption that any two propositions which are not equivalent are mutually exclusive" (p. 68). At least one of the ramifications of this confusion can be seen in the ready employment of false dichotomous questions. Such questions make a practice of opposing issues that are neither exhaustive nor mutually exclusive (Fischer, 1970). Notice that the Royal College of Physicians in the United Kingdom entitled their recent monograph on the health consequences of tobacco use "Smoking or Health" (Royal College of Physicians, 1977). To return to the analogy with the automobile, it is as if a book were entitled "Driving or Health." In neither instance is the dichotomy justified. Not all drivers suffer ill-health as a conse- quence of automobile use; some do. Not all smokers suffer ill-health as a con- sequence of tobacco use; some do. Neither are all nonsmokers and nondrivers certain to be healthy. O80899890g LESS-HAZARDOUS TOBACCO USE 313 False dichotomies deny the crucial middle ground and emphasize the ex- tremes. They polarize a question. They add to the memorability of the question. They might provide an image around which to rally contributions and interest in a problem. Yet it is perilous for scientists to treat them as any more than slogans or entertainments. To use the false dichotomy to guide research on the problem is, in fact, to base one's exploration on an unsupportable premise. Importance of Beliefs Beliefs and values are the first principles from which the creation and eradication of social problems flow (Lindblom and Cohen, 1979). Outside of the sometimes idealized world of public health education, complicated beliefs and circumstances contribute to the valuation of tobacco. Those who thought that announcing that "smoke kills" would lead to an exodus from the bondage of tobacco use might also have predicted that the high risk of earthquakes should have emptied California. If a patient in your care or a loved one dies or suffers from a tobacco-related disease, the costs of the activity may overwhelm the benefits. If you support your family through the sale of tobacco (or if you have gone to college because of a scholarship from a tobacco company), the benefits of tobacco may be salient. If you are a smoker who feels some pleasure in smoking, then the threat of a future death from smoking might be countered persuasively with the conviction that one must, after all, die of something and that, despite the most pampered life in the world, an accidental death from any number of causes could lurk around the next corner. Cigarette smoking is argued to be the "largest preventable cause of death in America" (U.S.D.H.E.W., 1979). The term "preventable" is a problem. The wish to prevent should not be confused with the ability to prevent. In later sections of this chapter, it will be argued that there are limits to the preventability of tobacco use. The limits on the preventability of tobacco use become some of the strongest arguments for developing less-hazardous modes of tobacco use. Beliefs and values influence what one chooses to be preventable. Many activities are preventable, given enough effort to prevent; however, drug use has not shown itself to be an area of easy prevention, despite large investments in wars on this or that drug. It is doubtful whether drug use is, in practice, preventable in a free society. (Prevention and deterrence are quite different concepts.) I would revise the quotation that opens this paragraph by stating that cigarette smoking is the largest cause of death that authorities are trying to prevent. Scientific Haggling Scientific dispute does not take place at a level above the usual mire of human argument. The mantle of science is worn by people whose conflicts with a colleague share much with arguments with a spouse. Such maneuvers as

314 LYNN T. KOZLOWSKI intimidation, throat, insult, belittling, evasion, and flattery, to give a partial list, are as readily found in rational argument as they are in everyday argument (Lakoff and Johnson, 1980). Lakoff and Johnson assert that both rational argument and haggling are grounded in the metaphor "argument is war." Anyone familiar with the manufacture of the scientific bullets used in the war against (and for) tobacco use should he aware that all is fair in love and rational argument, especially when an emotionally and economically charged issue is involved. Some of the most striking reactions to the topic of less-hazardous tobacco use have not appeared in print, but have occurred behind the scenes at scientific and medical meetings. In 1980, for example, a scientific meeting was convened by a major voluntary agency in the United States, in part to help set research priorities on smoking and health. The chairman of the smoking group had pre- pared a position paper that was to focus discussion on a list of research topics. In the opening plenary session, the distinguished chairman was asked, "Why isn't the issue of doing research on less-hazardous tobacco use on the agenda?" With no hesitation, he responded, "Better men than I have been ruined for proposing such a thing!" Be assured that although this comment influenced the proceedings, it did not appear in them, and neither did less-hazardous tobacco use appear in the list of topics in need of research. Fortunately, this section will not have to rely on undocumented anecdotes to establish the needed background. In 1978, Science (Marx, 1978) reported on an event that was distressing the highest "smoking or health" officials in Wash- ington, D.C. The Secretary of Health, Education, and Welfare (HEW) had been mounting a vigorous, high-profile campaign against cigarette smoking. Plans were being made for the 1979 Report on Smoking and Health of the Surgeon General. This report was to be released on the 15th anniversary of the landmark 1964 Surgeon General's Report. The 1979 Report was to be roughly three times the size of the 1964 Report. The promotion of healthier "lifestyles" was fast becoming a popular activity: everywhere running shoes were filled with jogging feet. In this atmosphere, a government scientist, Dr. Gio B. Gori (no less than the Deputy Director, Division of Cancer Cause and Prevention, National Cancer Institute) published a paper (with Cornelius Lynch) indicating that low-yield cigarettes, especially modern low-tar brands, were less hazardous than high- yield cigarettes. The paper encouraged smokers to wean themselves progressively to less-hazardous cigarettes as "an alternative to smoking cessation that is perhaps more effective than the self-denial approaches of current anti-smoking messages" (God and Lynch, 1978). Although Gori and Lynch were careful to avoid calling low-yield cigarettes "safe," the publicity surrounding the publication in the pres- tigious Journal of the American Medical Association announced that "tolerable" cigarettes were at last available. As described by Marx (1978), "... the sug- gestion by a government scientist that smoking might be 'tolerable' was not 1808~9~90~ LESS-HAZARDOUS TOBACCO USE 315 well received by health officials who were afraid it would undermine their anti- smoking efforts." [In an equally notable example of understatement, God wrote in a summary essay: "Public policy in smoking and health has been dominated for years by idealistic approaches with moderate sympathy for less-hazardous cigarettes" (Gori, 1980).] Those who were upset about the Gori paper and its impact included: the Secretary of HEW, the Surgeon General, and the Directors of the National Cancer Institute (NCI) and the National Heart, Lung, and Blood Institute. Dr. Gori is reported to have told the press that the Secretary of HEW was trying to have the NCI fire or at least discipline him. At the time of the Marx report, Dr. God had been removed recently from command of the NCI Smoking and Health program. Dr. Gori no longer works for the NCI. The Gori and Lynch (1978) and the earlier Gori (1976) papers did offer encouragement that those smokers who would not stop smoking completely could benefit from a switch to lower-yield cigarettes. Although this may be true under some circumstances, the Gori research has been the object of a great deal of scientific criticism (e.g., Gart and Schneiderman, 1979; Warner, 1979), and I would be toward the front of the line in criticizing the Gori work. One of my key objections concerns the uncritical acceptance of the lower yield ratings of recent cigarettes. Such acceptance places unwarranted confidence in the adequacy of the simulation of human smoking behavior by standard smoking machines (Kozlowski et al., 1980, 1982c; Kozlowski, 1981a, c). Why were officials afraid that a "less-hazardous cigarette" message would undermine antismoking efforts? The official antismoking efforts were directed at smoking prevention and cessation. In other words, the only message they wished to present was, "If you don't smoke, don't start; if you do smoke, stop." The addition of a further clause, "If you must smoke, at least smoke a lower- yield cigarette," was intolerable. Why should this additional message cause so much trouble? Does such a message actually spoil an antismoking campaign? 3. ARGUMENTS AND EVIDENCE, NOT FACTS AND PROOF When one deals with social problem-solving, despite the fondest wishes of practitioners and politicians, truly objective facts are rarely found. And without such facts, no incontrovertible proofs will be forthcoming. At most, one can argue and give evidence in support of the arguments (Lindblom and Cohen, 1979). So, when policymakers ask what should be dorte about the smoking and health problem, they should expect arguments and evidence rather than facts and proof. Though the policymakers and their advisors may wish to act as if revealed facts can lead to a course of action, the process depends unavoidably on .arguments.

LYNN T. KOZLOWSKI 4. ARGUMENTS AGAINST ADVOCATING LESS-HAZARDOUS TOBACCO USE I will try to present arguments against less-hazardous tobacco use that contain more meat than straw. (My position is not as partisan as might be supposed: a diverse clientrle needs to be served, and, for some, antismoking messages are probably the most useful prescription.) I will, however, feel no obligation in this section to take on the pore role of devil's advocate; some arguments for advocating less-hazardous tobacco will be touched upon while presenting the arguments against less-hazardous tobacco use. Damage to Cessation and Prevention Efforts It is widely believed that to advocate less-hazardous tobacco use is to undermine antismoking efforts. The rationales behind this belief are no doubt complicated and several. The most prominent concerns the information-pro- cessing abilities and motivation of smokers. Although support could be cited from psychological research on human information-processing abilities (e.g., Nisbett and Ross, 1980), the prevailing belief probably rests more squarely on common truisms such as People beheve only what they want to beheve, People want to have their cake and eat it too," or "People will want to take the easy way out." The smoker, it is thought, will gather from the less-hazardous tobacco message that it is acceptable to continue to use tobacco and will tend to ignore the advice that less-hazardous use should be employed only by those who cannot or will not give up tobacco use entirely. A related argument is that complicated messages will not be as persuasive or as memorable as simple messages: the less-hazardous use message, then, complicates the overall message to the det- riment of the antismoking message. These first two arguments concern problems with the reception of the smok- ing and health message. Another line of argument holds that in a world of limited resources one cannot do all that one might like to do to reduce the smoking and health problem. In terms of priority rankings, prevention and cessation activities are seen then as more important than reduced-risk activities. Will Recruitment to Tobacco Use be Encouraged by the Availability of Less-Hazardous Tobacco Products? No one knows the extent to which con- cern about the health consequences of tobacco use acts to deter those who are otherwise tempted to take up tobacco. One line of research does indicate that women, in particular, find it easier to take up smoking, given the modern, "milder" low-yield cigarettes (Silverstein et al., 1980, 1982). If advocacy of less-hazardous tobacco use adds enough recruits to the ranks of tobacco users, then the reduced risks to the individual user could be outweighed by the greater g~OS~9~90g TOBACCO USE 317 number of individuals at risk (see the section on the Prevention Paradox below). Trends in recruitment to tobacco use should be monitored. Will Tobacco Users Use Less-Hazardous Products Instead of Quit- ting? No one knows how many smokers would have given up tobacco use entirely if they had not known of the option of less hazardous use. Some smokers might switch to low-yield cigarettes to allay the pesterings of associates about the health consequences of smoking. However, it is doubtful that these individ- uals would be willing to give up tobacco entirely, unless greater social pressure were put on them. The group of smokers to be most concerned about is those who would have been able to abstain if they had not been offered the promise of reduced-risk tobacco use. I know of no estimates of how many individuals have been lost to smoking cessation or prevention because of the availability of presumably safer ways to use tobacco. A high priority should be given to empirical research that would estimate the size of the problem. Also, a high priority should be given to de- termining how to present the risks of tobacco use to individuals in ways that will have the greatest impact on health care decisions and health care behavior (cf. Slovic et al., 1977). Even if many individuals are lost to tobacco abstinence because of the less-hazardous tobacco use message, it does not follow that, therefore, the costs of the treatment outweigh the benefits. Being Faithful to One's Job Description It can be as important to know what is not part of one's job as it is to know what is part of one's job. Although tobacco once was a product that was crucial to the practice of the healing arts (Stewart, 1967), modem physicians believe that it is not within their job description to, in effect, advocate the use of any tobacco product: if less-hazardous tobacco use is to develop, it is thought to be up to the tobacco companies to be the advocates and developers. For the medical profession in general, tobacco has become an evil substance that is totally unfit for human consumption, unlike certain other potentially hazardous products (e.g., eggs, whole milk, salt, and sugar) about which the medical profession is willing to make recommendations concerning less hazardous use. For some reason, the tobacco industry has been especially easy to identify as the enemy, perhaps because of deep-set Calvinist convictions about the sin of drug use. Though a physician might be comfortable advising a low-salt or low-sugar diet (knowing that a no-added-salt or no-added-sugar diet, though probably less-hazardous, would receive little compliance), this same physician could not recommend the use of a less-dangerous tobacco product (knowing that abstinence may also result in little compliance). I juxtapose the tobacco and the food industries to illustrate an ironic inconsistency in the practice of public health and medicine: all these products may be optional but some are much more optional than others.

3 t 8 LYNN T. KOZLOWSKI • The Less-Hazardous Message is Already Well-Known The antismoking messages of prevention and cessation may inadvertently and unavoidably support the cause of less-hazardous tobacco use. Smokers on their own might tend to adop.t less hazardous uses of tobacco in response to clear messages that they should stop using tobacco. In other words, the message of less-hazardous tobacco use might occur by default as the antismoking message is spread. Similady, it can he argued that the less-hazardous tobacco use message is already well-known, because of the publicity surrounding the tar and nicotine yields of cigarettes. Standard tar and nicotine ratings have been supplied by governments, in part to encourage the use of lower tar and nicotine cigarettes by those who do not stop smoking (Friedman, 1975). The modem "tar derby" emphasizes that lower yield is better. Even if one considers the low-yield cigarette as the paragon of less hazardous tobacco products, the less-hazardous-tobacco- use message has been spread mainly in a superficial and dangerous way. (See Kozlowski, 1984, for a discussion of applications of less-hazardous tobacco therapies.) It is not enough simply to point a tobacco user to different products: advice and assistance should be given to help the user reduce exposure to toxic tobacco products. Less-Hazardous Tobacco Use as Boondoggle One argument against advocating less-hazardous tobacco use is that the promise of reduced risks is more apparent than real. In other words, the rec- ommendation to use less-hazardous products should not be made because there are no truly less-hazardous tobacco products. Low-Yield Cigarettes. Low-yield cigarettes are at the same time the most popular and the most questionable of the presumably less-hazardous tobacco products. Some evidence indicates significant, but small, reductions in risk to be gained from a switch to low-yield cigarettes (e.g., Lee and Garfinkel, 1981; Vutuc and Kunze, 1982); other evidence indicates no reductions in risk (e.g., Castelli et al., 1981; Kaufman et al., 1983; Robinson et al., 1982). The present evidence is far from conclusive (Russell et al., 1980b; Kozlowski et al., 1982b; U.S.D.H.H.S., 1981). It is possible that long-term use of low-yield cigarettes is required before a beneficial reduction in smoke exposure is seen and that those who are forced to switch brands are less likely to compensate for reduced yields than those who switch on their own (Russell et al., 1982). All cigarettes, along with other smoking tobaccos, make it difficult for users to know exactly what they are ingesting from these products (Kozlowski, 1984). It is not possible simply to read a product label and thereby know what one is getting from a cigarette or pipe. Actual smoke intake depends more on the details of a smoker's behavior (number of puffs, volume of puffs, depth of inhalation): 880t~9890~, LESS-HAZARDOUS TOBACCO USE 319 more on the smoker than on the product (Kozlowski, 1983). Cigarette smoke is more often inhaled than any other kind of tobacco smoke (U.S.D.H.E.W., 1979), and therefore cigarette smoke presents special risks to the lungs. There is also the question of risks to those who associate with or are exposed to smokers when they are smoking (U.S.D.H.H.S., 1982). AIso, the use of smoking tobaccos carries the risk of fires, and resultant death and suffering for both active and passive smokers (Bed and Halpin, 1978). Given (1) the controversy over the epidemiologic effects (in both active and passive users of tobacco smoke), (2) the difficulty in monitoring dosage, (3) the problem of inhalation, and (4) the issue of fire hazards, it is not easy to he sanguine about low-yield cigarettes as a treatment for the health consequences of smoking. Especially in light of other, more promising options for less-haz- ardous tobacco use, I am inclined to he very.pessimistic about the value of low- yield cigarettes. Even if low-yield cigarettes are poor less-hazardous tobacco products, it does not follow that other less hazardous tobacco treatments are therefore in- effective. The failure of one pharmaceutical is no grounds for closing down the pharmacopoeia. Pipes and Cigars. The available evidence suggests that pipes and cigars are less hazardous than cigarettes (Doll and Peto, 1976; U.S.D.H.E.W., 1979). People who start (and stay) with pipes or cigars tend not to inhale and hence reduce the exposure of their lungs to toxic smoke products. It is unclear whether smokers who turn from cigarettes to pipes and cigars continue the habit of inhaling. Some researchers have found inhalation among so-called secondary pipe or cigar smokers (Castledon and Cole, 1973; Turner et al., 1977, 1981); others have not found evidence of substantial inhalation by secondary pipe or cigar smokers (McCusker et al., 1982; Wald et al., 1981). Even if secondary pipe and cigar smokers do inhale, the epidemiologic evidence indicates that, while these smokers are at greater risk of dying than are primary pipe or cigar smokers, they are at a lower risk than those who continue to smoke cigarettes (Doll and Peto, 1976). If people were encouraged from the start to smoke pipes or cigars rather than cigarettes, this problem of inhalation among pipe or cigar smokers might not arise. Smokeless Tobaccos. There is really no dispute about whether smokeless tobaccos present fewer hazards to the user than do smoking tobaccos (Harrison, 1964; Russell et al., 1980a). Smokeless tobaccos expose the lungs to essentially no tobacco toxins. No carbon monoxide and no tar is produced. The oral cancers associated with oral smokeless tobaccos are substantially less lethal and are more easily diagnosed than lung cancers (U.S.D.H.E.W., 1979). In addition, smoke- less tobaccos pose no problems of second-hand smoke and no risks of fire. Clearly risks are reduced, but the residual risks are substantial enough to cause some authorities to refuse to advocate their use (Christen et al., 1979). A subclass of the boondoggle objection then is that reductions in risk are too small to warrant

320 LYNN T. KOZLOWSKI " support. (For a discussion of nicotine-containing chewing gum as possibly the least hazardous of the smokeless "tobaccos," see Kozlowski et al., 1982a, and Kozlowski, 1984.) 5. ARGUMENTS FOR ADVOCATING LESS-HAZARDOUS TOBACCO USE Job Descriptions Reconsidered Health Care Providers. Health care providers are sometimes preoccupied with their roles as opinion leaders and "moral forces" within their communities. Their function as authority figures can even obscure the more central parts of their job descriptions. Is it not best to try to do what one can to reduce death and disability in those who continue to use tobacco? Is not the reduction of death and disability a fundamental part of the job description of a health care provider? Tobacco Product Providers. For the health care provider to consider the support of less-hazardous tobacco use as a job for the tobacco industry may be naive. It could be risky to lee, re the development of less-hazardous products to an industry whose life's blood is the cigarette. Although the tobacco industry is best-suited technically to developing less-hazardous tobacco products, it should not be forgotten that it has a business to protect. Also, the tobacco industry has steadfastly denied that tobacco use causes any medical problems; this hardly puts them in a position to invest much in the development of products that reduce hazards that they assert are not there to begin with. Community Health and the Prevention Paradox Physicians who are not specially trained in public health and preventive medicine are apt to make a category mistake when considering the issue of less- hazardous tobacco use. This category mistake [i.e., allocating concepts to a category to which they do not belong (Ryle, 1949)] consists of mistaking the public health issue for a personal health issue written large. As a matter of personal health care, for a physician to recommend the less-hazardous use of tobacco can be seen (and felt) as a failure to use the positive powers of one's practice. It does not follow that a small benefit to the health of the individual will constitute a small benefit to the health of the community. Dr. William Castelli (1981) made the mistake of removing the less-hazardous tobacco use argument from the public health domain and placing it in the physician's office. Taking the most generous estimate from the report of Lee and Garfinkel (1981), Castelli noted that a pack-a-day smoker who switched from unfiltered to filtered cigarettes reduced his or her risk of lung cancer from 20 times that of a nonsmoker to 15 ~g0999890g iS-HAZARDOUS TOBACCO USE 321 times that of a nonsmoker. Castelli wrote: "I do not personally get much sat- isfaction encouraging someone to pursue a habit which increases the risk of lung cancer 15 times" (p. 642). This does describe the situation from the physician's perspective; however, from the perspective of one interested in community health, the satisfactions may be obvious: if 2000 pack-a-day smokers had been dying each year from lung cancer, now 1500 smokers would be dying. Five hundred people would still be alive; 25% fewer smokers would be dying from lung cancer. Rose (1981) describes the "prevention paradox .... a measure that brings large benefits to the community offers little to each participating individual." A treatment that is worthwhile and practical for the community may have trivial influence on the individual. Conversely, the treatment that may have the most benefit for the individual may be impractical and hence of little use for the community. Rose uses the treatment of hypertension as an example. Extremely high blood pressure can be controlled with drugs, but relatively few individuals have extremely high blood pressure. If the average diastolic blood pressure of the community were reduced by just 7-8 mm Hg (say, by altering the die0, then the number of disorders due to blood pressure would decline as much as if all those with pressures of 105 mm Hg or more were treated in a 100% effective way. The less dramatic therapy reaps appreciable net benefit because of the large number of people involved with the treatment. The continuing discussion of the low-yield cigarette has often ignored the relevance of this paradox for tobacco use (e.g., Marks, 1982). Basically, the principle behind the paradox is that small effects on a large enough scale can produce more net benefits than can effects of heroic proportions on a small scale. This principle is also manifest in Russell's (Russell et al., 1979) advice on the benefits of physicians' advice on smoking cessation. Each physician will have relatively little success in persuading patients to give up cigarettes, but given the number of physicians available to spread the word, the net effects could be many times larger than the effect of more expensive alternative therapies. Of course, one of the key assumptions involved with employing the pre- vention paradox as an argument for less-hazardous tobacco use is that the number of people enjoying the small benefit must be large enough to add up to a sub- stantial net benefit. One might think that if the prevention and cessation efforts became highly successful, there would be few tobacco users left to enjoy the small benefits of less-hazardous tobacco use; however, it must be remembered that for those individuals who continue to use tobacco, cessation is, by definition, not an alternative treatment to less-hazardous use. The Limits of Prevention and Cessation None of the arguments for the advocacy of less-hazardous tobacco use should be used to argue against the deployment of prevention and cessation programs. The tess-hazardous tobacco use message has no war with the anti-

322 LYNN T, KOZLOWSKI smoking message; in fact, the relationship between the two is symbiotic. As noted above, the less-hazardous use message is best viewed as an effort to deal with the failures of other efforts. Prevention programs in the schools have received great attention in recent years (e.g., Evans et al., 1981). Though these programs have shown some success in reducing recruitment to smoking, at least during the school years, no one would argue that any program has discovered a certain technique for drastically reducing the number of smokers in high school, say, below the level of 10% of the students. Similarly, formal smoking cessation treatment programs find in general that 80% of their clients will relapse to smoking within I year (Raw, 1978). One estimate of how well smokers succeed at stopping smoking after repeated attempts on their own indicates that about 40% will fail to abstain in the long run (Schachter, 1981). One of the best studies of the overall impact of the antismoking campaign comes from a cohort analysis of smokers in the United States population (Warner and Murt, 1982). Based on the percentages of smokers in different age groups before the antismoking campaign really got started (before the 1964 Report of the U.S. Surgeon General), Warner and Mutt estimated how many smokers would have been expected in these same age groups had the antismoking cam- paign not taken place. In 1964, 67% of the 21- to 24-year-old men were smokers; in 1975, only 41% of the 21- to 24-year-olds were smokers. In 1964, 42% of the 21- to 24-year-old women were smokers; in 1975, 34% were smokers. They estimate that, if it were not for the antismoking campaign, 61% of men 18-27 years old would have been smokers in 1978; only 39% of this group were smokers in 1978, a difference of 22 percentage points. For women of the same age, they estimate that 49% would have been smoking; 37% actually were smokers in 1978, a difference of 12 percentage points. Though these figures indicate sub- stantial success for antismoking efforts, they also clearly show that a potential market exists for less-hazardous tobacco. The Promise of Diminishing Returns. Tobacco users differ in how de- pendent they are on tobacco. A number of studies have shown that cessation interventions are more successful with less-dependent tobacco users (e.g., Fa- gerstrom, 1982; Kozlowski et al., 198 I). One of the clearest implications of this finding is that the pool of continuing smokers is becoming more likely to contain more-dependent tobacco users. [The population of smokers is made up increas- ingly of fewer and heavier smokers, (U.S.D.H.H.S., 1981)]. In other words, the antismoking campaign has probably tended to remove those who are most easily removed from the ranks of smokers. Those who remain are likely to be a hard core of recalcitrant and perhaps "reactant" smokers. Reactance is a tech- nical term that refers to an individual's assertion of freedom of action when faced with attempts to restrict that freedom (Brehm, 1966). Less-hazardous tobacco use may be one of the few treatments available for these smokers. ~809E9B90E LI:SS-HAZARDOUS TOBACCO USE 323 A Question of Class. Recently, there has been a growing concern about the social inequalities of health care delivery systems (e.g., Morris, 1980). If one looks carefully at smoking statistics, it is apparent that, in general, those of lower socioeconomic status are more likely to use tobacco than are those of higher socioeconomic status. [An exception is that higher-class women are smok- ing more than lower-class women (U.S.D.H.E.W., 1979)]. If one looks at some especially disadvantaged groups, one finds, for example, that in Canada only 23% of teenagers (ages 15-19) who are still in school are daily smokers, whereas 48% of those who are no longer attending school (essentially high school drop- outs) are daily smokers (Health and Welfare Canada, 1981). The same report finds that those with low levels of education, and those who are unemployed or in low-status jobs, are more likely to be current daily smokers. Moody (1980) finds that those from lower socioeconomic groups also take more puffs per cigarette and are exposed to more daily tar than are those from higher socio- economic groups. Has the antismoking campaign been less successful in reaching the lower classes? Has the antismoking campaign been less successful with those of low socioeconomic status that it has reached? Are those of low socioeconomic status more likely to be dependent on tobacco? Certainly it is fair to say that a school dropout may miss out on many of the antismoking efforts in the schools. When one has lost a job, is one also inclined to hold on to the compassion to be found in tobacco? Whatever the reasons, socioeconomic lines have indicated systematic limits to the power of current antismoking efforts. 6. THE LIMITS OF LESS-HAZARDOUS TOBACCO USE Prevention and cessation efforts are not alone in having limited power and success. Much of the speculation about less-hazardous tobacco use as a treatment has not, in fact, received empirical test. Some of these good ideas may not work in practice. The epidemiology of tobacco-related diseases can only serve as a guide to possible treatments. Epidemiologic samples are self-selected as tobacco users: epidemiologic studies generally show no more than correlations between tobacco use and disease. No one knows, for example, how the health consequences of smoking might change in a group of cigarette smokers who were randomly assigned to take up pipe smoking. No one knows how many participants in such a study would be able to comply with their instructions. We do have evidence that secondary pipe and cigar smokers do have less risk of disease than those who continue to smoke cigarettes (Doll and Peto, 1976); but we do not know if the change to pipes or cigars is the cause of the reduced risk. Perhaps those cigarette smokers who do change to pipes or cigars are very different (e.g.,

324 LYNN T. KOZLOWSKI constitutionally) than those who continue as cigarette smokers (Seltzer, 1972). Despite these reservations, the epidemiologic literature does form the basis for predictions about less-hazardous tobacco use. Technical versus Behavioral Interventions Technical interventions depend upon changes in the tobacco product. Be- havioral interventions depend upon changes in conduct. If, for example, a less- toxic tobacco tar could be developed, then, one might find reductions in lung cancer incidence even if there were no changes in tar intake. It has been argued that modern tars are less toxic (milligram for milligram) than the tars of the 1950s and that this reduced toxicity might account for the reduced incidence (e.g., Gori, 1976). If modern tars are less toxic or can be made less toxic, one would have a less-hazardous tobacco use treatment for the smoking and health problem that would (assuming that the compassion remained) pose essentially no problems of patient compliance. The ideal technical intervention involves the modification of a product that the tobacco user is already using. Being able to reduce the intrinsic risks of a product that tobacco users will not use provides little treatment for the tobacco and health problem. Some behavioral interventions might be directed to per- suading the tobacco user to use a less-hazardous product. Other behavioral interventions will be directed to the less-hazardous use of the product currently being used. Each of these kinds of behavioral intervention is truly easier described than done. For a discussion of some of the challenges involved with behavioral interventions, see Kozlowski (1984). Diet, Drugs, Occupation, and the Risks of Tobacco There may be adjunctive ways to engage in less-hazardous tobacco use. The epidemiologic literature suggests that it would be advisable for smokers to change other behaviors to reduce the health consequences of tobacco use. This literature is, for the most part, suggestive rather than conclusive. Those who work with asbestos and smoke cigarettes are at especially high risk of lung disease (see U.S.D.H.H.S., 1982, for a review). Similarly, cigarette smoking and birth control pills may act synergistically to increase the risk of cardiovascular disease in women (see U.S.D.H.H.S., 1980). Tobacco and al- cohol appear to act synergistically to increase the risk of cancers of the mouth, pharynx, larynx, and esophagus (see U.S.D.H.H.S., 1982, for review). It is possible that a continuing tobacco user could reduce the health consequences of tobacco use by being careful to avoid alcohol, asbestos, and birth control pills: in terms of practicality, it should not be prejudged which of these activities is optional for given individuals. As a positive measure to reduce the risks of cancer in the tobacco user, there is growing evidence that a diet rich in pro-vitamin A 9g0939990g LESS-HAZARDOUS TOBACCO USE 325 has a protective effect against lung cancer (Doll and Peto, 1981; Shekelle et al., 1981). Less-hazardous tobacco use might, then, be established by modifying (1) the tobacco use, (2) a cofactor for risk, or (3) both. 7. LEGITIMIZING THE TOPIC AND THE NEED FOR RESEARCH The arguments for and against advocating less-hazardous tobacco use have certainly not been exhaustive. This chapter has tried to legitimize the study of less-hazardous tobacco use as a beneficial treatment for the smoking and health problem. Despite the impression that some antismoking readers might have, I am uneasy about a blanket endorsement of the less-hazardous-tobacco-use ther- apy. Data may indeed emerge in the future that will show the less-hazardous movement to have been ill-advised. Current research should, however, not fear to show both the advantages and disadvantages of all aspects of the war against tobacco-related maladies. As is the case in many areas of applied research, in this area it is not possible to wait until all of the data are in to decide what should be done about less- hazardous tobacco use. As a self-administered therapy, "less-hazardous tobacco use" exists already. Many tobacco users will not be persuaded to give up their use of tobacco, despite the best efforts of the antismoking campaign. If there are ways to reduce obvious errors in this self-administered therapy, then the consumers of these therapies should know about them (Kozlowski, 1982b). Without research on the would-be forms of less-hazardous tobacco use, we are not able to establish their actual, rather than supposed, net worth. ACKNOWLEDGMENTS The author thanks R. Frecker, C. P. Herman, S. Herling, L. Jelinek, M. Pope, and K. Wagner for their assistance. REFERENCES American Psychiatric Association, 1980, Diagnostic and Statistical Manual of Mental Disorders, 3rd. Ed. American Psychiatric Association, Washington, D.C. Bain, J., 1896, Tobacco in Song and Story, Caldwell, New York. Bed, W. G., and Halpin, B. M., 1978, Human fatalities from unwanted fires, The Johns Hopkins University--Applied Physics Laboratory, December, p. 8. Brehm, J. W., 1966, A Theory of Psychological Reactance, Academic Press, New York. Brooks, J. C., 1952, The Mighty Leaf: Tobacco through the Centuries, Little Brown, Boston. Castelli, W. P., 1981, Filter cigaretles and heart disease, Lancet 2:642. Castellt, W. P., Dawber, T. R., Feinleib. M., Garrison, R. J., McNamara, P. M., and Kannel, W. B., 198 I, The filter cigarette and coronary heart disease: The Framingham study, Lahcet 2:109.

326 LYNN T. KOZLOWSKI "Castledon, C. M., and Cole, P. V., 1923, luhalation of tobacco smoke by pipe and cigar smokers, Lancet 2:21-22. Christen, A. G., Arr~trong, W. R., and McDani¢l, R. K., 1979, Iraraoral l~u~oplakta, perktontal breakdown, and tooth It)as in a anuff dipper, J. Am. Dent. Assoc. 98:584--586. Cohen, M. R., and Nagel, E., 1962, An Introduction to Logic, Harconrt, Brace and World, New York. Doll, R., and Peto, R., 1976, Mortality in relation to smoking: 20 years observation on male British doctors, Br. Med. J. 2:1525-1536. Doll, R., and Poto, R., 1981, The causes of cancer:. Quantitative estimates of avoidable risk~ of cancer in the United States today, J. Natl. Cancer Inst. 66(6):I 192-I 308. Evans, R., Hill, P. C., Rnines, B. E., and Henderson, A. H., 1981, Current behavioral, social, and educational Im3graras in control of smoking: A selective critical review, in: Perspec~h, es on Behavioral Medicine (S. M. Weiss, J. A. Head, and B. H. Fox, eds.), pp. 261-284, Academic I:~ss, New York. Fagerstr6m, K.-O., 1982, A comt:uu'ison of psychological and pharmacological treatment in smoking cessation, J. Behav. Med. 5(3):343-351. Fischer, D. H., 1970, Historian's Fallacies: Toward a Logic of Historical Thought, Harper and Row, New York. Friedman, K. M., 1975, Public Policy and the Smoking-Health Controversy, D. C. Heath and Company, Lexington, Mass. GarL J. J., and Schneider)nan, M. A., 1979, "Low-risk" cigarettes: The debate continues, Science 204:690-691. Gerstein, D. R., and Levison, P. K., eds., 1982, Reduced Tar and Nicotine Cigarettes: Smoking Behavior and Health, Committee on Substance Abuse and Habitual Behavior and Social Sciences and Education, National Research Council, National Academy Press, Washington, D. C. God, G. B., 1976, Low-risk cigarettes: A prescription, Science 194:1243-1245. God, G. B., 1980, A summary appraisal, in: Banbury Report 3. A Safe Cigarette? (G. B. God, and F. G. Back, eds.), Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. God, G. B., and Lynch, C..L, 1978, Toward less hazardons cigarettes, JAMA 240:1255- 1259. Harrison, D. F. N., 1964, Snuff--its use and abuse, Br. Med. J. 2:1649-1651. Health and Welfare Canada, 1981, The Heahh of Canadians. Report of the Canada Health Survey, Minister of Sul~ply and Services, Ottawa, Canada. (Catalogue 82-538E). Kanfman, D. W., Helmrich, S. P., Rosenbarg, L., and Miettinen, A. S., 1983, Nicotine and carbon monoxide content of cigarette smoke and the risk of myocardial infarction in young men, N. Engl. J. Med. 308:409-413. Kozlowski, L. T., 1981a, Tar and nicotine delivery of cigarettes: What a difference a puff makes, JAMA 245(2):158-159. Kozlowski, L. T., 1981b, Smokers, non-smokers, and low-tin" smoke, Lancet 1:508. Kozlowaki, L. T., 1981c, Application of some physical indicators of cigarette smoking, Addict. Behav. 6:213-219. Kozlowski, L. T., 1982a, The determinants of tobacco use: Cigarette smoking in the context of other forms of tobacco use, Canad. J. Public Health 73:236-241. Kozlowaki, L. T., 1982b, Tar and Nicotine Ratings May Be Hazardous to Your Health, Alcoholism and Drug Addiction Research Foundation, Toronto. Kozlowski, L. T., 1983, Perceiving '~he risks of low-yield vontilated-filter cigarettes: The problem of hole-blocking, in: Proceedings of the International Workshop on the Analysis of Actual vs. Perceived Risks, (V. Covnilo, W. G. Flamm, J. Rodricks, and R. Tardiff, eds.), Plenum Press, New York. Kozlowski, L. T., 1984, Pharmacological approaches to smoking modification, in: Behavioral Health: A Handbook of Health Enhancement and Disease Prevention (J. R. Matarazzo, N. E. Miller, S. M. Weiss, J. A. Herd, and S. M. Weiss, eds.), .Iobn Wiley, New York. L80~,9890~ LESS-HAZARDOUS TOBACCO USE 327 Kozlowski, L. T., Rickert, W., Robinson, .I., and Grunherg, N. E., 1980, Have tar and nicotine yields of cigarettes changed? Science 209:1550--I 55 I. Kozlowski, L. T., Director, .L, and Hafford, M. A., 1981, Tobacco dependence, restraint, and time to the first cigarette of the day, Addict. Behav. 6:307-312. Kozlowski, L. T., Appel, C.-P., Frecker, R. C,, and Khouw, V,, 1982a, Nicotine, a prescdbable drug available without prescription, Lancet 1:334. Kozlowski, L. T., Frecker, R. C., and Lei, H., 1982b, Nicodnc yields of cigarettes, plasma nicotine in smokers and public health, Prey. Med. 11:240-244. Kozlowski, L. T., Rickert, W. S., Pope, M. A., Robinson, I. C., and Frecker, R. C., 1982c, Estimating the yield to smokers of tar, nicotine, and carbon monoxide from the "lowest-yield" ventilated-filter cigarettes, Br. J. Addict. 77:159-165. Lakoff, G., and .lohnson, M., 1980, "Metaphors We Live By," University of Chicago Press, Chicago. Lee, P. N., and Garlinkel, L., 1981, Morality and type of cigarette smoked, J. Epidemiol. Commun, Health 35:16-22. Lindblom, C. E., and Cohen, D. K., 1979, "Usable Knowledge," Yale University Press, New Haven, Connecticut. ' Marks, L., 1982, Policies and postures in smoking control, Br. Med. J. 284:391-395. Marx, .I.L., 1978, Health officials fired up over "tolerable" cigarettes, Science 201:795-798. McCusker, K., McNabb, E., and Bone, R., 1982, Plasma nicotine levels in pipe smokers, JAMA 248(5):577-578. Moody, P. M., 1980, The relationship of quantified human smoking behavior and demographic variables, Sac. Sci. Med. 14A:49-54. Morals, J. N., 1980, Are health ~rvices important to the people's health? Br. Med. J. 280:167-168. Nlshett, R., and Ross, L., 1980, "Human Inference: Strategies and Shortcomings of Sncial ,Iudg¢- ment," Prentice Hall, Eaglewood Cliffs, New Jersey. Raw, M., 1978, The treatment of cigarette dependence, in: Research Advances in Alcohol and Drug Problems, Vol. 4, (Y. Israel, F. B. Glaser, H. Kalant, R. E. Popbam, W. Schmidt, and G. Smart, eds.), pp. ,M1-485, Plenum Press, New York. Robinson, J. C., Young, J, C., and Rickert, W. S., 1982, A comparative study of the amount of smoke absorix~ from low yield ('less hazardous') cigarettes, part 1: Non-invasive measures, Br. J. Addict. 77:383-397. Rose, G., 1981, Strategy of prevontion: Lessons from cardiovascular disease, Br. Med. J. 252:!847-1851. Royal College of Physicians, 1977, Smoking or Health, Pitman, London. Russell, M. A. H., Wilson, C. Taylor, C., and Baker, C. D., 1979, Effect of general practitioners' advice against smoking, Br. Med. J, 2:231-235. Russell, M. A. H., Jarvis, M. J., and Feyorahend, C., 1980a, A new age for snuff7 Lancet 1:474--475. Russell, M. A. H., Jarvis, M., Iyer, R., and Feyerahend, C., 1980b, Relation of nicotine yield of cigarettes to blo~l nicntine concentrations in smokers, Br. Med. J. 280:972-976. Russell, M. A. H., Sutton, S. R., lye.r, R., Feyerabend, C., and Vesey, C. J., 1982, Long-term switching to low-tar low-nicotine cigarettes, Br. J. Addict. 77:145-15g, Ryle, G., 1949, The Concept of Mind, Barnes and Noble, New York. Scbachter, S., 1981, Self-treatment of smoking and obesity, Canad. J. Public Health 72:401-406. Sehachter, S., Silverstein, B., Kozlowski, L. T., Pertick, D., Herman, C. P., and Liebling, B., 1977, Studies of the psychological and phannacelogical determinants of smoking, J. F.~. Psychol. [Gen.] 106:3-40. Seltzer, C. C., 1972, Differences between cigar and pip~ smokers in healthy white veterans, Arch. Environ. Health 25:187-191. Shekelle, R. B., Liu, S., Raynor, W. J., Jr., Lepper, M., Maliza, C., Rossof, A. H., Paul, O., Shyrock, A. M., and Stamler. L, 1981, Dietary vitamin A and risk of cancar in the Western Electric study, Lancet 2(8267):1185-1190.

LYNN T. KOZLOWSKI $ilverstein, B., 1982, Cigarette smoking, nicotine addiction, and relaxation, J. Pers. Soc. Psychol. 42(5):946-950. SHverstein, B.. Feld, S., and Kozlowski, L. T., 1980, TI~ availability of iow-nicotine cigarettes as a cause of cignre~ smoking arrm, rtg teenage female.s, J. Health So~. Behav. 21:383-388. Silverstein, B., Kelly, E., Swan, J., and Koziowski, L. T., 1982, Physiological predisposition toward becoming a cigarette smoker: Experimental evider~m for a sex diffczence, Addict. Behav. 7:83-86. Slovic, P., Fischoff, B., and Lichtenstein, S., 1977, Behavioral ¢~cision theory, Armu. Rev. Psychol. 28:1-39. Stewart, G. G., 1967, A history ofth~ medicinal uses of tolmcco, Med. Hist. 11:228-268. Turner, L A. McM., Sillett, R. W., and McNicol, M. W., 1977, Effect of cigar smoking on carhoxyha~moglobin and plasma nicotine concerarations in primary pipe and cigar smokers and ex-cigm'ett* smokers, Br. Med. J. 2:1387-1389. Turner, J. A. McM., Sillett, R. W., and McNicol, M. W., 1981, The inhaling habits of pipe smokers, Br. J. Dis. Chest 75:71-76. U. S. Department of Health, Education and Welfare, Public Health Service, 1979, Smoking and Health: A Report of the Surgeon General, (Publication No, (PHS) 79-50066), U. S. Government Printing Office, Washington, D. C. O. S. l~nt of Health and Human Services, Public Health Service, 1980, The Health Con- sequences of Smoking for Women, U. S. Government Printing Office, 1980 0-326-003, Wash- ington, D. C. U, S. Department of Health and Human Services, Public Health Service, 1981, The Health Cen- sequences o/Smoking: The Changing Cigarette, U. S. Government Printing Office, Washington, D.C. U. S. Department of Health and Human S~wices, Public Health Service, 1982, The Health Con- sequences o/Smoking: Cancer, U, S, Government Printing Office, 1982 0-367-198/579, Wash- ington, D. C. Vntuc, C., and Kurme, M., 1982, Lung cancer in women in relation to tar yields of cigarettes, Prey. Med. 11:713-716. Waid, N. L, Idle, M., Boreham, L, Bailey, A., ~nd Van Vunakis, H., 1981, Serum cotininc icveis in pipe smokers: Evidence against nicotine as a cause of coto~ry heart disea~, Lance~ 2:775-777. Waruer, K. E., 1979, Toward less hazardo~ts cigarettes, JAMA 241:2143. Warner, K. E., and Mutt, H. A., 1982, Impact of the anti-smoking campaign on smoking prevalence: A cohort analysis, J. Public Health Policy 3:374-3~). Index Accidental injuries as a consequence of alcohol consumption, 150, 164-165, 168, 173-174 Acupuncture, effect on endorphin levels in human drag addiction, 1 I Adeaylate cyclase, effect of opiates o~, 7-9 Ad~icotrophic hormone (ACTH), effect of naloxone on, 1 I Age as variable in study of driver impairment and alcohol-related collisions, 242, 243, 247-248, 250-251 as variable in study of etiology of alcoholism, 280-281 as variable in study and treatment of nonopiate dependency, 292 as variable in study of tobacco use, 322, 323 Alcohol abuse animal models of, 36-38 definitions, 185-187 environmental factors leading to, 52-55 etiology, contribution of prospective studies to understanding of, 265-285 genetics, 27-28, 47-87, 269 psychological and social consequences of, 164, 271 typology, 279 assessment of use by patients in treatment, 183-203 consumption effects of drugs on, 30-31 methods of studying, 24-27 Alcohol (cont.) censumption (cont.) and occurrence of ischemic heart disease, 99-140, 162 techniques and problems in measurement of, 209-223 metabolism, 58-61, 195-196, 234 as reinforcer, 25.26, 28, 29, 32, 33-34 role In driver impaimmm and collisions, 227-258 studies of self administration by experimental ~mimals, 23-39 physical health probicms associated with, 149-176 Alcohol dehydrogenase (ADH) in alcohol metabolism, 59 Aicohol Dependence Scale (ADS), 188 Alcohol dependence syndrome, 186; see also Alcohol abuse; Alcohol consumption AMoholics Anonymous, 284-285 Alcohol Level Evaluation Roadside Tester (A.LE.R,T.), 231,242 Aldehyde dehydrogenase (ALDH), and facial flushing, 60-61 Alpha-methyl-p-tyrosiue, and alcohol intake, Amphetamines, 302 Anemia, as consequences of alcoholism, 150 Angina pectoris (AP), association with alcohol consumption, 116 Antidepressants, ~cyclic, in treatment of nonopiate drag abuse, 295 Anxiety as consequence of alcohol consumption, 271 etiologic role in alcoholism, 272 329

2063628039

O~ONZ9890Z CHAPTER Keynote Address: The Control of Lung C¢ ncer t RICHARD PETO and RIC HARD DOLL Cancer Studies Unit, Radcliff, Infirmary. Oxford OX2 6HE, Great Britain ICRF Cancer Epidemlo|ogy tnd Clinical Trials Unit, Radcliffe Infirmary. Oxford OX2 6lIE, i ;rear Brilain ABSTRACT During the 1980s, alxmt 1 million Americans anti 2 m Ilion European.'~ are likely to die or tobacco-induced lung cancer, and there is increasingly' h.eaw marketing of manufactured cigarettes in developing countries. This will produce large increases in lung cancer in the next century. These increases are inevitable, just as the increases in United States cigarette usage 40 and more mars ago are responsible for the la:'ge increases in United States lung cancer rates t~May. This increased usage overwhelms the beneficial effects of cigarette tar level reductions. By contt ast, in Britain and Finland mat, lung cancer rates in middle age had already stabilized Ix'fore the large tar reductions begat , and in early middle age the maid lung cancer death rates have already halved, and are s ill dropping fast. Unless recent tar reductions have implausibly large adverse effects on v t~ular or respiratory disease, they have perhaps bccn underrated as practical public he.dth measnrcs. (United States and United Kingdom tar reductions still can be accelerated, md at present many other countries lag well behind.) Although the control of fossil fuel combustion product ~, ionizing radiations, asbestos, and other occupational factors is certainly worthwhile, it can tot avoid any substantial fraction of lung cancer deaths, and although substantially protc~ live nutritional modifications may exist, none has yet been reliably identified. Meanwhile. recent European experience shows that govern,ncnts can, if they wish, achieve substantial ~ cductions in both tar deliveries and cigarette nsagc without materially affecting personal fr 'edoms, and nnless one or both of tbese arc achieved in many different countries there is lltlle prospect of avoiding any substan- tial fraction of the many millions of lung cancer deaths hat threaten to occur over the next few decades. Kqy Words: Lung cancer prevention, cigarette sales tr rods, cigarette tar reduction, lung cancer trends, smoking duration," quantitative inforn~ atlon, clgarette-lung cancer inter- national cnrrelation, cigarette taxation, tobacco prlc" increases 'This article is expanded [mm a report prepared for the WHO ~ancer Unit, b.t it does not necessarily reflect the views of the WI tO or of its Cancer Unit. ~' t~8~ Vedog Cher'n~ Intemot~. Inc. L~ Conce~ C~uses acid Pa~,,,entlon.

2 Rtchord Peto and Rk~hard For the control of each type of cancer, three strategies are available--prevention, screening (early detection), and treatment. Over the past few decades cancer research has produced (or suggested) some important results in all three areas. These include, for example, the effects of hepatitis B virus on the risk of liver cancer, of lower tar cigarettes on the risk of lung cancer, of screening on the prog- nosis of cervical cancer, and of cytotoxic drugs on the treatment of Hodgkin disea~. It therefore is reasonable to be optimistic about, but impossible to predict reliably, what cancer research will produce over the next few decades. If, however, we consider the promise not of what future research may one day offer but of what present-day knowledge can already offer, then the most prac- ticable, and cost-effectlve, opportunities for avoiding premature death from cancer probably involve not screening or improved treatment, but prevention, and this is particularly true of lung cancer. This conclusion does not involve the unrealistic assumption that tobacco can be eliminated: instead, it merely assumes that cigarette sales can be somewhat reduced (eg, by politically realistic price changes, or by the type of education that already appears to have had a substantial effect on white-collar cigarette usage), that the tar delivery per cigarette can likewise be somewhat reduced, and that gro~'~ occupa- tional exposures (eg, to levels of asbestos far higher than are nowadays permissible in, for example, the United States) can be avoided. Nor does this conclusion in- volve the assumption that lung cancer is already a major health problem every- where, for it applies not only in countries where cigarette smoking has been widespread for decades (eg, the United States, where lung cancer already accounts for some 25-30% of all cancer deaths) but also in countries in which cigarette smoking has become widespread only in recent decades (eg, China, where lung cancer as yet accounts for only al:x~ut 5-10% of all cancer deaths). This is because the main rise in lung cancer produced by cigarettes may take as much as halfa cen- tury to materialize, so countries where cigarette smoking is only now becoming or has only recently become widespread can expect large increases in lung cancer dur- ing the 1990s or early in the next century unless effective actiou against the health effects of tobacco can be achieved. The reasons that the prevention of lung cancer is of such overwhelming importance are, first, that the disease is extremely common~; second, dmt it is el'he types of cancer that cause most deaths worldwide are cancers of the lung and of the stomach. Re- cent International Agency for Re~earch on Gaacer/~/HO joint estimates (1) maggest that by 1975 the annual number of new ca~s of lung cancer was already about 0.6 million (developed "west," 0.3 million; developed "east," 0.1 million; China and rest of "third" world, 0.2 million), a total similar to that for stomach cancer. For both diseases, about 90% of affected patients are likely to die. However, whereas in m~t countries stomach cancer rates are either steady or decreasing, lung cancer rates are (with a few interest{ng exceptions that may result from cigarette tar delivery reductions) either steady or increasing. ~, during the 1980s lung cancer will account for more deaths than any other tyl~ of cancer; indeed, the annual number nf lung cancer deaths is pro]mbly already, in the early 19B0~, alw~ut two- thirds of a million, and it may well exceed a million by Ihe end of the century. SEven with expensive hlgh-technotogy support, current surglcM management of lung cancer cures only about 10% of all case~, and ahbough thi~ percentage is snllic~eot to justify the widespread use of surgery (at least in developed countries), it i~ small, has sh~r~wn little change ira recent decades, and is hardly ha- proved at all by the addition of any other current cremes of therapy. l~o~Eg~90e The Control of Lung Can :e~ 3 generally incurableS; third, that effective preventiv ~" measures are already reliably known~; and last, that implementation of these me ~sures will also have a substan- tial impact on many other diseasesL Consequently, what will follow is a description not of early detection or treat- ment, but only of the preventability of lung cancer where chief emphasis must evltably be on the effects of smoking, either alon.: or in combination with other causative factors. There already have been many cxcellent reviews of the effects of tobacco on lung cancer (and, of course, on many ,,thee diseases), and of the pros- poets of controlling the smoking epidemic (:2). The~ efore, the present text is intend- ed not as a balanced (and repetitious) account ef the whole problem, but as a discussion centering on those particular aspects of the relationship between smok- ing and lung cancer that commonly engender iml,orlant misunderstandings, and of those particular aspects of prevention that corn nonly are underemphasized. For a comprehensive review of the overall heath effects of tobacco, the U.S. Surgeon-Gener,'d's 1979 report (3) may be consult.'d [although for a description of the effects of tobacco just on cancer the Surgeon-G ;neral's 1982 report (4) is much to be preferredI. For a review of the overall epidt Iniology of cancer in developed countries, the report of Doll and Peto (5) may be c~,nsulted, especially as Appendix E reviews in some detail the effects of past chanFes in tobacco usage and recent changes in cigarette tar deliveries on lung cancer trt ads. Finally, for reviews of prac- tical steps toward the avoidance of smoking, in addition to the United States Surgeon-General's 1979 report (3), the UICC (6), World Health Organization (2) and Ontario Council on Health (7) reports are im ;duable. Epldemlology of Smoking a~ld Lung Cancer The Need for Prolonge(I Exposure There are a few key features of the effects of 'trbacco on lung cancer that are slightly counterintuitive, and these are discussed ::t some length by Doll and Peto (5). Ohlef among them, and the key to any prol~ r understanding of tobacco car- clnogencsis, is the extraordinary relevance of the d tradon of smoking to hmg cancer onset rates. For example, after 45, 30, and 15 year : ofclgarette star,king, the excess annual incidence rates of lung cancer may be about 0.5%, 0.1% and under *The most effective step is to avoid increases, or to produce t rcreases, in cigarette consumption, but changes in the harmfulness of cigarettes also may be]p. There may be consld~erable pelitlcal difficulties in taking any actions that will affect cigarette consumption sub~ tantially, but it is nevertheless likely that ~omc such actions will be practicable in at least some counh ies over the next few ycar~. Moreover, substantial decrea~s in the sales-weighted mean tar deliverie.~ of manufactured cigarett¢~ are likely to be p~litically practicable in many coimtrles, as they have l'ttle economic impact on governments, rl'obaceo is al.~ an important can~ of various less common tyI cs of cancer, eg, of the mouth, pharynx, larynx, aml e~phagus, and p~bably aim of the panc~m~ and trina~ tract. Mo~ im~flantly, at least in devrlol,d cmmtries sm'h as Britain, to~c~ p~bly ]ills more ~ople from ~splrato~ and va~ulnr di~a~ than irma cancer.

l~chard Pelo and l~hard Doll 0.01%, respectively (Table I), The annual lung cancer incidence rates to be ex- pected among smokers may be. estimated by adding up a background (nonsmoker) rate, which, like the onset rates of many other types of cancer, depends strongly on age (but not, of course, on tobacco exposure), plus an excess rate, which depends Table 1. Approximate" effects of various durations of cigarette smoking on annual incidence of hmg cancer Annual excess incidence Years of Moderate Heavy cigarette smok|ng smokers smokers % 15 0.005 0.01 30 0.1 0.2 45 0.5 1 (60) (1.57) (3?) aF, stimated from data reported by Doll and Peto (8) for male British doctors. The cumulative risks would be far greater than the.~, annual risks, of course, so an eventual total of over 10% of reg.lar cigarette smokers may die of tnba¢co-induced cancer, depending on the number and type of clgaretles smoked. Annual lung cancer death rate per 100,0OO men (standardized for amount smoked) 1000 100 10 Smoked cigarettes ~lnce age . _ ,~m Never ~" - smoked - / regularly 20 40 60 80 Duration of exposure in years (age - 221/2 for smokers age - 2th for nonsmokers) Figure I. Background and excess risks: lung cancer death rates autong (a) non-smokers (lower line) in relation to age, and (b) regular cigarette smokers (upper line) in relation to ap- proximate years ofsmoking. From Ref. (9). These two lines can be used directly to indicate the approximate baekgrmmd and excess risks, for in middle and old age the lung cancer in- cidence rates anmng people who have snmked cigarettes tt~r.ughuut aduh life greatly cxrccd the rates among nonsmokers of similar age. (This might not, however, be. true for t~oplc who did not I~gin to smoke substantial nmnbe.rs of cigarettes until middle age, for the background and the excess annual risks per 10O,000 men indicated by these lines are, rcspcc- lively, approximately IOs times the fourth power of years of age 10"4 times the fourth power times years of regular cigarette smoking.) g~O~g9890E The Conlrol of Lung Car 0.3% 0.2% O.1% 0.0% ~Y Before 15 15-19 20-24 25or Never o/er Age (years) when s arted to smoke cigarettes Figure 2. Tim relevance uf star,king iu early aduh life: , lationsbip, in prospective survey data of regular smokers, between the age when regular 'igarette smoking began in early aduh life and hmg cancer deatt~ rates at age 55-64 (mean * age 60) for Unitod States males. From Ref. (,5). Data are presented separately for heavy a td for moderate smokers. strong!y on duration of regular tobacco exposure (I ut not otherwise, at least to a first approxi~nation, on age). Typical background ;md excess rates for males are depicted in Figure 1 (9), and those for females ma5 be about two-thirds as great. The most surprising consequence of the overwl'elming effects of the duration of smoking is illustrated, using real datas, in Figur,. 2, which shows how strongly the annual excess risk of death from lung cancer ;~t 60 years of age depends on whetber men started smoking at 15 or at 25 years of age tie, on whether by the age of 60 they had smoked for 45, or for only 35, yt ars). Failure to appreciate the relationship illustrated in Figure 2 has led to a v ~riety of unjustifiable conclu- sions, eg, that cigarettes do not cause lung cancer cr, less perversely, that low-tar cigarettes have at least as great an effect as high-tar ones (10); that air pollution is of comparable imp(~rtance to tobacco [see, however, Cederl6f et al (11)]; or that new on.sos tff hmg cancer (rather than the dela)cd effects of past changes in tobacco usage) are chiefly responsible for the rapid ncreases in lung cancer in re- cent years. In each case the point that often is over'ooked is that current patterns ofhmg cancer mortality rates in late middle age or ,n old age depend strongly not only on current patterns of tobacco usage, but als ~ on the patterns of cigarette usage anlong young adults as mncb as half a cent try ago. t'llw data utilized are frmn the third largest prospective survey ) 't reporled anti are similar to the find- ings in lhe larger two surveys; the corresponding results from all ~hree of these surveys are presented in the II.S. Si.gcon.(;rnr~'al's 1982 trl~tl (.I).

r'dchord Pefo and Rlchord Doll Therefore, current trends, current urban/rural differences, and current inter- national differences in lung cancer reflect, among other things, past Ircnds, past urban/rural differences, and past international differences in cigarette usage by young adults. Consider, for example, the extent to whicb current trends in United States lung cancer mortality rates among men now aged 70 may be affected by the large trends in cigarette consumption 50 years ago among people then aged 20 (5). (For details, see Appendix E of Doll and Peto, (7). In 1930, United States cigarette consumption was increasing rapidly among young men, and national sales rose from 1 cigarette per adult a day in 1915 to about 10 per adult a day in 1945. The effects of those increases are only now becoming fidly apparent, and largely as a very long delayed result of them, United States male lung cancer rates in late mid- dle and old age are still rising steeply, despite the fact that cigarette sales per adult have remained at approximately 10-1'2 a day ever since 194-5, and tbat tar levels per cigarette have fallen substantially (Figure 3). Contrary to various suggestions, the "discrepancy" that has been seen for the past 25 years in the United States be- tween rising lung cancer rates (see Figure 4) and falling tar levels does not imply, or even suggest, that Americans are exposed to increasing levels of carcinogenic pollutants other than tobacco, or, as the recent (10) National Academy of Sciences-National Research Council (NAS-NRC) report suggested, that tar level reductions in cigarettes have been ineffective. Indeed, but for tar-level reductions, the current increases in United States lung cancer mortality rates probalfly wouhl be appreciably more rapid. Likcwi~, in many countries the smoking of manufactured cigarettes by young adults was a habit that tended to become established first in the towns before it spread to the surrounding countryside, rather than the converse. Consequently, 15 Actt~al cigarette consumption per adult 0 ._~ O /- cigarette equivalents) 19~ 1920 1940 1960 1980 Fi~re 3. Trend~ in United ~latc~ consumption: mean daily ~ale~ of mamlrarl,red ciga- rel~c~ l~r Untied Slale~ athlll aged over lfl year~, It~elher with a £rtltle e~timal~ of lar yield l~r atlull. Fr.m Ref. (5). "ltm estimate of tar yield allow~ approximately fi~r tlecl'ea~es ~hlce the 1950s i, lar yield ~r cigare,e sm.ked in a standard maturer, hut not fi)r any syslemalic changes in lhe manner in which cigarettc~ are smoked. The Control of Lung Can :er 7 half a century ago cigarette smoking was probably more prevalent among young men in towns than among young men in the corn try. Disparities in recent years between urban and rural lung cancer rates amon! today's old smokers therefore may rcsuh chicfly not from air pollution but rom a delayed effect of past urban-rural differences in cigarette usage among t ve people who were then young but who are now old. Finally, it is wholly wrong to suggest that the p~ or international correlation be- tween current smoking habits and current lung cmcer rates indicates tbat smoking is not the chief determinant of worldwide hmg ca~,cer mortality. For, such a cor- relation effectively relates lhe lung cancer rates of t ~e grandparents to the smoking habits of their grandchildren. If instead the nati ~nal lung cancer rates for one generation arc related to national cigarette consum! ~don rates when that generation were young adults, a moderately close relationsbi], does emerge (Figure 5). 25,000 20,000 15,000 10,000 6000 2000 0 "Estimated rates among nonsmokers ~ Mouth, esophagi l:lh~ rynx or laryn ~ Figure 4. Recent trends in United States cancer mow Mity rates: age-standardized death ccrlificadow ral~ (Iwr I(~ million i~.~q~le a~cd .rider Ti~ ) in flw 19~Os (lop bar), flw I~ (middle I,ar) aml dw 197(E (t~)ttom bin') G)r vmi,m~ ~ .'~ .f ennecr in lhc Untied From Rcf (5). For cancers of the long avul .])l~'r rcspir~ h,ry and digeslivv tracls, estimatt'd rates for lifelo.g IlOnsl.nkers are alst~ given (asterisked ;m', nlxwc the rates for the 8~O~g9890g !~

r'~hord Peto and I~herd Belglu.m Canada ~ France New Zealan Germajy ~ / b~~Greece " ~ 60 ~pain / ~ ~ ~ . • Rates based on o~r tOO deaths m ~ ~ ~~ Den~ ~a/e~sed on 25-~00 deaths ~ ~o n D Noway 500 1000 1500 2000 25~ 3000 Manufactured clgaretles ~r adult In tg50 F~gure 5. I,un~ crower m~d mno~h)~ hi Ihe ~ame ~enera(ion: rclmi.aship hclween hmg m;malhrl..'d (exrl.dc~ tmndmlled rig~rettes in I~l~ium cigarette rousumpli~m when flint gcncralioa of ~ople were in early adult life: (lala various cou.tries, and for US non-smokers estimated hy fittin~ Che p~s[~ectk'e s.rvey d.¢~ re.reed by dm American Cancer ~'icCy (25). Other Features Dose-Response Relationships In Table I, it may be seen that doubling the dose may approximately double the excess risk at each age. Partly because ofdifficuhies of dosimetry~, it is not rcaUy known whether, as Doll and Pcto have tentatively suggested (8), a doubling of the true dose rate produces an approximately fourfold increase in the age-specific effect, or whether, as is suggested by much other data, it merely prodnces a twofold increase. Whatever the exact truth, however, it is clear that two packs a day for 20 years is far less hazardous than one pack a day for 40 years, so any reports based on inappropriate concepts such as "pack-years" should be treated warily. Effects of Stopping Smoking When smoking ceases, the annual excess risk remains roughly (perhaps to within a factor of two) constant thereafter. Referring to Table 1, it may be seen that the annual excess risk after 30 years of s~noking is about 0.1%, ~ if a smoker stops el'he cffedivc (k~se may .at ~x: simply proportional to Ihe nnmhcr ofcigarc.es sm.ked per (lay, for the CO uptake fx'r cigarette apl~ars to be less fi~r heavy than for mc;~lerate srm~ker.~. AI~, t~.¢ause the chief target area i.~ the main airway~, r~pid inhalatiem may ¢Jcp,'~sit les~ rnl thegn/Iron ~h~w inhalali~m does. 'lhi~ s.ggestl.n j~ rr'~'~tfi;rr'erl by relents (12) that i. ~mw, altlto.gh .at all, ~t.rlies he;try .'o.oker,.~ who describe themselve.~ as "n,r~t inhaling" get m~z~ hmg cancer than do co.q~arably heavy smnke~s wh¢, "do inhale"!. ~o~ggggog The Ca431rol of Lung Can :er 9 after 30 years, then approximately this annual exce :s risk may persist indefinitely. Thus, for example, 15 years later the annual excess ~ isk may still be about 0.1% in- stead of the 0.5% that it would have been had smokh~g continned, so about 80% of the excess risk is being avoided. It is not true, bower :r, that the annual ab~lutc ex- cess risk decreases substantially, and still Icsa is it tn c that it decreases to zero after 10 years; only one prospective study has suggest~ I that, and the others clearly refute it. But, the large increases in risk that would otherwise happen are avoided by stopping stnoking. Tf'~9 Importance of C!gareffes as Opposed fc Pipes In Britain and tile United States, cigarettes app~ ar to have a far greater effect than pipe or cigar tobacco did, and so the switch ea "lier this century from pipes to cigarettes has produced vast increases in lung cat car. The reasons for this dif- ference are not adequately known, especially as the ~moke from pipes and cigars is about as carcinogenic as that from cigarettes for lal ,oratory animals. One sugges- tion is that the difference depends chiefly on the ,t rearer alkalinity of the smoke from pipes and cigars, which may both make inhalalion less pleasant and facilitate the transport of nicotine across the oral mucosa, tht reby obviating the need to in- hale (13, 14). This suggestion may not be difficult o test and, if confirmed, may point to an important way of diminishing the hazat tls of cigarettes, bat at present this remains speculative~. A related suggestion is tit ~t the "air-cured" tobacco of, ft." cxamph', certain French cigarettes ~nncwhat ~cscmblcs pipe tobacco and is therefore substm~lially less carcinogenic than the "quc-curcd" tobacco typical of British aud American cigarettes, but the intcrnation d differences in lung cancer on which this suggestion rests owe so much to differ 'nces in duration of cigarette smoking that it is still unclear whether there are also any material differences in the hazards of the various cigarettes. (During the 1.¢ 30s and 1940s, for example, British cigarette cousmnption was four times that i t France.) Interaction with Other Causallve Factors A variety of other causative factors for lung cancc • are known, of which the best studied are asbestos, ionizing radiations, and urban air pollution. All thcse have a far greater absolute extra effect on smokers than ¢ n nonsmokers (illustraled for asbestos in Table 2), as may various other causative actors. ,Some of the benefits of control of certain other causes o flung cancer thereto c may be attainahle indirectly by reducing tobacco exposure. However, because e ffectlve tobacco exposures are currently increasing in many countries (and even v.here they are decreasing, the immediate decreases are unlikely to be enormous), the theoretical possibility of avoiding tobacco exposure clearly does not justify in action where other substantial causes of lung cancer can be reduced materiallyL qn the stn(ly of Cc¢led~3f el al (15) in Sweden, p~pe ~mokcrs had the same tcnfokt cxces~ of lung cancer that cigarette smokers had, which rather sugges1~ that the sm ,llness of the effects in Britain or the United State.¢ may res.h more from traditions almut h~w pil~ ~ are smoked than frmn the pharma- colr~p/of dw smoke--and it i.~ unlikely that such traditinn~ wil dwmsclves Ix. wholly dcternfincd pharmacoh~gic faclnrs. °Apart fi~...~rm,kh~g, a~l.'stos, kmizlng tmlintlons, aml comb. li.n pr~*dn('t~ .f fi,~il Ihcl~, the reliably cstabli.~hcd can~c.~ nfhmg cancer are. bi~ehloromelhyl)ed'cr (BflME), mu.~tard ga~, and certain comlxr.nds .r oxidation ~tat~ of As, Gr, and Ni (5).

I(9 Richard Peto and Fdchard Doll Table 2. Multiplicative effects of heavy asbestos exposure and of smoking on lung" cancer risks" Relat|,ce risk of I_u.0ag c~a ricer for: Nonsmokers Smokers No known asbestos Heavy asbestos exposure (prolonged employment as a lagger before 1968 United States dust controls were introduced) ! (reference category) 5 aDma rmm Selikoff (16). Note that although such heavy asbestos exposure is no longer permitted in many countries, places where heavy occupational exposures do still occur may offer excellent opportunities for limited disease prevention, because even if the workers do not smoke (=, the excess risk of bronchial carcinoma is low), the risk of me*othelioma, which does not depend on synergy with tobacco, will mill be high. Mlscerllflcatlon of Lung Cancer Deoths People, and especially old people, dying of lung cancer may never have their disease recognized and may be miscertified as dying of ~me other condition. Pro- gressive rectification of such errors produces large, purely artifactual, increases in hntg cancer death certification rates. In middle age such effects were substantial during the first half of the century, even in developed countries--for example, when diagnostic radiology was introduced dtnring the 1920s, it prodnced alxmt a threefold incream in British lung cancer death ce~ification rates--but in midclle age st,oh effects are now limited chiefly to underdeveloped countries. In old age, however, large (eg, twofold) artifactual increases have continued to occur since 1950, even in various devclotx'd countries, whereas among old people in many underdeveloped countries lung cancer death certification rates are still grossly unreliable [as are "age-standardized" lung cancer death certification rates, unless standm~dization is to the truncated age range 35-64 recommended by the Inter- national Agency for Research on Cancer (IARG), (17) to circumvent such difficulties.] Tar Deliveries The effects of changes in tar deliveries need to be properly understood by anyone concerned with the avoidance of cancer, for at least in developed countries they may offer one of the more important cancer control strategies. Between the 1930s and the 1970s there have been reductions of more than 50% in the mean tie, sales- weighted) tar delivery per cigarette in the United States, Britain, Scandinavia, and a few other places. These changes were small until the late 1950s and then they sud- (h'nly l)vvame rapid, with dcrrcases fi'om 30-odd mg per clgarc/te in ei~e mid-1950s down m alqm~ximau'ly 15 mg IWU clgmelle hydtc 1970s. The chau~l{es me n~t ex- pensive, and involve d~e use of fihcrtips, porous paper (or even, as an extreme measuH', "ventilated" fillers 0rot allow air to tiller hllo the side nf the filler t(i The Control of Ltn~ Cancer 11 dilute the smoke) aod modified types of tobacco (which may in some instances ac- tually h,e less expensive than unmodified tobacco). There is, of course, a reduction not truly in the unwanted componenls of the smoke but also in those substances (eg nic'odne?) to which some smokers are adclicted, atttl when snch reductions occur many smokc~s a~ likely to com~nsate, either by smoking more cigarettes~ or, ~rhaps more commonly, by taking in more smoke per cigaretteu. It appears, however, that the latter form of compensation is not Mways su~cient to outweigh the reduction in tar (19), in which c~e the net ~sult will ~ inhMation of less tar into the lung. This conclusion is suggested ~th by common scn~ and by obse~a- tion, but cvcn if it is accepted it does not prove that the h~ards will ~ correspon- dingly reduced, for despite some 30 years of la~ratory resea~h the im~rtantly carcinogenic factors in cigarette smoke have not yet been identified reliably. Moreover, it is di~cult to predict how changed patterns of inhMation will change what is de.sited on the main target areas--which, for lung cancer, are not the peripheral tissues, but in the large airways~ the smoke streams past them. Gonscqucntly, it is necessary to discover by direct epidemiologic observation whether the risks of lung cancer are materially reduced by the widespread switch to lower tar cigarettes. Unfortunately this is not easy to do, for not only are smokers of low-tar brands self-selected but al~, just as it is only a~er some decades of smoking that the full risks matcri~izc, ~ ~rhaps it is only after some decades of using low- tar cigarettes that the fitll benefits will materialize, Therefore, even if the effects in late middle age will one clay be substantial, they may not yet I~. Any substantial effects that arc going to materialize in ca@ middle age should ~ beginning to be evident by now io Brilaln, however, for although the tar rcdnction~ of the 1950s were nnly moderate, lhosc of the 1960s were substantial in Brilain, North America and Scnndinavia. Thus, a 40 ycnr aid in 1980 will have been smoking from about 1960 to 1980, Ihroughoul mosl of which time tar levels were substantially lower than in previous decades. Two main pieces of epidemiologic evidence are currently available, the first being the rcsuhs from classical case-control or pros~ctive surveys. Unfortunately~ such data as are currently available are limited by the fact that they relate chiefly to late middle or old age, when most of the lung cancers occur, and even recent ]sin principle, tar reductkms could either increase or decrease the number of people who smoke (by making it less of an ordeal for nonsmokers to acquire the habit or by making the habit leg'~ addictive) and coukl either increa~ or decease the number of ciga~ttes one iodlvklual smnkrr consumes (by in- creasing the rmmber needed to achieve a given do~ or by decreasing the satisfaction ~r cigareHr}. In practire, lmwcver, the patterns of cigarette consumption in different conntries do not ap~ar to l~ in- flnenced consislenlly in either direction by chants in cigarette consumption. ttSurptisin.gly, there appears to b~ little reliable information on which of the many characteristics of the cigarette (eg, nicotine, draw resistance," taste) im~antly affect "com~nsad~." If these cot]hi be i<h,nlifivd and lntalified {<.g, b~ invreasing the niroline deliver y. draw resislan<'e, or whatever of law-tar l~lii In'e~uoIMIl7 iln~'lliel Ihe reel'nl di~alll~finlloI ~ndloI h7 K~il[rtia~ ~1 al (111) Ihat ill* ri~ of lii7~w~fdilil hifiirrlhlli all" irli lnalerllill7 di~erenl miumt lillt~ker~ of differeni 171~ ill £~'08E9,£908 .., • Shiffman Medical Library, 4325 Brusl~ St,'

12 f'dchord Pefo ond r'dchord Doll studies relate chiefly to people who have smoked low-tar cigarettes for ouly a frac- tion of their smoking lives. This dittqcuhy is exaccrhated in studies perfi~rmcd dur- ing the 1960s (or early 1970s) by the fact that the tar reductions then availablc study wcrc not only more rcccnt, but also less extreme, than those curttally available. A related source of di~culty is that as overall tar levels dccrcasc, the highcr tar levels simply cease to exist, so direct concurrcnt compari~n of people now on low-tar cigarettes can ~ only with ~oplc on re@crate-tar cigarettes, anti not with the old very high tar brands. ~spite the~ di~cnltles, when I~e and Gar- finkcl (20) reviewed MI the case-control and pros~ctivc studies then available they concluded flint: a reasonably clear picture has emerged. 33fis is that smokers of fihcr (or low tar/nicotine) clga~ttes have a lower mortality than smokers of plain (or high tar/nicotlne) cigarettes for tho~ dise~s ~st ~trongly a~iatcd with smok- ing .... 33~cs~ reductions in mortality have l~en ~n in those who have smoked the more modem ty~s of cigarette for only a small part of their smok- ing livc,. "~e fact that those who have smoked them fi~r longer show even grcawr ~cluct~ns in mortMity ~uggests that the ovrrall pictu~ will impure even more in ~ea~ to Com~. Becausc of di~cuIfieB oE sclf-sclccdon, of comparln~ the ncw with thc old c(m- currcndy, and of characlcrizing individuals' recent hmg cancer ralcs in early middle age tic, the rates among people who have smoked low-tar cigarcttcs for much of d,~ir aduh lives), d~c case-control and prospective survey data cau I~ sup- plcmcntcd Uschdly by a second type of cpidcmiologic data, ic, thc stody of nadnual trends in early mkkllc age. However, fi~r reasons that already have ~cn discusscd, it is not advisable to use for this pur~se data (such as those from the Unitcd States) in which any downward trends caused by tar reductions arc likely to ~. diluted or even rcvcr~d by upward trends resulting from the delayed effects of past incrca~s in tobacco consumption. Instead, it is ~tter to use the British data. For by the 1950s (when thc rapid tar decreases began) British mMe lung cancer rates in early middle age had Mrcady approximately stabilized (Table 3). Table 3 also descries their subsequent evolution, and the reductions are extremely impressive. They are most unlikely to result f~m changes in air ~llution, for not only are any effects of air pollution likely to ~ far smdlcr than this (11), but ~so similar hMvings in early middle age have ~en seen over the last 20 years in un~lluted Finland. Moreover, ~th in Finland and in Britain the changes appear, if anything, to ~ accelerating downward, so if this pattern carries on into late middle age during the next decade or two, thcn at least in these two countries (where the male death rates are at pres- ent uniquely high) lung cancer may some day decrease for a few years~: as fast as it once incrcascd. A finM piece of human evidence that tends indirectly to confirm the reality of these changes is provided by a comparison of histologic sections from American "I1 will not de('rea~ to anywhere near non smoker rates, however, unle.~ there is widespread abandon- meat of ciga~tte smoking. ~milarly, in tho~ other ~pulations where lung caner rate~ have n(H y~t completed/heir ri~, even a tar-level ~luction dmt halve~ the c~rcitmgeniclty nf ciga~ttes may merely ~low, rather than reverse, the progressive increaw of the di~ase ~er ~l~e next few decades. 9~OB39B903 The Control of Lung Can, er Table 3. Recta! trrnds in England an,d Wales male h, tg cancer death ccrlificafion rates in early nti(hlh, age"'t' Ikath certification ratea per million ~ ,en from cancer~ of the re~plratory Age 1951-55 1956-t 0 1980 Ratio (r") (") (b) (0 (~b) 30-# 3B~ 37~ 13 0.3 35-9 lOP 95~ 45 0.5 40-~ 253~ 256~ 134 0.5 45-9 58~ 59~ 37~ 0.6 annie bo¢~ tlw approximate constancy heft)re tar deliver|e~ I'x'gan In Iw greally ~t~d and the la~e drr~axe thereafter. bsol~: (I)'[lw~'lrrnd~nlr.~tmalrtiallyaffi'rlrdhy~hml~r~in~l~rali~, I,ralmrnl~,ft[wdi~a~' (2}Sah,~-~.rlghtrd Mran ~ igarrlte t ~mxU~nlaitm Iwr Brilixh male aged ~-50 did n,,I hangr greatly until the pa,I few yrarx and in 1955. H~5 anti 1975 wn~ rr~pr(liw'ly, 10.5.9.9. and 10.2 121 ) Ihe I0*~1% d~'rea~e in (tm~umpti,m thai have likrwi~" }~'t'll appr.M~natrly Iml~ m'er the im~t 20 years. 1 '~d in l~lb n~mlr~ the de~ rra~x apl~ar, ir dlligh inlnke t~llly in fir~ t~ or m~ tff ~noking hi~tory. aulopsics in the 1950s aud in Ihc 1970s (22). In lhc IqS0s smokers I~atl a high, dose- related prevalence of what were thought to be prene q~lastic lesions, whereas by the 1970s such lcsious were an order of magnitude less ~onunon among smokers. The exact biologic significance of lhese lesions, howcw r, remains obscure, especially because their prevalence decreased so sharp/y durin ; a period in which lung cancer rates were rising. (They may be indicators not so n,uch of the extent to which the main neoplastic endpolnts are occuring, but of tl e extent to which one of the "stages" of carcinogenesis is occurring.) Practical Actlo~: Discouraging Sales and Decre~slng Tar Levels Sales A varicly of WltO and UICC expert reports hart. been prepared on how volun- tary organizations and governments qan decrease ci! arette consumption, and these deserve carefid scrutiny for they contain much ~ ::ll-judged advice. There are, however, two important respects in which they m;~v be somewhat deficient. The first is that lax increases may .nnt be suffieientl) emphasized. Because many gtlvt'rnltltqllS drrive large tax yichls fi'om tobacct, sales, all hut this one of the strategies dmt may be considered for reducing cigar,'ttc sales, will also, if effective, r(.ducc Ia× rcv,.nucs. Ahhough in principle governn ents may t~/ieve they act only for Ihe good uf their chizens, in practice they may tend to dccide that what is

F~hord Peto ar~:l FOchord ~ Table 4. Elasth:ity: Predicted change in cigarette sales per 10% increase in real price Country' studied Estimates in 9~ different papers ofthe change in annual cigarette sales associated with a 10% price increase (%) United States - 5, - 8, - 4, - 4 Canada - 7 United Kingdom -6, -5 Switzerland - 8 Finland - 3h aBelween 1950 aml 1968 no studies of the rla,~ti,'ity of cigarelte demand were published, but since 1968 at least ten have Ix.ca. All are cited except G~r thai of Atkinson anti Skcgg (1974), which is s.pcrseded by the reanalysN by rl Peru (t974) of the identical data. For re.re.fences, see Ontario C, ouncil of lleahh (7). This estlmalr of only 3% was published in 1974 and would have b~en mm'e extreme if 11 had I~n ~sslble ~o in- dude the ~ul~eq.cm large incrca~ in prke and decrease in constanp~br~ tha~ ~k place in Finland i~ mid- 197~. cconomically easiest fclr thc government is bcst for thc citizcns. Consequently, the otto strategy--increaslng the tax on tobacco--that increa~s cadger ~han decreases ~ax revc..es l)crhaps deserves more emphasis than it usually ,gels, esl~:ially because it is one of ¢l~c few straWgies for which thcrc is clcar, direct evidence nf ef¢~:ct, In ]gBI, for example, increases by n total nf at~m~ 2(~% in British cignrcltc prices pr¢~duccd decreases that, althoogh substanti~, were Icss than %)% in cigarette sales, so the tobacco manuhcturcrs complained of unemployment in thc industry while tim government collected marc t~, The same thing hap~ncd in Finland in the mid-1970s. Several reports during the past 15 years have examined marc formally the quantitatlvc relationship ~twcen price and consumption in these and various other countries with remarkably consistent findings ('Fable 4). At Ic~t for the subsequent year or two, a 10% increase in price appears to produce a~ut a 5% decrease in consumption. If such a decrease were hrgcly permanent, dmn it would in the king term prevent abont 10,~0 lobacco-induccd deaths per miffion cigarette smokers, I~ is more di~cuh to prmlucc a reliable estimate of the extent to which these year-to-year changes in consumption, produced by price changes, persist over longer ~riods, for so many other hctors also may bc in- valved. Despite this, however, some dircc~ cvidcncc for the common~nsc notion that price does affect long-term, as well as short-term, consumption is afforded by the general tendency for cigarcttc consumption to ~ high in many countries where the price is low (7). [n view of such data, the promlsc ofdclibcratc shifts finm other g~x)ds to tobacco may deserve greater emphasis than it oficn rccclvcs. A second dcfi¢'icncy of emphasis is that lhcre may have bccn insu~cicnt stress on ~hc longJcrm advantages of getting quantitatively infi~rmativc material across abrupt ft.: l~lal risks from tobacco, and tim cxtc.t l~ which, at least in countries, thc~ exceed all other reliably known causes of death. "l~c reasons any ~rious i.~gram of canccr prevention must strcss thc hcalfl~ clTccts of tobacco arc illustrated by Table 5, which has t~cn abstracted from thc chapter nn Cancer Epidcmiology in the Oxford 7~xt~k ~M~icine (~3), This ~rs~ctivc, howcvcr, is Li~O~ggggog The Contro~ of Lung Can :er 15 Table 5. Reliably established, practicable" ways of a, aiding the onset of life-threatening cancer in the United States or United Kingdomb. Percentage of all US/UK cancer "teaths known to be thus avoidable Avoidance of tobacco smoke 30 Avoidance of alcoholic drinks or mouthwashes 3 Avoidance of obesity 2 Regular cervical .'z'reening and genital hygiene 1 Avoidance of incs~ntlat medical use of hormones or < l radlnlogy Avoidance of unnsnal exposure to sunligh~ < I Avoklance of known eff~ts on ~ple of current levels of ex~,snre m carcinngens Occupational context < V F~, waler or urban air < 1 aExcl.dlng ways sm'h as pmphylaclk p~statcclomy, mas~ccto, ~y, hys~ereclomy, ~pho~mmy, arlificial hFmm [~[ a~d P~'¢¢' (23). rThc p~orlion of current United Sla~es cnnrer deaths Ihat are I kc[~ to rcsuh from ~cupal~nal factors was or pas~ exi~,~u~ ~o aslwsl~ may ~c~ ~mnt fiw I-~% ~ all curre, ~ United ~ate~ caner deaths, slill rising and lhnl eve.l.ally may well Iw 2-3%. However. l~ca ,¢ Ihe nl,pmxhna~c magnit.de ¢ff dw health rffi,t Is .f rx[~.~ to ndwslt~s I~amr wJdr}y nvrrplrd, ex~,sure I," cls have Iwen grcally reduced and are now in ca~in,~ns (~, [~n~kfine), the ~la~ cff~'ts M ~dch a~ ~ill apf ~.afing. l~mg after ~tand~ ~t~ in ex- ~ ha~ ~aken pfa~. unfamiliar even {o mos¢ cancer research workers. Jet ~]one to most nonmedic~ ~ople. Indeed, in r~ent surveys in Britain most ~ ~ple mistakenly imagined that {ra~c caused marc deaths than tobacco; in fact [ ~bacco causes over 20 times as many UK dca{hs as trn~c. Likewise, in the Unit~ ~I States recent surveys have in- dicated flint runny people believe that backgroun~[ radiation from nuclear power plants is n greater hcnhh risk than tobacco (24), ~ hereas in fact tobacco is several thousand times more iml~rmnt. Such gross misp,.rccp{ions, of cou~e, may have substnnfial effects on behavior. Indeed, the ch~irm an o~ R.J. Reynolds, America's largest cigarette manufacturer, rc~r¢cdly (New Ye,k Times, April 12, 1981) said to his sh~rcboJders [hat the reason ~he cancer.scare was no longer hitting cigarette sales ~ hard was that so many things have ~en linked to cancer that ~ople were "beginning [o take a more objcctlve [sic] view of the heath evidence"~ He may well I~ right about [he effect of the string of ~"~rts a~ut new carclnogcns, l~causc truly a remarkably l~rciplcn[ newspa~ reader or televiewer wash{ able [o gncss, after rcadlng abou{ ~nc new cancer scare n~ter another, {hat old sou-newsworthy [nbacco was still causing a}~u{ one-tiffed of dcnlhs~nn effect {ca times as large as the next mos¢ im~rmn[ reliably known cffcel. It is admittedly difficult (o comm.nirat~" risks in a way Ihat will be tmdcrstood and rcmcml~rcd approximately cor~ ,'ctly, csl~ciMly by pcoplc who have no framework oft~flmr risks with which to co,aparc them, tlowever, it should ~ ~ssiblc as }ong as the main message is set clcm~ y apart from the lcsscr messages

16 i"dch~rd Peto and l~chard Doll that qualify it and Ihat may help prevent people from rationalizing it awayL~. Aflcr all, the chief message is merely that "ABOUT A QUARTER OF AI,I. REGULAR CIGARETTE SMOKERS WII.,L BE KILLED BEFORE THEIR TIME BY "FILE IIABIT," which is consldcral)ly less complicated than tim mass of quantitative infornmtion about house prices, groceries, car prices, clc, that already has become part of the folklore of consumer societies. How exactly this main men, age should be put over is a matter for experiment; comparisems with other condidtms may (especi,',Jty in Britain) be helpful, eg, SM()KING IS BRITAIN'S BIGGF~'q'I' KII.I,ER Am(rag 1000 young adults who smoke cigarettes rc~darly, - at.xml 1 will be murdered - alxmt 6 will be killed on the roads - al~mt 250 will be killed by tobacco. "For example, one aright follow the main me,age with a few explanatory notes, such a~: - Some of th,-L,w kill~ hy I~bacco wmdd have di~l ~n anyway, hut other~ ndghl have lived m~ 5, 10, 20, 30, dr more extra yenr~; the average amount of life Io~t by them l~ing 10-15 year~. - If yo~ glvc np I~-fi~re yo~ have ~a ~i~nt~ hrml dilate, hmnchili~, or rnncer, Ihen yogi nlrr~l ,ff lhe ri~k of~'n~h from sm~king. - llzmnge Io I~ }~ly from smoking accurnnlale~, m linage w~ ~1~ in their l~u~ will I~ nt greale~l risk in mk~le a~. Even in the United Slates, where road accident death rates are more than double those in Britain mad murder rates are about ten times those in Britain, some such comparlso.ns may be helpful (ahhough it may then be advisable to start with only 100 young United Stales adults, and to threaten about I, 2, and 25 of them with death). Whatever format is preferred, however, the central point remains: The reason one wants to prevent smoking is not just because it is dangerous--dozens of things are dangerous--but because it is so dangerous. This indicates getting some sort of quantitative information over, both about the effects of smoking itself on mortality and, perhaps at least as importantly, about how much srnallcr all reliably known other carcinogenic effects are. Such information may in the short term make only a few people give up, but over a few years wide acceptance of such a pcrspcctlve may have substantial effects, either on individual behavior or on making other actions politically acceptable. Tar Levels The foregoing epidemiologic evidence (especially that on trends in lung cancer morlality among English men in early middle age) strongly suggests Ihat, even without any substantial changes in cigarette sales, practicable reductions in sales- weighted tar deliveries may well reduce the lung cancer mortality from smoking ~08~9~90~ The Conlrol of Lung Con~ er 17 sul~stanlially. There was, moreover, in Lee ant Garfinkel's review (20), no evidence that any other disease was aggravated b) such changes mnong smokers of similar numbers of cigarcttestL Also, although tar and nicotine decreases do prcxluce some compensation in tim manner in whi h cigarettes m:e smoked, they do not appear to be important determinants of wl -ther or not people smoke, or of the number of cigarettes that they smoke. Tar reductions can IJ¢ implemented with no ;ttbstant~al political problems (especially if they are done centrally, or at least ~ ithout advertising campaigns that may suggest misleadingly to non- or ex-smol ers that low-tar cigarettes are safe), for they do not adversely affect the groa, ers, manufacturers, taxers, distributors or advertisers, and the smokers apl car hardly to notice gradual changes in tar deliveries. It is, therefore, unfortunate that while the WttO and UIGC have organized several meetings on stunk ng avoidance, some of which have produced reports listing a variety of practice,/suggestions for govermnents or fur vohmtary groups In consider, no similar reports arc available to help governments accelerate tar reductions. "/'he prob ~'ms are, nf course, quite dif- ferent from one conntry to another, depending o t whether the country is a to- bacco grower, a cigarette manufacturer, an export i~r, or an importer, on whether cigarette mannfactt,rc or distribution is virtually ia govermncnt monopoly, on whclhcr advertising is altowcd, on the era'rent tar I ,vels, and so on. For countries such as Brilain, dilli'renlial laxaliot| (which has ~]rcatly been used once sue- ccssfully to cut off the highest tar levels) could be 'l~scd again to cut off the next highest levels, and restrictions could be imposed im the advertising of, for ex- ample, all brands delivering more than 10 mg of t~, r. For countries such am China and Russia, where cigarettes are manufactured at d distributed by the slate with little or no advertising, and where typical tar levels exceed the upper limit of what is currendy sold in Britain, large changes could be ~roduced at little cost and with great benefit to their people early in the next con ury~s. However, the practical problems of how to help governments decrease ta " levels (without inadvertently encouraging a belief that low-tar cigarettes are sah ) is a large question, almost as deserving of carefully thought out, practical repot ts as the problems of smoking avoidance are. Ultimately, of course, the aim is to produce cit'~umstances in which very few people choose to smoke, but in a world where cig;wctte sales are still increasing, rather than decreasing it is not wise to let the perfc~ I be the enemy of the imssiblc. "In view of the extent to which smokers of low-tar cigarettes r o "compensate," however, and of the uncertainly as to which smoke components chiefly affect heal ! disease and chronic ol~tructive h,ng di.'~ease, there shmdd be no implicit expectation that these dis 'ases will also he avoided, and indeed the large case-cnntrot study of Kaufman et al (18) soggests ap woxlmate equivalence of the effects on myocardial infarction of the different types of cigarette thai ~ re currently available. ~During the 1950s, men in Finland still smoked "Russlan-st le" cigarettes, and in 1960 male hmg cancer iucideurc rates in early middle age were ~imilar in Bus :ia and in Finland. By the early 1980~. Imwcver, lyph'al Finnish rignrclle tar dcliverie~ had dr, pped m only 10-15 m~., whih" typical Rus- sian cigarette ~ar dcllvcric~ were slill about 20-30 my.. Recc ,t Finnish male hmg cancer ~midcncc rate~ in early middle age have decreased by nearly half, whil' those in Russia have hardly ahcrvd.

18 f'dchard Polo and I'dchatd Doll Table 6. lnfi+rmallon for governments on simple measures fi~r the control ofhmg cancera Price increases will pnxltwe fewer deaths anti more revenue (as long as they do nt~t feed Imrk into wage demands). Tar reduction~ ~uld ~ enconra# (e~p~iMly in countrie~ ~och as Russia and China whe~ typical tar levels are still of the or~r of ~30 rag, which i~ ext~mely high). Advertifing could ~ t~ed, ~t~ct~, prohibited, or limit~ to ciga~ttes delivering un~r I0 mg tar. Simple, clear, quantitative information could t~ communicated effectively to the general ~pulatkm: ABOUT A QUARTER OF ALL REGULAR SMOKERS ARE KILLE1) BEFORE THEIR TIME BY TOBACCO. GeneraJ no e: rvt on menda mn of hose few stmple measures (whwh might have a mtb~ an ial cffec m just a few off ex~) thm~ nt~, of court, detract from the need for a wide range of other meamtre~, including many ofth~" integrated by W}IO (2). UICC (6), and Tar-level reductions are not the only simple possibility for governments (Table 6), and they may do little for vascular or respiratory disease. But they tnay well offer one of the more immediately practicable means of avoiding an appreciable proportion of the mass of lung cancer deaths that can otherwise be cxpcclcd In oc- cur dttring Ihe first few dccatles of the next century. References 1. Parkin DM, Stjernsward J, Muir CS. Estimates of cancer occurrence througbout the world. WHO Bulletin (in press). 2. World Health Organization. Gontrolling the smoking epidemic: report of the WHO.expert com- mittee on smoking control, tFechnical Report Series 636. Geneva: WHO, 1979. 3. U.S. Surgeon-General. Smoking and health. U.S. Department of Health, Education and Welfare Publ No (PHS) 79-50066. Washington, DC: U.S. Government Printing Office, 1979. 4. U.S. Surgeon-General. The Health Gonsequences of Smoking--Cancer. U.S. Department of Health Education and Welfare Publ No (PHS) 8'7-50179. Washington, DC: U.S. Government Printing Office, 1982. 5. Doll R, Prig R. Quantitative estimates of avoldable risks of cancer in the United States today. JNCI 1981.66:1191-1308. 6. Gray N, ed. Lung Cancer Prevention: Guidelines for smoking control. Geneva: Union lnterna- tionale Contre le Cancer, 1977. 7. Ontario Council of Health. Smoking and health in Ontario: a need for balance. Report of the Task Force on Smoking of the Ontario Council of Health. Toronto: Ontario Government Bookstore, 1982. B. Doll R, Peto R. Cigarette smoking and bronchial carcinoma: dose and time relationships among regular smokers and lifelong non-smokers. J Epidemiol Community Heahh 1978; 32:303-13. 9. Doll R. The age distribution of cancer: implications for models of carcinogenesis (with discus- sion).J R Statist Soc A 1971; 134:133-6. 10. National Academy of Sciences-National Research Council. Reduced tar and nlcotine clgarcttes: smoking behavior and health. Washington, El'G: National Academy Press, 1982. I 1. CederI,Bf R, D,.~II R, Fowler B, Frieberg L, Nelson N, Vouk V. Air pollution and cancer: risk assessment methodology and epidemiological evidence. Environ Health Perspect, 1978; 22:I-12. 12. l)nll R, I'elo R. Mortality in relation |n snmklng: 20 years' ol)servallm;s nn male British din:lots. Br MedJ, 1976; 2:1525-36. 13. Wald N, I~ll R, Gopeland G. Trends in tar, nicotine, and carbon monoxide ylehls of UK cigarettes manufactured since 1934. Br Med J 1981; 1:763-5. 6~0~9~90~ The Cooled of Lung Cancer 19 14. Wald N, hlle M, !~reham J, Bailey V, Van Vunakls H. Serum codnine levels in pipe smokers; evidence against nicotlne as a cause nf ct~mnary heart disea~, l.ancet 1981 ; ii:775-7. 15. (:ederlfif R, Friberg I,, ltrub<'e Z, I,orieh U. The relatlon~hlp of smoking anti t~me rovariahles to mmtality and cancer mnrlfidity. Sto, rkholm: Karollnska Institute, 1975. 16. Selikoff 1.1. (:onstraiuls in estimating occopational cancer mortality. In Polo R, St hneiderman MA, ed~, Q.uanlilication of occupational cancer. New York: Cold Spting Itarbor Publications, 1981. 17. International Agency for Research on Cancer. Cancer incidence in llve continents, Vol III. Geneva: World tlealth Organization, 1976. 18. Kaufman DW, Helmrich SP, Rosenborg L, Miettlnen OS, Shapiro S. Nicotine and carbon monoxide content of cigarette smoke and the risk ofmyncardlal infarction in young men. N Engl J Med 1983; 308:409-413. 19. Wald N, Idle M, Boreham J, Bailey A. Inhaling habits among smokers of different types of cigarette. Thorax 1980; 35:925-8. 20. Lee PN, Garfinkel L. Mortality and type of cigarette smoked. J Epklemiol Community Health 1981; 35:16-22. 21. Lee PN. Statistics of smoking in the United Kingdom. Research Paper No l, London: Tobacco Research Council, 1976. 2'2. Auerbach O, Hammond EC, Garfinkel L. Changes in bronchial epithelinm in relation to cigarette smoking 1955-19~J0 versus 1970-1977. N Engl J Med 1979; 300:381-6. 23. Doll R, Peto R. (1983). Epidemiology of cancer. In: Oxford Textbook of medicine. Weatherall I~I, l.edlngham J, Warrell I_)A, eds., Oxford: Oxford University Press, 1983: 4.51-4.79. 24+ l.lplml A. The bit~lngical effects of Io',v-level ionizing radiation. Sol Am 19'82; 246(2):29-37. 25. Garfinkel I,. Cam'er mortality in tmnsmokers: Prospeetlve Study by tile Amerk'an Cancer Fka'icty. J NCI 198t1; li5: I I lig- I 173.

2063628050

Tt~E 'LANCET, NOVEMBER 16, 1985 1111 Occasional Survey IS THERE A FUTURE FOR LOWER-TAR-YIELD CIGARETTES? Participants of the Fourth Scarborough Conference on Preventive Medicine* Summary An international workshop was held to consider whether the policy adopted in many countries to encourage the decline in cigarette tar yields was beneficial. The consensus was that the policy had been beneficial and that tar yields should be further reduced. In addition the yield of other smoke components should be reduced even in the absence of conclusive evidence of their specific toxicity. The lower-tar policy should be monitored to ensure that the concentration of smoke components (or their metabolites) in smokers declines as the yields decline. The public need to be made aware ofthe uncertainties of the policy with respect to its effects on the risk of diseases other than lung cancer and that the benefits from smoking lower-yield cigarettes are smaller than those derived from avoiding cigarettes altogether. INTRODUCTION CIGARETTE smoking is the most pressing health issue in economically developed countries. Public health policy has been directed at discouraging non-smokers from starting to smoke and encouraging smokers to stop. By the early 1980s in both the US and the UK, cigarette consumption per head had decreased. In the UK it had decreased by about 35% in men iom a fairly steady maximum spanning the period 1940-75 end decreased by about 25% in women from a peak value in 1976. In the US it decreased, in both sexes combined, by about 20% from a peak in 1963. The trends are shown in fig 1~,2 (and Tobacco Advisory Council, unpublished). Since the early 1970s, the US and UK authorities have also recommended that people who are unwilling or unable to give up smoking switch to cigarettes of lower tar and nicotine yield, in the expectation that the adverse health effects of smoking could be reduced.la The policy was never intended to be an alternative to encouraging smokers to give up smoking; nor was it expected that the benefits of smoking lower delivery cigarettes would be as great as those to be derived from stopping smoking altogether. Despite the decline in sales-weighted tar yields in the US and UK (fig 2),l't doubt has been expressed about "whether "l)articipams: N. Benowitz, M. Eeinlcib, C. Feyerabend, L. Garfinkel, R. Grcenberg, T. Guarino, J. tladdow (Co-chairman), V. Hawthorne, S. Jones, \V. Kanncl, D. Kauliuam G. Knight, L Kozlowski, M. Kunze, J. Luoto, G. Palomaki, N. Pride, G. Rose, M. Russell, R. Stepney (Rapporteur), H. van Vunakis, N. Wald (Co-chairman), J. Wilkenfield, E. Wynder. No. Gi~aret tes¢ Adult/Year ,ooo 4000 2g~ 1500 --- US men & women 500 ]- ..'* ...... UK women 0 20 25 30 35 40 45 50 55 60 65 70 75 8085 Fig 1--Annual consumption of manufactured cigarettes per adult in the USA and UK from 1920 to 1985. (Data for men and women separately are not available for the USA.) the lower-yield policy has been beneficial.%6 In this paper, which arose from discussions at a meeting in Maine, USA, in 1984, we consider the issues surrounding the advisability of a lower-tar policy. The conclusion expressed in this paper represents the general view of the group involved but on some issues there were one or two dissensions. HAVE I.OWER-YIEI.D CIGARETTES BEEN OF HELP SO FAR? Lung Cancer The carcinogenic activity of tobacco smoke seems to reside in the tar,7 so it is reasonable to expect that cigarettes yielding less tar will be less likely to cause lung cancer. However, the relation may not be straightforward. One cause of uncertainty involves "compensatory" smoking--the tendency of smokers to increase the amount of smoke inhaled from a cigarette of lower tar yield and, to a lesser extent, to increase the number of cigarettes smoked. Several studies in which the intake of carbon monoxide or nicotine have been used as an indirect measure of tar exposure have found that'the estimated reduction in tar intake is only about half of what might be expected from the difference in cigarette tar yields.8Jl Prospective epidemiological studies1~14 of lung cancer show, on average, an approximate 20% reduction in risk associated with lower-tar (or filter) cigarettes compared with higher tar (or plain)--a difference that is very much what would be expected from the intake studies. Most lung cancers still occur in filter cigarette smokers who have switched from plain cigarettes, so the full effects of falter cigarettes have not yet been seen. One case-control study that has looked at IifeiongTiiier smokers Suggests that the reduction in risk may be between 30 and 40%.15 Secular trends in lung cancer mortality and cigarette consumption in Britain indicate that the lower risk of lung cancer in smokers of lower tar compared with high-tar I. lh~pe.Simpson RE. The nature of herpes zoster: A long term study and a new hypothes~s, Pre¢ R Sac Med 1965; 58: 9-20. 2. Thomas 3,t, Robertson W], Dermal transmission of virus as a cause of shingler. Lancet 1971; it: 1149-50. 3. Berlin BS, CampbellT. Hospltal-acquired herpeszoster following exposuretochicken- pox, J,-/3L~/1970~ 2:11: 183,1-3,3,, 4. Morens DM. Bregman D J, West M, et al. An outbreak of varicella-zo*ter viru~ three- lion among cancer pat ienl~. Ann Intern Med 1980; 98." 414-19. 5. Ederer F, 3tyer~ MH, Mantel N. A statistical problem in space and time: Do leukemia ¢as~ come in clusters? Bwmetrlcs 1964; 20: 626-36. Weller TH. Varicella and herpe~ zoster, N EnglJ Med 1983; 30~: 1362-68. S. R, PALMER AND OTHERS: REFERENCES 7. Ross CAC, B ro,,~-n WK, Clarke A~ et al. Herpes zoster an general practice,.7 R Coll Pratt 1975; 25: 29-32. 8. Cradock-Wat~.on IE, Ridehalgh blKS, Bourne MS. Specific immunoglobulin res- ponses after varicella and herpes zoster. J H)~ (Camb) 1979; 82: 319-36. 9. Weller TH. Varicella and herpes zoster. N EnglJ 3led 1983; 30~: 1434-40. 10. Arvin AM, Koropchak CM, Witter AE. tmmunologtc evidence of reinfection with varicella zoster virus. J Infect Dis 1983; 148: 200-05. 11. Gershon KA, Steinberg SF, Gelb L. Clinical reinfection witi~ varicella-zoster virus. Infect Dis. 1984; 149:13,7-42. 12. Gersbon APt, Steinberg SP, Ceil.mediated immunity to varicella-zoster virus measured by virus inactivation: Mechanism and blocking of the reaction by specific antibody. Infect Immun I979; 25: 164-69. 0 O

mt3/ci£t 30 25 20 15 ]0 5 0 2.00 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 T~r u~ USA ~ I i I I I 68 70 72 74 76 78 80 - Nicotine I 82 84 UK USA I I I I I I I I 68 70 72 74 76 78 80 82 84 Year Fig 2--Sales-weighted tar and nicotine yields in the UK and USA from 1968 to 1984. cigarettes observed in epidemiological studies is not the result of self-selection. Male per head cigarette consumption changed only slightly between 1946 and 1975 (fig 1), but tar levels per cigarette decreased substantially, beginning in 1965 (fig 2). In the next few years the incidence of lung cancer began to decrease in younger men.16 Since then, a similar trend has become apparent for oIder men as well. During the same period, while tar yields decreased, cigarette consumption per head increased among women, and no reduction in female lung cancer mortality occurred. Coronary lIeart Disease The component of cigarette smoke that is responsible for the excess risk of coronary heart disease is not known, though nicotine and carbon monoxide are suspect. The nicotine and carbon monoxide yields of lower yielding cigarettes have, on average, been reduced by less than their tar yields.; Therefore, compensatory smoking results in an intake of nicotine and carbon monoxide which is not much less than that of smokers of higher yielding brands. On the assumption that coronary heart disease is due to nicotine, carbon monoxide, or a smoke constituent closely related to either, the disease is thus unlikely to be materially reduced by smoking currently available cigarettes with lower nicotine or carbon monoxide yields. Recent epidemiological observations indicate that the risk ofcoronary heart disease is not greatly affected by the yield of the cigarette. The American Cancer Society Study suggested a small decrease in coronary heart disease mortality among smokers of relatively low tar or nicotine cigarettes compared with smokers of higher yields,tz In the Framingham study, however, filter cigarette smokers had a greater risk of coronary heart disease than smokers ofnon-filter cigarettes.17 Data from the Whitehall Studyt4 were inconclusive, and the West of Scotland Study~3 found no significant difference in coronary heart disease mortality between smokers of plain and filter cigarettes. A recent study from BostonIs showed clearly that the risk of non-fatal myocardial infarction, while THE LANCET, NOVEMBER 16, 1985 increased threefold in cigarette smokers, was unrelated to nicotine or carbon monoxide yield. Therefore, apart from one study (the largest),~2 reductions in tar and nicotine yields have been found to have essentially no effect on the risk of coronary heart disease. Chronic Obstructive Lung Disease Chronic obstructive lung disease has not been extensively studied in relation to tar yields and, though the smoke components responsible for it are unknown, interest has extended to oxides of nitrogen as a possible cause. Several cigarette brands yielding lower amounts of tar and nicotine have relatively high deliveries of nitric oxide and other gases (unpublished results, UK Laboratory of Government Chemist). The evidence that lower-tar cigarettes confer a health advantage rests mainly on results from only two prospective studies. The American Cancer Society Study~9 found an association between lower-tar-yield cigarettes and a (non-significant) reduction in deaths due to emphysema. The Whitehall Study reported that lower-tar smokers produced less phlegm2° and had a slightly higher FEV, 2~ than smokers of the same number of high-tar cigarettes. DOUBTS ON TIlE FUTURE OF THE I.OWER-YIE1.D POLICY The wisdom of advocating further reductions in cigarette yield has been challenged on three main grounds. We present the argument and the response in each case. Diminishing Returns and Possibility of Encouragh~g Smoking The reductions in male lung cancer risk observed so far in the UK and US are largely attributable to the switch from non-filter to filter cigarettes during the 1950s and 1960s. There is no direct evidence that the beneficial effects which accompanied thc reduction in yield from around 35 mg to around 18 mg tar will also bc found when yields fall from the present average of 15 mg to 10 mg or below. More research on the effects of smoking modern lower-tar cigarettes is needed. Compensation might increase with further reductions in yield, leading to diminishing returns in disease prevention. If this were true and if the lower-tar policy were to encourage people to start smoking or discourage smokers from giving up the habit, the balance could be tipped against the lower-tar policy. On a larger scale, by appearing to legitimise the habit, the lower-tar policy may also militate against government efforts to encourage the avoidance of smoking. The importance of compensatory smoking should not be overemphasised. Even if further reductions in tar yields produce proportionately less benefit, any benefit would be worthwhile. Concerns that a lower-tar policy will encourage smoking do not seem to be well grounded and tar-reduction programmes may actually help people to give up smoking. In both the US and the UK, which have active tar-reduction programmes, there have been notable reductions in general smoking rates and cigarette consumption. In the American Cancer Society Survey, people who had switched to lower-tar cigarettes at the start of the study were more likely to have become ex-smokers by the end, irrespective of the number of cigarettes originally smoked. Possibility that Cigarette Enghzeering Might Increase Risk of Disease Changes in cigarette design might increase the risk of chronic obstructive lung disease or cardiovascular disease by increasing the concentration of harmful smoke components 0 O~ C~ O~ 0 THE LANCE" other than t" by increasin However, component~ lower yield~ lower-tar cardiovascu comparison Possibility t. It has bee purpose of~ on a standar may misleat tar strategy. "cheat the ~ (the US br represents a the ventilat These cigar allow smol~ 'Barclay' is is approprk typically ct smoking b3 machine sm these cham rather than Canadian unpublishc. counter tht Altering tht affect the ra are therefo~ do not retie, attempts sh having an machine sc cigarettes a' more closel gripping d~ efforts hart Continue t/ Other Nox, Despite ~ tar yield fi policy alto emphasis t mono~de i of tar yield Governme: the UK, a~ specified s; encourager This polic~ withdrawn cigarettes ~ The rob policy in tl exports of industry, 1 controls or the groum .t

, THE LANCE~T, NOVEMBER 16, 1985 other than those specifically being reduced and possibly also by increasing the extent of inhalation. However, although it is possible that certain toxic smoke components may be increased in. cigarettes with otherwise I~'lower yields, there is at present no satisfactory evidence that lower-tar cigarettes materially increase the risk of cardiovascular or chronic obstructive lung disease in comparison with current higher tar brands. Possibi]#y that Tar-tables May Mislead S~nokers It has been argued that compensatory smoking defeats the purpose of Government tables of tar yields, based as they are on a standard set of machine-smoking variables.22 Tar-tables may mislead smokers and reduce the credibility ofthe lower- tar strategy. Furthermore, it is possible for manufacturers to "cheat the machine". One brand of cigarette, in particular (the US brand 'Barclay', which uses the 'Actron' filter) represents a special case of a general problem of blockage of the ventilation holes of cigarettes with ventilated filters. These cigarettes have holes in the side of the filter tip which allow smoke drawn through the cigarette to be diluted. 'Barclay' is alleged to occupy a lower rank in the tar-table than is appropriateJ~ Its filter has ventilation channels which are typically crushed and blocked in the normal course of smoking by lips and fingers, but are not obstructed during machine smoking. With ventilated cigarettes that do not have these channels, hole blocking is thought to be a sporadic rather than a systematic consequence of normal smoking. Canadian and British data-'~'-~5 and the results of unpublished studies by the UK Government Chemist largely counter the argument that the tar-tables mislead smokers. Altering the machine-smoking conditions does not materially affect the ranking of different brands. As intended, the figures are therefore fair indications of relative yields, although they do not reflect the absolute yields to the smoker. Undoubtedly,. attempts should be made to prevent particular brands from having an unreasonable advantage under conditions of machine smoking. For example, the grip with which the cigarettes are held in the machine could be made to simulate more closely that produced by human lips--for instance, the gripping device could be elliptical instead of circular, but efforts have to be made to ensure that there is no leak. T t I E \VAY FORWARD Continue the Lower-tar Approach while Reducing Yield of Other Noxious Agents Despite uncertainties about the medical effects of reducing tar yield further, there is insufficient reason to abandon the policy altogether. It needs to be modified, though. More emphasis needs to be given to the reduction of carbon monoxide yields and those of other noxious agents. Control of tar yields, albeit by "voluntary agreements" between the Government and the tobacco industry, are already in force in the UK, and similar controls could be instituted for other specified smoke components. Yield reductions could also be encouraged by taxing higher-yielding brands more heavily. This policy was followed in the UK-~6 but was unfortunately withdrawn, despite successful reductions in the sale of cigarettes with tar yields over 20 rag. The tobacco industry has complied with the lower-tar policy in the US and the UK. It has recently also curtailed exports of high-tar cigarettes to developing countries. The industry, however, may be less enthusiastic about price controls or the control ofthe yield of other noxious agents, on the grounds that evidence of toxicity for a specific smoke 1113 Fig 3~Diagram by which smokers can gauge their intensity of smoking. Middle circle indicates the staining of a cigarette butt with standard smoking procedure; left circle represents relative "under-smoking" and right circle relative "over-smoking". (Supplied by L. Kozlowski.) component should be provided before the yield of that component is restricted. At first sight, this seems reasonable. However, the chances of showing that any single component of tobacco smoke is responsible for a particular disease is small, not because the component is harmless, but because the studies required are difficult, if not impossible, to carry out. Cigarette smoke inhaled into the lungs is one of the most toxic environmental hazards in general life, but the exact chemicals responsible and their modes of action remain largely unknown. In the face of these difficulties it is unreasonable to demand evidence of toxicity for individual • chemicals before preventive action is taken. To do so would be like resisting demands for clean drinking water until the precise microorganisms responsible for disease were known. it is sensible public health policy to focus attention on broad components of tobacco smoke for which there is general evidence of toxicity, such as tar, while at the same time ensuring that the concentration of other components likely to be harmful are reduced as well. A gradual reduction in the concentration of components such as carbon monoxide and oxides of nitrogen, based on knowledge of their biological effects, is more likely to change mortality and morbidity for the better than for the worse. Publicising the reduction in yields other than tar (preferably on the packet as ffell as in separate tables) will draw attention to the risk of diseases other than lung cancer, such as coronary heart disease, which are caused by cigarette smoking. Implement Biochemical Epidemiological Monitoring The continuation of the lower:tar policy and its possible extension to other noxious agents in cigarette smoke needs to be monitored. There are practical difficulties in doing this with disease or death as end-points. An alternative and more manageable approach is to measure exposure to smoke components directly, by the use of biochemical markers such as cotinine and carbon monoxide in blood. The application of such "biochemical epidemiological" techniques may help predict changes in mortality and morbidity without having to wait for the full pathological effects. Investigate Contpensatory Smoking A medium-nicotine low-tar cigarette has been proposed as one which might reduce the extent to which smoke from a cigarette is inhaled.-'7 The effect ofnicotine yield (and other features of a cigarette, such as i~g draw resistance) on the extent of compensation in the general population needs further investigation. The public health position on whether nicotine yields should be maintained can then be clarified. Increase Awareness of Possible Dangers of Compensatory Smoking At the same time as the lower yield approach is pursued, governments should make smokers more aware of the reality and potential risks of compensatory smoking. Kozlowskie.~

, • '•° , , 1,114 has suggested that cigarette p~ckets might contain a simple illustration (fig 3) showing the extent to which the end of a filter is stained by smoking. Darker staining would suggest oversmoking relative to the machine and so provide a guide to the absolute yield being obtained, and how it can be reduced. CONCLUSION There is a future for lower-tar yield cigarettes, but the aim should be to reduce the yield of other smoke components as well as of tar. Biochemical monitoring of the concentration of smoke components (or their metabolites) in smokers can ensure that exposure is on average reduced even if this reduction is less than would be expected from the reduction in machine-smoked yields on account of human compensatory smoking. The public needs to be made aware of the uncertainties of the policy, particularly those arising from compensatory smoking, and also that the benefits of smoking lower-yield cigarettes can only be small compared with those of avoiding the smoking habit altogether. The lower-yield approach is but one facet of a general strategy aimed at reducing the extent of disease caused by smoking in societies in which some people will continue to smoke regardless of the adverse long-term consequences to their health. This paper arose from papers and discussion at the Fourth Scarborough Conference on Preventive Medicine held in September, 1984, in Scarborough, Maine, USA. The meeting was sponsored and supported financially by the American Cancer Society, Maine Division, Esther G. Dachslager Fund, and the American Heart Association, Maine Affiliate Inc. We thank the Tobacco Advisory CounCil, the Laboratory of the Govern- ment Chemist, the US Office of Smoking and tleatth, and Stephanle Kiry|uk for helping to providc certain data referred to in thc paper. Correspondence should be addressed to N. J. W., Department of Environmental and Preventive Medicine, Medical College of St Bartholomew's ttospitat, Charterhouse Square, London ECIM. 6BQ. REFERENCES 1. The health consequences of smoking: A report of the surgeon general. US Department of 11eahh and tluman Serwccs, l'ubhc Ilealth Service, Oll'tce of Smoking and Ileahh, 198l. 2. Lee PN, ed. Statistics of smoking m the Umtcd Kingdom, research paper 1.7th ed. I.ondon: Tobacco Research Gounod, 1976. "L Third Report of the Independent SctcnUfic Committee on Smoking and health. Chairman: Peter F~ogg~tt. London: tim Stauonery office, 198~: 11. 4. Wold N J, Doll R, Copeland G. "Fronds in tar, nicotine, ~nd carbon monoxide yields of UK cigarettes manufactured since 1934. BrMedff 1982; 28~." 76]-65. 5. Prevention of coronary heart disease: report of a WHO expert committee. 1~'I10 Tech Rip Set, 678. Geneva; World tlcalth Organisatinn, 19~2: 27. 6. Gersteln DR, 1.evison PK, eds. Reduced tar and nicoune cigarettes: smoking behavior and health. Washington DC: National Academy Press, 1982: 5. 7. Wynder EL, Hoffmaa D. Tobacco and tobacco smoke: studies m experimental carcinogenesis. New York: Academic Press, 1967: 529. 8. Wold NJ, Idle M, Boreham l, Bailey A. Inhaling habits among smokers of dilTerent types of cigarette, Thorax 1980; 35: 925-28. 9. Ashton tt, Stepney R, Thompson JW. Self.titration by cigarette smokers. Br Med] 1979; it: 357-60. 10. Russell MAH, Sutton SR, lyer R, Feyerabend C, Vesey CJ. Long term switching to tow-tar low nicotine cigarettes. BrffAddict 1982; 77: 145-48. I 1. Stepney R. Would a medium-nicotine, low-tar cigarette be less hazardous to bealth? Br .44edff 198t; 283:1292-96. t2. Hammond EC, Garfiokel L, Seidman H, Lew EA. "Tar" and nicotine coatem of cigarette ~moke ha relation to death rates. Env Res 1976; 12: 263-74. IL Hawthorne VM, Fry JS. Smoking and health: the association between smoking behaviour, total mortality, and cardiorespiratory disease in W est Central Scotland.ff Epidemiol Comm tllth 1978; 32: 260-66. 14. l-liggenbmtam T, Shipley MS, Rose G. Cigarettes, lung cancer, and coronary heart disease: the effect~ of inhalation and tar yield. J Epidemiol Comm Hlth 1982; 36: 113-17. 15. Lubin JH, Blot WJ, tlertino F, et at. Panerna of lung cancer according to type of cigarette smoked, lot ~ Cancer 1984; 33: 569-76. 16. Wold N'J. Smoking. In: Vessey MP, Gray MJA, eds. Cancer risks and prevention. Oxford: Oxford University Press, 1985: 44-67. 17. Castelli WP, Dawbet TR, Feinleib M, Garrison R J, McNamara PM~ Kannel WB. The filter cigarette and coronary bean disease; the Framingham study. Lancet 1981; it: 109-1J. 18. Kaufmzn DW, Helmrlch SP, Rosenberg L, Miettinea OS~ Shapiro S. N.icotine and carbotx m~noxide ¢~ntent of cigarette smoke and the risk ~f myocardial irffarct iota in young men. NEngl.~Med 198J; 1108: 410-1~. References conthnted at foot of next column THE LANCET, NOVEMBER 16, 1985 Point of View COMMUNITY GENERAL PRACTITIONER DAVID MANT PETER ANDERSON Oxfordshire Health Authority, Manor House, Headley l~ay, Oxford OX3 9DZ Summary The attainment of quality in general practice entails explicit recognition of the public-health content of primary care. General practitioners should accept responsibility for auditing the state of the practice health, monitoring and controlling environmental disease, planning local services, auditing the effectiveness of preventive programmes, and evaluating the population effects of medical intervention. This requires specific training in the skills of population medicine, reallocation of scarce resources, and cooperation with existing public-health doctors. Eventual integration of community medicine and general practice is desirable. INTRODUCTION "The world needs a new kind ofdoctor, one who combines clinical skills with the skills of population medicine." This exhortation by Hart~ applies not only to general practice but to all areas of medicine. It is ironic that this dichotomy exists in the United Kingdom, which boasts a population-based system of primary care, an expertise in population medicine which is unrivalled in Europe, and a National "Health" Service. The divisions in the structure of the health service which have led to a community-medicine specialty without access to the community and a primary-health-care system without responsibility for the community's health is not an evolutionary accident. Some of the political forces that led to a tripartite structure in 1946 have not weakened. However, a growing number of doctors working within the constraints imposed by these damaging divisions are eager for change. The Royal College of General Practitioners has set the lead with its recent emphasis on prevention, and in practical terms the involvement of general practitioners in this field is increasing: for example, the proportion of cervical smears taken by general practitioners in Oxfordshire has increased from 40% to over 70% in the past decade. SimiLarly the number of community physicians who wish to see community medicine live up to its name is also growing. It was therefore disappointing to read the two recent papers on quality of care published by the Royal College~'3 which fail to include the integration of the skills of popu)ation N. WAI.D AND OTIfERS: REFERENCES--COtIti?Iu¢d 19. Lee PN, Garfinkel L. btor tality and type of¢igarene smoked, ff Epidemlol Gomm tilth 1981;85: 16-22. 20. nlgenbonam T, Sbapley MJ, Clark TJtt, Rose G. Lung function and symptoms of cigarette smokers related to tar yield and number of cigarettes smoked. Lancet 1980; i: 409-12. 21. Lee PN. Low tar ctgarene smoking. Lancet 1980; i: 1365-66. 22. Hatriman E. Turning the tables. The Guardian, May 2, t984: p 11. 23. Harriman E. Tar table 'chca~' are sued. NewSciemist July 21, 1983: 175. 24. Rickert WS, Robinson JC~ Young JC, Collishaw NE, Bray DF. A comparison of the )fields of tat, nicotine, and carbon monoxide of ~6 brands of Canadian cigarettes tested under three conditions. Preheat Afed 198,3; 12: 682-94. 25. Rawbone RG. Switching to low tar cigarettes: are the t~ league tables relevant? Thorax 1984; 39: 657-62. 26. Editorial. Silent prevention. Lancet 1979; i: 705-06. 27. Russell MAH. Low-tar medium-nicothae cigarettes: A new approach to ~afer smoking. Br3ledff 1976; i: 1430-33. 28. Kozlowski LT, gickert WS, Pope MA, Robinson JC. A color-matching technique for monitoring tat/nicodna yields to smokers. Am ff Publ Hlth 1982; 7~: 597-99. THE LANCI medicine a be argued commitme However, to make o debate we of"qualit3 The "co is a revol democrac~ define the care team achieved communit achieved i practice a~ same agail | The State It is per public-he: general p health of~ for the Ct The State documenl primary c main det~ the publk on health accident ~ drinking ~ local high the punk be taken. informin~ communi to his ad' from his The "Dr, The in sewers ir There is reLating t better do is prima~ collected and bact Health I extra di~ perhaps of sentil establish There tasks wh disease t controllt exclusio disease ~ prophyk neither t is the environ1 areas an l

2063628055

WINKELSTEIN, W. "SOME ECOLOGICAL STUDIES OF LUNG CANCER AND ISCHAEMIC HEART DISEASE MORTALITY IN THE UNITED STATES" BOOK 53; TAB 7 HERE REMOVE WHEN ARTICLE HAS ARRIVED

2063628057

The value of preventive medicine _'stablished in IBA-GEIGY medical and each year on papers and ,n also holds ide scientific Ciba Foundation Symposium 110 ~11 the rides on in the house UNIVERSITY of NORTH CAROLINA 1985 Pitman London NAR 14 1985 HEALTH SCIENCES LIBRARY

Control of tobacco-related disease RICHARD PETO Cancer Studies Unit, Radcliffe Infirmary, Oxford OX2 6HE, UK Abstract. As ways of discouraging tobacco consumption, the effects of increases in price and in quantitative information may have been under-emphasized. To decrease the hazards of tobacco, switches from cigarettes to pipes, cigars or 'smokeless' tobacco may be useful, as may a reduction in cigarette tar delivery. Indeed, the spread of existing tar level reductions from capitalist to socialist countries might prevent tens of thousands of lung cancer deaths each year in the early decades of the next century, and (perhaps by attempts to engineer cigarettes so that smokers of lower tar cigarettes are less likely to "compensate' by taking more smoke) it should eventually also be possible to change cigarettes so as also to reduce their effects on heart and lung disease. Changes in consumption and in composition of tobacco products are complementary, not competing, strategies. If both are pursued effec- tively, then although the life expectancy of old people may not be much improved, the proportion of adults who die before reaching old age will decrease substantially. 1985 The value of preventive medicine. Pitman, London ( Ciba Foundation symposium 110) p 126-142 The obvious way to avoid tobacco-related disease is to avoid smoking, and eventually the most important way of controlling tobacco-related disease will be by vast reductions in the extent to which tobacco, particularly in the form of cigarettes, is smoked. But at the moment, although cigarette consumption is going down in certain countries, worldwide cigarette consumption is going up. It is therefore an extremely destructive form of idealism to consider only the discouragement of cigarette consumption but to avoid considering other ways of modifying the extent to which cigarettes kill people. There are, at least in principle, three different ways of reducing tobacco-related disease. First, one may be able to modify the "host'--the smoker. Mention has been made at this symposium of the Japanese, who appear as a nation to eat less fat than we in Britain, and who moreover appear to eat fat with a higher "P:S ratio' (i.e. with a higher ratio of polyunsaturated to saturated fats). Although smoking has been widely prevalent in Japan for more than 30 years, Japanese death rates from heart disease are still, even in early middle age, very much lower than those in the US or Britain. In other words, the absolute heart disease risk per cigarette is much lower in Japan than in Britain or America. 126 TOBACCO-Rt Studies of Ja: non-genetic A second particular eit some less ha: changes in c filter tips envisaged. British gove: Health (1982- cigarette, bt about the c~ tobacco in a alternatives tobacco (inc is, powderec exact hazart liminary evit unlikely to produced b~ the young er The third sumption. Ir because wid risks are un. two hundred continued to moment, ho must also cc tobacco prot on other top consumptior Changes in tt Probably in, aged: from c haps. cigars) consider the merely becat

ase lects of increases in price and . To decreabe the hazards of ,s tobacco may be useful, as existing tar level reductions asands of tung cancer deaths ups by attempts to engineer x to "compensate" by taking garette550 as also to reduce ,tion and in composition of :s. If both are pursued effec- not be much improved, the •ase substantially. rton symposium 110) O avoid smoking, and • co-related disease will • ticularly in the form of zarette consumption is nsumption is going up. rn to consider only the :onsidering other ways • There are, at least in ed disease. ker. Mention has been • a nation to eat less fat vith a higher 'P:S ratio' rated fats). Although aan 30 years, Japanese niddle age, very much :Is, the absolute heart n Britain or America. TOBACCO-RELATED DISEASE 127 Studies of Japanese in America suggest that these differences are due chiefly to non-genetic 'host factors'. A second way is to affect the manner in which tobacco is used, and in particular either to encourage a switch from cigarettes to the use of tobacco in some less hazardous form, or to modify the composition of cigarettes. The chief changes in cigarette composition thus far have involved the introduction of filter tips and the lowering of tar levels, but other modifications could be envisaged. One possible change was discussed in the most recent report to the British government of the Independent Scientific Committee on Tobacco and Health (1983). This was the possible introduction not of a low-nicotine, low-tar cigarette, but of a medium-nicotine, low-tar cigarette. As well as thinking about the composition of cigarettes, we should also think about the use of tobacco in a much less hazardous form, including perhaps not only the usual alternatives such as pipes and cigars, but also various forms of 'smokeless' tobacco (including nasal snuff, chewing tobacco, and 'dipping' tobacco--that is, powdered tobacco that is usually held between the gum and cheek). The exact hazards associated with such habits are not yet known, but the pre- liminary evidence thus far available (Winnet al 1981) suggests that they are unlikely to be substantial in comparison with the vast mortality now being produced by tobacco smoking (unless promotion of smokeless tobacco among the young engenders nicotine addiction that eventually leads to smoking). The third and most important way is, of course, to decrease tobacco cor;- sumption. In the end, this is going to be the solution that is finally adopted, because widespread tobacco use is (at least if the product is smoked and the risks are understood) unacceptably hazardous. Eventually--perhaps one or two hundred years from now--people will perhaps find it amazing that tobacco continued to be smoked so widely in the second half of the 20th century. For the moment, however, cigarette consumption is going up Worldwide, and so one must also consider the first two factors (modification of the host, and of the tobacco product being used). I will leave modification of the host to speakers on other topics, however, and will be concerned only with the composition and consumption of tobacco products. Changes in the way in which tobacco is used Probably in decreasing order of efficacy, three types of change may be envis- aged: from cigarettes to smokeless tobacco, from cigarettes to pipes (or, per- haps, cigars), and from more hazardous to less hazardous cigarettes. I shall consider the latter at greatest length, not because it is the most effective, but merely because politically it may be the easiest to achieve.

128 PETO TOB~~ Smokeless tobacco In the South-Eastern United States, many women have, throughout their adult lives, habitually 'dipped snuff--that is to say, they have placed powdered tobacco between the gum and cheek, and have thereby absorbed a number of pharmacologically active substances. It is noteworthy that among snuff- dippers in the South-Eastern US. the proportion who smoke (15%) is much lower than that among other women (45%) (Winn et al 1981). Moreover, even among those who do smoke the consumption of cigarettes per smoker is slightly lower among the "dippers'. These observations suggest that snuff dipping discourages smoking in that particular population, giving some hope that it might also do so elsewhere. Vigorous commercial promotion of snuff dipping has begun in America, and is just beginning in Europe. If this or some other such habit were to become widespread and did to any substantial extent replace smoking (particularly of cigarettes), then the net effect would be likely to be a reduction in tobacco-induced mortality. For, although snuff dipping causes a vast increase in the relative risk of cancer of the gum and cheek (together with the same sort of risks of cancers of other parts of the mouth that smoking produces), the absolute excess risks of death from oral cancer associated with the habit in the South-Eastern United States appear to be at most a few per cent of the total risk of death produced by cigarette smoking (Winn et al 1981). Although the absolute risks in other populations might, of course, be con- siderably different (especially if some diseases other than oral cancer are found to be increased by tobacco 'dipping'), the use of smokeless tobacco is still likely to be much less hazardous than is tobacco smoking, especially of cigarettes. Switch of smoking from cigarettes to pipes and/or cigars In prospective observations of male British doctors, (i) lifelong non-smokers, (ii) men who currently smoked only pipes and/or cigars (most of whom smoked only pipes), and (iii) men who currently smoked cigarettes had age- standardized death rates in the ratio 1: 1.09:1.64 (Doll & Peto 1976). This may not exactly reflect the relative hazards of the two forms of tobacco, but, to- gether with much other evidence, it does correctly suggest that habitual use of pipes is much less hazardous than is habitual use of cigarettes. Since, more- over, about half of the men who smoked only pipes and/or cigars had pre- viously smoked at least some cigarettes, it also strongly suggests that a wide- spread switch from cigarettes to pipes could save many lives, even though men who switch may inhale their pipe smoke in ways that are more hazardous than the usual inhalation patterns among lifelong pipe smokers. Chan Sever prod1 takin This effec~ smok lung than ~ less h Thirc cigar Th cflnc~ evidt the cigar (i) L Lun smo adul of c: age age redti rele 193~ (i.e Scar 195t per the por, to e unl~ corf

PETO hroughout their adult rye placed powdered • absorbed a number iv that among snuff- moke (1597) is much ~81). Moreover, even per smoker is slightly ,t that snuff dipping ng some hope that it orion of snuff dipping If this or some other atantial extent replace ould be likely to be a auff dipping causes a cheek (together with mouth that smoking incer associated with at most a few per cent g (Winn et al 1981). . li~ourse, be con- 3~lncer are found ; to5-acco is still likely ially of cigarettes. .-'elong non-smokers, ars (most of whom l cigarettes had age- )eto 1976). This may of tobacco, but, to- that habitual use of rettes. Since, more- l/or cigars had pre- uggests that a wide- Is, even though men rare hazardous than TOBACCO-RELATED DISEASE 129 Changing the composition of cigarettes Several questions are relevant. First, scientifically, how can a cigarette be produced that is less hazardous in use? (N.B. Smokers tend to 'compensate' by taking more smoke from today's low-tar than from the old high-tar cigarettes.) This first question may conveniently be divided into (a) assessment of the effects of changes that have already been introdiaced on the three principal smoking-related diseases (lung cancer, heart disease, and chronic obstructive lung disease), and (b) design of cigarettes that are likely to be less hazardous than those currently being sold. Second, politically, how can the switch towards less hazardous cigarettes be encouraged in Western and in Eastern countries? Third, psychologically, to what extent (if any) will the pursuit of less hazardous cigarettes impede measures to reduce cigarette sales? The three main diseases associated with cigarette smoking are (i) lung cancer, (ii) heart disease, and (iii) chronic obstructive lung disease. For lung cancer, but not for the other two diseases, there is now reasonably good evidence that the changes in cigarette composition that have taken place over the last few decades in Western countries have reduced the hazard per cigarette. (i) Lung cancer Lung cancer risks in adult life depend surprisingly strongly not only on recent smoking habits, but also on smoking habits many decades beforehand, in early adult life. Thus, for example, among 60-year-old habitual smokers of one pack of cigarettes per day, those who began to smoke cigarettes regularly at about age 15 have more than twice the lung cancer risks of those who began at about age 25 (Doll & Peto 1981, Appendix E). This suggests that if the tar level reductions have any worthwhile effect, then tar levels in early adult life may be relevant to lung cancer risks in middle age, mar.y decades later. Between the 1930s and the 1970s there have been reductions of more than 50% in the mean (i.e. sales-weighted) tar delivery per cigarette in the United States, Britain, Scandinavia, and a few other places. These changes were small until the late 1950s and then they suddenly became rapid, with decreases from 30-odd mg per cigarette in the mid-1950s down to approximately 15 mg per cigarette by the 1970s. The changes are not expensive and involve the use of filter-tips, porous paper (or even. as an extreme measure, 'ventilated' filters that allow air to enter into the side of the filter to dilute the smoke) and modified types of tobacco (which may in some instances actually be less .expensive than unmodified tobacco). There is, of course, a reduction not only in the unwanted components of the smoke but also in those substances (e.g. nicotine?) to which

130 PETO TOBACCO-RE[ some smokers are addicted, and when such reductions occur many smokers are likely to compensate, either by smoking more cigarettes* or, perhaps more commonly, by taking in more smoke per cigaretteS-. It appears, however, that the latter form of compensation is not always sufficient to outweigh the reduc- tion in tar (Wald et a11980), in which case the net result will be inhalation of less tar into the lung. This conclusion is suggested both by common sense and by observation, but even if it is accepted it does not prove that the hazards will be correspondingly reduced, for despite some 30 years of laboratory research the importantly carcinogenic factors in cigarette smoke have not yet been identified reliably. Moreover, it is difficult to predict how changed patterns of inhalation will change what is deposited on the main target areas--which, for lung cancer, are not the peripheral tissues, but the large airways--as the smoke streams past them. Consequently. it is necessary to discover by direct epidem- iological observation whether the risks of lung cancer are materially reduced by the widespread switch to lower tar cigarettes. Unfortunately, this is not easy to do, for not only are smokers of low-tar brands self-selected but also, just as it is only after some decades of smoking that the full risks materialize, so perhaps it is only after some decades of using low-tar cigarettes that the full risk avoidance will materialize. Therefore, even if the effects in late middle age will one day be substantial, they may not yet be. Any substantial effects that are going to materialize in early middle age should be beginning to be evident by now in Britain, however, for although the tar reductions of the 1950s were only moderate, those of the 1960s were substantial in Britain, North America and Scandinavia. Thus, a 40-year-old in 1980 will have been smoking from about 1960 to 1980, throughout most of which time tar levels were substantially lower than in previous decades. Two main pieces of epidemiological evidence are currently available, the first being the results from classical case--control or prospective surveys. Unfor- * In principle, tar reductions could either increase or decrease the number of people who smoke (by making it less of an ordeal for non-smokers to acquire the habit or by making the habit less addictive) and could either increase or decrease the number of cigarettes one individual smoker consumes (by increasing the number needed to achieve a given dose or by decreasing the satisfac- tion per cigarette). In practice, however, the patterns of cigarette consumption in different coun- tries do not appear to be influenced consistently in either direction by changes in cigarette composition. "1" Surprisingly, there appears to be little reliable information on which of the many characteristics of the cigarette (e.g. nicotine, draw resistance, taste) importantly affect 'compensation'. If these could be identified and modified (e.g. by increasing the nicotine delivery, draw resistance or whatever of low-tar cigarettes), then maybe the intake of many toxins could be decreased simul- taneously. Such compensation presumably underlies the recent disappointing finding by Kaufman et al (1983) that the risks of myocardial infarction are not materially different among smokers of different types of cigarette. tunately, such relate chiefly t and even rece cigarettes for~ bated in studi~ the tar reducti. less extreme, t overall tar lev~ concurrent co with people o brands. Despi all the case-c~ that: 'a reasonabl tar/nicotine tar/nicotine smoking.. smoked the smoking liv~ even greate: improve ev~ Because of, concurrently, early middle a~ tes for much o can be supplen the study of na (Doll & Peto those from tht reductions are from the delay better to use t! began), Britist~ mately stabiliz and the reduct from changes i likely to be far in early middle Moreover, bot accelerating dc the next decad

PETO TOBACCO-RELATED DISEASE 131 , smokers are ,erhaps more ~owever, that ~h the reduc- ~lation of less .,ense and by tzards will be research the ot yet been ~d patterns of s--which, for -as the smoke :.irect epidem- 'ly reduced by is not easy to ,o. just as it is so perhaps it isk avoidance ill one day be are going to .nt by now in re only and g from about antially lower tvailable, the rveys. Unfor- ople who smoke ng the habit less dividual smoker sing the satisfac- ~ different coun- ges in cigarette ty characteristics ~sation'. If these aw resistance or :lecreased simul- ling by Kaufman nong smokers of tunately, such data as are currently available are limited by the fact that they relate chiefly to late middle or old age. when most of the lung cancers occur, and even recent studies relate chiefly to people who have smoked low-tar cigarettes for only a fraction of their smoking lives. This difficulty is exacer- bated in studies performed during the 1960s (or early 1970s) by the fact that the tar reductions then available for study were not only more recent, but also less extreme, than those now available. A related source bf difficulty is that as overall tar levels decrease, the higher tar levels simply cease to exist, so direct concurrent comparison of people now on low-tar cigarettes can be only with people on moderate-tar cigarettes, and not on the old very high tar brands. Despite these difficulties, when Lee & Garfinkel (198I) reviewed all the case-control and prospective studies then available they concluded that: 'a reasonably clear picture has emerged. This is that smokers of filter (or low tar/nicotine) cigarettes have a lower mortality than smokers of plain (or high tar/nicotine) cigarettes for those diseases most strongly associated with smoking... These reductions in mortality have been seen in those who have smoked the more modern types of cigarettes for only a small part of their smoking lives. The fact that those who have smoked them for longer show even greater reductions in mortality suggests that the overall picture will improve even more in years to come." Because of difficulties of self-selection, of comparing the new with the old concurrently, and of characterizing individuals' recent lung cancer rates in early middle age (i.e. the rates among people who have smoked low-tar cigaret- tes for much of their adult lives), the case-control and prospective survey data can be supplemented usefully by a second type of epidemiological data, namely the study of national trends in early middle age. However, for obvious reasons (Doll & Peto 1981). it is not advisable to use for this purpose data (such as those from the United States) in which any downward trends caused by tar reductions are likely to be diluted or even reversed by upward trends resulting from the delayed effects of past increases in tobacco consumption. Instead, it is better to use the British data. For, by the 1950s (when the rapid tar decreases began), British male lung cancer rates in early middle age had already approxi- mately stabilized (Table 1). Table 1 also describes their subsequent evolution, and the reductions are extremely impressive. They are most unlikely to result from changes in air pollution, for not only are any effects of the air pollution likely to be far smaller than this (Cederlrf et al 1978), but also similar halvings in early middle age have been seen over the last 20 years in unpolluted Finland. Moreover, both in Finland and in Britain the changes appear, if an.vthin~, to be accelerating downward, so if this pattern carries on into late middle age during the next decade or two, then at least in these two countries (where the male

132 PETO TABLE 1 Recent trends in England and Wales in male lung cancer death certification rates in early middle age"'b Death certification rates per million men from cancers of the respirator),' tract, excluding larynx Age 1951-1955 1956-1960 1983 Ratio (yrs) la) (b) (c) (c/b) 30-34 38c 37c 10 0.3 35-39 101¢ 95c 37 0.4 40-44 253c ~6~ 112 0.4 45-49 589¢ 597~ 2950 0.5 • Note both the approximate constancy before tar deliveries began to be greatly reduced and the large decrease thereafter. b Note: (1) These trends are not materially affected by changes in curative treatment of the disease. (2) Sales-weighted tar levels started to fail rapidly only at the end of the 1950s but are now less than half of what they then were. (3) Mean daily cigarette consumption per British male aged 30-50 did not change greatly until the past few years, and in 1955, 1965 and 1975 was respectively 10.5, 9.9 and 10.2 (Lee 1976): the 10-20% decrease in consumption that existed in the second half of the 1970s is too small and too late to be chiefly responsible for the large decreases in lung cancer mortality rates in 1983. (4) In unpolluted Finland, where male cigarette smoking also began so long ago that the trends had nearly flattened out by the late 1950s, male lung cancer rates in early middle age have likewise been approximately halved over the past 20 years, and in both countries the decreases appear, if anything, to be accelerating. (In Finland, as in Britain, no large changes in cigarette consumption were evident before the mid-1970s.) ¢ High mean tar intake throughout smoking history. d High intake only in first decade or so of smoking history. death rates are at present uniquely high) lung cancer may some day decrease for a few years* as fast as it once increased. (ii) Heart disease Perhaps because of the substantial extent to which smokers 'compensate' for tar delivery reductions, there is disappointingly little evidence of any favour- able effect of such reductions on heart disease. The studies reviewed by Lee & Garfinkel (1981) did in aggregate suggest some slight benefit, but the rather * It will not decrease to anywhere near non-smoker rates, however, unless there is widespread abandonment of cigarette smoking. Similarly, in those other populations where lung cancer rates have not yet completed their rise, even a tar-level reduction that halves the carcinogenicity of cigarettes may merely slow, rather than reverse, the progressive increase of the disease over the next few decades. TOBACCO better stuc no evident cancer ris numbers o rarity of ti slight but v disease (ai chief need tunately, t vestigated. If so, a I hazardous responsiblt ants, dra~ importantl minimize t This 'bla tant carci~ disease firs by random modified t\ The effe~ smoker is ~ she will ta difference study phar~ twice as mt There max only a fe~ nicotine, d, cigarette t~ tive lung di (iii) Chron, COLD is a ponse relat disease ma~ rates are ri that if we d should be a

PETO ifieation rates in early Rtltlo tc b~ o.3 0.4 (}.4 0.5 :atb reduced and the of the di.~ease. but are nov, less than ~nge greatly until the 10.2 {Lee 1976): the , is too small and too tv rates in 1983. :o that the trends had .,e have likewise been decreases appear, if garette consumption ne day decrease :ompensate' for .~ of any favour- iewed by Lee & • but the rather there is widespread re lung cancer rates : carcinogenicity of he disease over the TOBACCO-RELATED DISEASE 133 better stud)' of Kaufman et al (1983) suggests none. At least, however, there is no evidence of any adverse effect to set against the apparent reductions in lung cancer risk. and perhaps larger case-control studies, with even greater numbers of people in early middle age (i.e. in their forties or even, despite the rarity of the disease among young adults, in their thirties) will reveal some slight but worthwhile differences between one cigarette and another. For heart disease (and, probably, for chronic obstructive lung disease), however, the chief need is to design a cigarette that will minimize 'compensation'. Unfor- tunately, the exact determinants of compensation have not been properly in- vestigated. Perhaps. for example, many smokers smoke largely for nicotine. If so, a medium-nicotine, low-tar cigarette might be substantially less hazardous (for there is some evidence that nicotine itself is not chiefly responsible for the cardiotoxicity of cigarettes). Alternatively, perhaps flavor- ants, draw resistance, acidity variations or other manipulable factors might importantly affect "compensation', and might therefore be modifiable to minimize the amount of smoke taken per cigarette. This "black box" approach to cigarette design does not require that the impor- tant carcinogens, cardiotoxic agents and causes of chronic obstructive lung disease first be identified, and hence it could be pursued immediately, perhaps by randomized trials measuring compensation in smokers of various suitably modified types of cigarette. The effects of changes in tar delivery on compensation are quite marked: if a smoker is given a high-to-medium tar cigarette and a low-tar cigarette, he or she will take twice as much smoke from the low-tar cigarette. A twofold difference like that• which occurs immediately, should be relatively easy to study pharmacologically. What ingredient in the smoke tells the smoker to take twice as much from it? Is it the nicotine? Is it the draw resistance? Is it the taste? There may be several determinants of compensation, but there are probably only a few major ones. If we could identify one or two of them that, like nicotine, don't appear to be important toxins, it should be possible to produce a cigarette that would give less risk of lung cancer and less risk of chronic obstruc- tive lung disease. (iii) Chronic obstructive lung disease (COLD) COLD is as specifically related to smoking as is lung cancer, as the dose-res- ponse relationship among British doctors shows (Doll & Peto 1976). This disease may be decreasing in Britain, but in some countries, such as the US, the rates are rising rapidly. The dose-response relationship for COLD suggests that if we decrease the extent to which people take smoke from cigarettes, we should be able to produce a cigarette conferring less hazard of this disease. It is

134 PETO TOBACCO-REL. difficult to test this directly, however, because of the natural history of COLD (Peto et al 1983). We cannot use the approach used on trends for lung cancer, because although almost all cases are caused by tobacco, there are other impor- tant determinants of chronic obstructive lung disease. At a time when there were few marked differences in the smoking habits of the different social classes in the UK, COLD was much more common as a certified cau_se of death in the lower than in the upper social classes, in both men and women. Since then, from the late 1960s to the late 1970s, death rates among middle-aged men have halved from this disease in Britain, and are continuing to fall. It is very difficult to produce a specific explanation for these trends. One cannot confidently ascribe them to changes in cigarette composition. Air pollution decreases pro- vide an obvious explanation, but I am not sure that it is the correct one. Thus, the study of national mortality trends may not be directly informative about the effects of the post-1960 changes in cigarette composition on COLD mortality. Classical case-control studies are likely to be even less informa- tive-indeed, since severe COLD decreases cigarette use, they might even yield the inverse of the truth. A possible compromise might be to study the relationship between cigarette composition and the one-second Forced Expira- tory Volume (FEV) in early middle age (e.g. 35-44), before the FEV had got low enough to have much effect on smoking habits, but no recent such study has attempted this. Thus, for COLD, there is as yet no direct assessment, on an individual basis, of whether there are any important differences in the disease onset rates (i.e. in FEV loss) that are produced by different types of cigarette. In any case, even if there were, one would still not know which components of smoke were chiefly responsible. So, the above recommendations for seeking ways of producing decreases in compensation (with respect to all but a few smoke components) may be the most immediately promising means of cigarette modification to explore, for COLD as for heart disease. Changes in the amount of tobacco used Although the type of change in cigarette composition that has been introduced over the past few decades may reduce the risk of death per cigarette by a small but worthwhile amount, and although experimental investigation of the physi- cal and pharmacological determinants of compensation may lead to the design of cigarettes with still lower risks, cigarettes are still likely to kill about 20 or 30% of those who use them regularly. Effective discouragement of their use is difficult (and, in different countries, these difficulties may be quite different), but it is so uniquely worthwhile that it deserves even more attention than it gets. A variety of voluntary orgar and these deser There are, hob deficient. The fi many governm~ that may be con tax revenues, e principle goven in practice the3 government is b the tax on toba when sales are d especially becaL evidence of effe A second de! stress on the Ion ial across about : developed count admittedly diffic remembered aF framework of ot! possible, so long sages that qualfl QUARTER Ol~ KILLED BEFC less complicated groceries, car p consumer societ matter for exper Britain) be helpf SMOKING IS Among 1000 3' habout 1 will --about 6 will --about 250 wi Even in the Uni double those in E some such comp~ start with only 10 25 of them with d Whatever forn o o~ o%

PETO .ral history of COLD ends for lung cancer, aere are other impor- ,t a time when there i fferent social classes cause of death in the women. Since then, :ddle-aged men have all. It is very difficult cannot confidently ution decreases pro- e correct one. directly informative nposition on COLD . even less informa- ~e, they might even ight be to study the :ond Forced Expira- ~re the FEV had got ecent such study has • an individual basis, ;e~et rates (i.e. in • ~I~Y case, even if smoke were chiefly ways of producing ,moke components) ~tte modification to aas been introduced cigarette by a small gation of the physi- y lead to the design • to kill about 20 or .ment of their use is be quite different), re attention than it TOBACCO-RELATED DISEASE 135 A variety of WHO and UICC expert reports have been prepared on how voluntary organizations and governments can decrease cigarette consumption, and these deserve careful scrutiny, for they contain much well-judged advice. There are, however, two important respects in which they may be somewhat deficient. The first is that tax increases are not sufficiently emphasized. Because many governments derive large tax yields from tobacco sales, all the strategies that may be considered for reducing cigarette sales will also, if effective, reduce tax revenues, except for an increase in the taxation of tobacco. Although in principle governments may believe they act only for the good of their citizens, in practice they may tend to decide that what is economically easiest for the government is best for the citizens. Consequently, the one strategy--increasing the tax on tobacco---that increases rather than decreases tax revenues, even when sales are decreased, perhaps deserves more emphasis than it usually gets, especially because it is one of the few strategies for which there is clear, direct evidence of effect. A second deficiency of emphasis is that there may have been insufficient stress on the long-term advantages of getting quantitatively informative mater- ial across about the total risks from tobacco, and the extent to which, at least in developed countries, these exceed all other reliably known causes of death. It is admittedly difficult to communicate risks in a way that will be understood and remembered approximately correctly, especially by people who have no framework of other risks with which to compare them. However, this should be possible, so long as the main message is set clearly apart from the lesser mes- sages that qualify it. After all, the chief message is merely that 'ABOUT A QUARTER OF ALL REGULAR CIGARETTE SMOKERS WILL BE KILLED BEFORE THEIR TIME BY THE HABIT', which is considerably less complicated than the mass of quantitative information about house prices, groceries, car prices, etc. that already has become part of the folklore of consumer societies. How exactly this main message should be put over is a matter for experiment: comparisons with other conditions may (especially in Britain) be helpful, for example: SMOKING IS BRITAIN'S BIGGEST KILLER Among 1000 young adults who smoke cigarettes regularly, --about 1 will be murdered --about 6 will be killed on the roads --about 250 will be killed by tobacco. Even in the United States, where road accident death rates are more than double those in Britain and murder rates are about ten times those in Britain, some such comparisons may be helpful (although it may then be advisable to start with only 100 young United States adults, and to threaten about 1, 2 and 25 of them with death). Whatever format is preferred, however, the central point remains: the

I I|111111111111 II II II III IIIIIIIII III 136 PETO reason one wants to prevent smoking is not just because it is danger- ous--dozens of things are dangerous--but because it is so dangerous. This indicates getting some sort of quantitative information over, both about the effects of smoking itself on mortality and, perhaps at least as importantly, about how much smaller all reliably known other carcinogenic effects are. Such in- formation may in the short term make only a few people give up, but over a few years wide acceptance of such a perspective may have substantial effects, either on individual behaviour or on making other actions political!y acceptable. CONCLUSION Large changes in cigarette usage can be produced by socially acceptable means (Table 2). In Britain, for example, cigarette usage per adult has decreased by about 30% over the past decade alone, Likewise, tar level reductions can fairly easily be implemented, especially in countries such as Russia and China where cigarettes are manufactured and distributed by the State with little advertising and where typical tar deliveries exceed the upper limit of what is currently sold in Britain. TABLE 2 Information for governments on simple measures for the control of lung cancera Price increases will produce fewer deaths and more revenue (as long as the)' do not feed back into wage demands) Tar reductions should be encouraged (especially in countries such as Russia and China where typical tar levels are still of the order of 20-30 mg, which is extremely high) Advertising could be taxed, restricted, prohibited, or limited to cigarettes delivering under I0 mg tar Simple, clear, quantitative information could be communicated effectively ~o the general population: ABOUT A QUARTER OF ALL REGULAR SMOKERS ARE KILLED BEFORE THEIR TIME BY TOBACCO I General note: recommendation of these few simple measures (which might have a substantial effect in just a few years on exposure) does not, of course, detract from the need for a wide range of other measures, including many of those suggested by WHO (1979), UICC (Gray 1977), and others. Modificztions of the consumption and of the composition of cigarettes are complementary, not competing, strategies that together could lead to avoidance of most of the 100 000 or more tobacco-induced deaths each year in Britain, most of the 300000 or more in the US, and some hundreds of thousands in the rest of the world. Such changes will, perhaps surprisingly, TOBACCO-REL not greatly cha developed cou~ in middle age. REFERENCE~ Cederl6f R. Doll assessment meth Doll R. Peto R It. doctors. Br Med Doll R. Peto R 19~ United States to,_ Gray N (ed) 1977 1. Contre le Cancer Independent Scien Stationery Offic~ Kaufman DW. He monoxide conter 308:409-413 Lee PN 1976 Stat~ Research Counc~ Lee PN, Garfinket 35:16-22 Peto R. Speizer FE Richards SM. Gi but not of mucus 128:491-500 Wald N, Idle M, B cigarette. Thorax Winn DM, Blot cancer among w~ World Health Org~ committee on sm DISCUSSION Doll: I don't ' disease in Brita diagnostic habit 20 or 30 years a~ term. The diff, differences, bec Koplan: You obstructive lung

PETO ~cause it is danger- ~ so dangerous. This ~ver, both about the as importantly, about effects are. Such in- ve up, but over a few tantial effects, either =ally acceptable. tlly acceptable means lult has decreased by reductions can fairly sin and China where ,'ith little advertising .'hat is currently sold I of lung cancer~ ~ot feed back into ~ussia and China where s delivering under 10 mg ectively to the general LED BEFORE THEIR night have a substantial need for a wide range of .~ICC (Gray 1977), and 9n of cigarettes are ~er could lead to deaths each year in some hundreds of rhaps surprisingly, TOBACCO-RELATED DISEASE 137 not greatly change the life expectancy of old people, but they will, at least in developed countries, appreciably decrease the proportion of people who die in middle age. REFERENCES Cederlrf R. Doll R. Fowler B, Friberg L, Nelson N, Vouk V 1978 Air pollution and cancer: risk assessment methodology and epidemiological evidence. Environ Health Perspect 22:1-12 Doll R. Peto R 1976 Mortality in relation to smoking: 20 years' observations on male British doctors. Br Med J 2:1525-1536 Doll R, Peto R 1981 The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J Natl Cancer Inst 66:1191-1309 Gray N (ed) 1977 Lung cancer prevention: guidelines for smoking control. Union Internationale Contre le Cancer, Geneva Independent Scientific Committee on Smoking & Health 1983 Third Report. Her Majesty's Stationer3' Office. London Kaufman DW. Helmrich SP, Rosenberg L, Miettinen OS, Shapiro S 1983 Nicotine and carbon monoxide content of smoke and the risk of myocardial infarction in young men. N Engl J Meal 308:409-413 Lee PN 1976 Statistics of smoking in the United Kingdom. Research Paper No. 1. Tobacco Research Council, London Lee PN, Garfinkel L 1981 Mortality and type of cigarette smoked. J Epidemiol Community Health 35:16-22 ~ Peto R, Speizer FE, Cochrane AL, Moore F, Fletcher CM, Tinker CM, Higgins ITr, Gray RG, Richards SM, Gilliland J, Norman-Smith B 1983 The relevance in adults of airflow obstruction. but not of mucus hypersecretion, to mortality from chronic tung disease. Am Rev Respir Dis 128:491-500 Wald N, Idle M, Boreham J, Bailey A 1980 Inhaling habits among smokers of different types of cigarette. Thorax 35:925-928 Winn DM, Blot WJ. Shy CM. Pickle LM, Toledo A, Fraumeni J 1981 Snuff-dipping and oral cancer among women in the Southern United States. N Engt J Med 304:745-749 World Health Organization 1979 Controlling the smoking epidemic: report of the WHO expert committee on smoking control. Technical Report Series 636. WHO, Geneva DISCUSSION Doll: I don't think one can really compare trends in chronic obstructive lung disease in Britain and the US. There were formerly enormous differences in diagnostic habits between the two countries, which have diminished in the last 20 or 30 years as a result of physicians getting together and deciding to use this term. The differences in trends can be largely attributable to nosological differences, because the national statistics have started from a different base. Koplan: You say that there are other aetiological factors involved in chronic obstructive lung disease (COLD), in addition to smoking. In lung cancer itself,

138 DISCUSSION you suggest that one means of comparing low tar cigarette smokers to high tar sm6kers is to compare data over time. Tar content clearly is the major compo- nent in lung cancer, but are there likely to be other aetiological agents for lung cancer as well, in parallel to COLD? Peto: There certainly are other host factors in lung cancer. There are obvious ones, like asbestos and ionizing radiation. It is also possible that there are nutritional components in lung cancer, or that infective processes could be relevant. I don't think changes in air pollution can be plausibly invoked as causes of the changes over time. partly because air pollution was never an important cause of lung cancer anyway, and partly because one sees the same trends in Finland, a country that has never been seriously polluted (outside the sauna baths!). Blanpain: The pine woods in Finland produce substantial air pollution: substances such as turpentine are turned into carcinogenic smog under the influence of sunlight. Peto: Changes in air pollution from pine forests are not responsible for the 40% decrease seen in Finnish lung cancer rates between the early 1960s and the late 1970s. The number of cigarettes per man smoked in Finland, like that in Britain, stayed roughly constant until the mid 1970s and only then began to fall, with large price increases. The tar deliveries fell enormously, because there was a switch from Russian-type cigarettes to modern Western low tar cigarettes. In fact, the fall in tar delivery in Finland was greater than anywhere else in the world, and we saw a 40% decrease in lung cancer incidence in early middle age by the late 1970s. By now, the decrease may be beyond 50%, as it is in Britain. Miteva-Toncheva: We are confronted here with a very complex problem, namely the interaction between environmental factors and the individual's genotype (for example, the effects of tobacco on potentially susceptible indi- viduals). Have you investigated al-antitrypsin variants by isoelectrofocusing? Homozygotes for deficiency of this protease inhibitor, and heterozygous car- riers, are highly susceptible to chronic obstructive lung disease (COLD) under the influence of tobacco. Such individuals form a high risk group, and it is very important to give more care to them. Peto: For homozygotes, that is true. People deficient in this inhibitor finish up with proteases coming out of their alveoli, so wrecking their lungs complete- ly, and may die of COLD at the age of 40. Homozygotes with al-antitrypsin deficiency are rare and are at enormous risk of this lung disease, if they smoke. Heterozygotes are an appreciable proportion of the population and some studies have suggested that they are at increased risk but, overall, there seems to be no substantial difference in risk between heterozygotes and people of normal genotype. But there must be genetic deter)ninants of the response to cigarette smoke of the heart, lungs, and other systems. However, we can change the environment, but we can't change the genotype. TOBA Met smoki peopl~ ischae decre: Pet, could variot, On smoki and ht switch cigare moutl- don't becau their i~ that ci manta in the chang~ Hje., (CHD becau, the 0, sclero, reasor Pet~ CHD so extz from .~ fivefol high a Sec~ extren diseas, causal Dan smokil only c, (Fribe" young, 0

DISCUSSION okers to high tar ~e major compo- agents for lung here are obvious e that there are ,cesses could be ~ibly invoked as n was never an ~e sees the same ~ted (outside the air pollution: ~mog under the responsible for early 1960s and and, like that in _~n began to fall, cause there was ir~rettes. In ~glse in the arI~middle age it is in Britain. ~plex problem, he individual's ~sceptible indi- ectrofocusing? erozygous car- COLD) under ~, and it is very nhibitor finish tngs complete- al-antitrypsin if they smoke. on and some 1, there seems md people of e response to ever, we can TOBACCO-RELATED DISEASE 139 Meade: What is the current status of the evidence on cigars in relation to smoking-related disease, cardiovascular or malignant, now that so many more people are smoking them? You also mentioned the high relative risk of ischaemic heart disease in younger cigarette smokers, and the fact that it decreases with increasing age. Have you any ideas why that is? Peto: I don't know why the relative risk changeswith age like that. You could produce models in which you eliminate susceptible sub-groups in various ways, but that is a mathematical exercise, not a biological one. On the question of cigars, one has to distinguish between two types of cigar smoking. The first is in people who smoked cigars when they began smoking and have smoked them throughout their lives. There are also people who have switched from cigarettes to cigars because they want to avoid the health risks of cigarettes. Long-term cigar smoking confers a certain risk of cancer of the mouth and throat, but nothing like the lung cancer risk of cigarettes. But we don't know what will happen when people switch from cigarettes to cigars, because this could produce quite different effects. These people may carry over their inhaling patterns to cigars. There is not much evidence, in animals at least, that cigar smoke is less hazardous than cigarette smoke. It is perhaps chiefly the manner in which the cigar is smoked that is protective, rather than differences in the make-up of the smoke, except insofar as these differences produce changes in the manner of smoking. Hjermann: I believe that smoking is causally linked to coronary heart disease (CHD), but I am not quite sure how firmly I should believe this! It is partly because of the lack of controlled trials in this field. In a prospective study like the Oslo trial, we could not find any significant correlation between athero- sclerosis and smoking. It seems to be hard to show. What are your main reasons for believing in this causal relationship? Peto: The fundamental reason for believing that the relationship between CHD and smoking is a causal one is that the relative risk in early middle age is so extreme. Among people in their thirties, CHD is rare but the relative risk from smoking is as much as 10-fold. In the 40-44 age group there is still a fivefold relative risk. So the relative risk between the ages of 30 and 50 is so high as to make non-causal explanations rather implausible. Second, peripheral vascular disease and aortic aneurysm have even more extreme relative risks among smol~ers than CHD does. For peripheral vascular disease the association is so extreme as v~,rtually to preclude anything except a causal explanation. Danielsson: There is still some question about the causal correlation between smoking and ischaemic heart disease, in my view. A twin study showed that the only certain correlation between smoking and disease was for lung cancer (Friberg et al 1973). The problem with evidence for causality dra~vn from the younger smokers is that you don't have a good control group. We should always

-- - IlII lilllllll ..... I III , ............. ,, ,,,, , ........... 140 DISCUSSION TOBACCO-F bear in mind that perhaps a smoker is different from a non-smoker; that is, smokers may be more prone to this disease, whether or not they actually smoke. This is why it is so difficult to do a conclusive study. Peto: This is not a plausible hypothesis where the relative risks are so extreme. The twin studies are much overrated, Like randomized trials, they would be fine if done on a scale one or two orders of magnitude bigger than they • .are done. When the evidence from homozygous and heterozygous twins is studied, and one looks at the actual numbers of events and how much differ- ence there was in smoking, there is not enough evidence to add usefully to that from much larger studies. Hiatt: Is any effect on cancers other than those in the respiratory tract visible yet, with the changes in cigarette smoking? And what are your present thoughts about interactions between cigarettes and other carcinogens? Peto: We can't really assess the effects of changes in cigarette composition on say, cancer of the bladder, although Paolo Vineis has suggested, on the basis of his large but unpublished case-control study, that black tobacco (like that in Gauloises, for example) is peculiarly hazardous. One would probably learn more by measuring the extent to which the urine of smokers is mutagenic when they are smoking different types of cigarette, taking the mutagenicity of the urine as a surrogate for epidemiology. Some types of cigarettes deposit sub- stantial amounts of benzidine into the bladder. The chief source of benzidine that people are now exposed to in Britain is from cigarettes, rather than the industrial use of dyes and other chemicals. Doll: Mortality from cancer of the bladder and of the pancreas---diseases that are not very strongly related to tobacco use--is either falling or static in the UK. However, the incidence rates of some cancers that are closely related to tobacco--cancer of the mouth, oesophagus and larynx--are going up. These last changes presumably result from the increase in alcohol consumption, which is even more strongly related to these cancers. These changes illustrate the complexity of drawing conclusions about the effect of specific aetiological agents from mass statistics. If I wanted to argue that cigarette smoking was not proved to be the cause of any type of cancer, I would point to the trends in cancer of the larynx in various countries. Although this cancer has been closely related to cigarette smoking in all case-control studies, the trends in incidence and mortality have sometimes been in the opposite direction to the trends for cancer of the lung. To my mind, the only reasonable explanation is that other factors of which we are still ignorant account for these exceptions. But if you want to allow an exception to disprove a massive set of evidence, you point to the trends in cancer of the larynx. These cases are important to investigate because they are exceptions, but we ought to beware of allowing such exceptions to undermine a mass of contrary evidence, when we ar and the pro. Shephara naive to thi into sugges other risks will make i With reg obtained bx In fact, wit cigarettes, Peto: Th airways. I a the alveo/i reductions, young peoI continue t~ tendency tt levels--the levels were when Japm took up th~ that high t~ Koplan: cigarette s~ non-tobacc Peto: I d, between cc The comm public pres really seri~ European efforts; it c health, or problem s~ where the serious, an reduction, years, whi, Eddy: V~ of tobacco

DISCUSSION • moker; that is, ~t they actually ve risks are so zed trials, they qgger than they zygous twins is ,w much differ- usefully to that ory tract visible ,resent thoughts composition on • on the basis of co (like that in probably learn ~utagenic when genicity of the es deposit sub- ze of benzidine ,, or static in the ~sely related to ~ing up. These consumption, ions about the anted to argue pe of cancer, I ries. Although 11 case-control .~s been in the mind, the only • still ignorant on to disprove • larynx. These ,ut we ought to rary evidence, TOBACCO-RELATED DISEASE 141 when we are trying to interpret the relationship between aetiological agents and the production of disease. Shephard: You discussed the tactic of emphasizing tar reduction. It may be naive to think that cigarette manufacturers are not going to pervert this view into suggestions that we believe low tar cigarettes to be safe. There may be other risks with this tactic, one being that the lower levels of tar and nicotine will make it easier for young people to become addicted to cigarettes. With regard to the figures for the tar content of cigarettes, these are all obtained by smoking machines which presuppose a certain pattern of smoking. In fact, with the change in pattern of smoking that one gets with low tar cigarettes, there may be very little reduction in the tar delivered by them. Peto: There may be a substantial difference in tar delivery to the main airways. I agree that there may be no great change in the amount that reaches the alveoli. However, in countries where there have been good tar level reductions, there have also been considerable reductions in the extent to which young people are taking up smoking and also in the extent to which adults continue to smoke. That is not always true. but at least there is no systematic tendency the other way. Conversely, consider countries with very high tar levels--the USSR at present, or Japan in the 1950s. At that time in Japan, tar levels were high. Smoking by minors was forbidden during those years, but when Japanese males reached adulthood, about 80% of the male population took up the smoking of high tar cigarettes. So there is no suggestion, either, that high tar levels discourage people from starting to smoke. Koplan: Are there any differences in the level or type of effort to curtail cigarette smoking among countries that are tobacco producers, as against non-tobacco producers? Peto: I don't really know. There does, however, appear to be a big difference between countries where tobacco manufacture is in private and in public hands. The communist countries have been terrible; I suppose there is no effective public pressure on those governments to act, and so they have done nothing really serious to discourage smoking in the USSR, China or any Eastern European country. In at least some Western countries there have been serious efforts; it depends very much on concerned individuals within departments of health, or within the population generally. One or two people who take the problem seriously can make a vast difference to a whole country. In France, where the State manufactures tobacco, the government has done nothing serious, and in Britain it has. Perhaps partly because of this, Britain has seen a reduction of one-third in the number of cigarettes smoked over the last i0 years, which is a considerable success that should be emulated elsewhere. Eddy: What role do you think nicotine chewing gum might play in the control of tobacco-related diseases?

142 DISCUSSION Peto: A relatively minor one. Various randomized trials have suggested some effect, though not a large one. But, because the problem is so important, a minor effect on a major disease represents a useful public health advance, so it is a worthwhile aid that should be introdu_ced, encouraged and used. Any- thing that helps to get people off cigarettes is a good idea. We have to find out what works for different people, and nicotine chewing gum helps a few people. REFERENCE Friberg L, Cederl6f R, Lodch U et al 1973 Mortality in twins in relation to smoking habits and alcohol problems. Arch Environ Health 27:294-304 Screenir related Malm6 " BO PETERSSOt' Section of Preve: S-214 O1 MalmO, Abstract. related di: in a large underlyin! In a pr~ death in n most pred cant corre tively, ant In the r admissior group, af" prcventir related di as the ret high/non 1985 The p 143-16 The preventi broad progrz as individual complete eli~ Three mai phasized, na subject, and The purp~ individual st.

2063628076

WORLD HEALTH ORGANIZATION INTERNATIONAL AGENCY FOR RESEARCH ON CANCER TOBACCO: A MAJOR INTERNATIONAL HEALTH HAZARD Proceedings of an International Meeting organized by the IARC and co-sponsored by the All-Union Cancer Research Centre of the Academy of Medical Sciences of the USSR, Moscow, USSR held in Moscow, 4-6 June 1985 EDITORS c-' ~ D.G. ZARIDZE R. PETO H S,\ LT]-i °~"-":'- ~,~,~_c~,cS LII3~ARY IARC Scientific Publications No. 74 INTERNATIONAL AGENCY FOR RESEARCH ON CANCER LYON 1986

CHEMICAL CONSTITUENTS AND BIOACTIVITY OF TOBACCO SMOKE D. HOFFMANN & E.L. WYNDER Naylor Dana Institute for Disease Prevention, American Health Foundation, Valhalla, NY 10595, USA INTRODUCTION The occurrence of cancer of the respiratory tract and of the upper digestive tract is causally related to smoking of cigarettes, cigars, pipes and bidis, while malignant tumours of the bladder, renal pelvis and pancreas are causally related to smoking of cigarettes (IARC, 1986). Epidemiological studies have demonstrated an association of tobacco chewing with cancer of the oral cavity (IARC, 1985). These conclusions have been supported by a large number of bioassays. The application of tobacco extracts, the inhalation of tobacco smoke and the application of tobacco smoke condensate induce cancer in laboratory animals (Wynder & Hoffmann, 1967; Hoffmann et al., 1983; IARC, 1985, 1986). It has been the joint task of chemists and biologists to identify those compo- nents in tobacco and tobacco smoke that contribute to their carcinogenic effects. However, it would be an insurmountable task to evaluate each of the more than 2500 constituents in tobacco leaf and more than 3900 compounds in tobacco smoke for possible tumorigenic effects (Table 1). Therefore, the research programme has to be limited to the identification of those tumorigenic and carcinogenic agents that can account for most of the carcinogenic activity of tobacco products. Despite this limitation, remarkable progress has been made by the laboratory scientists. This progress is well reflected in the reduced carcinogenic potential of 'low-tar' cigarettes (IARC, 1986). In evaluating the carcinogenic risk of environmental tobacco smoke exposure (passive smoking), the knowledge of the phys- icochemical nature of sidestream and mainstream smoke and the principles of chemical carcinogenesis were the primary data bases which led the IARC (1986) to conclude that 'passive smoking gives rise to some risk of cancer.' THE PHYSICOCHEMICAL NATURE OF TOBACCO SMOKE The combustion of tobacco products leads to the formation of mainstream smoke (MS) and sidestream smoke (SS). MS is generated during puff-drawing in the burning cone and - 145-

146 HOFFMANN & WYNDER Table 1. Estimates of constituents in tobacco smoke (~__3900 known compounds) Major classes of compounds~ NO Amides, imides, lactones 240 Carboxylic acids, anhydrides 240 Lactones 150 Esters 475 Aldehydes 110 Ketones 520 A~cohols 380 Phenols 285 Amines 200 N-Nitrosamines 22 N-Heterocyclics 920 Hydrocarbons 755 Nitriles 105 Carbohydrates 45 Ethers 310 Total 4865 ~ Some compounds contain multiple functional groups, thus this list exceeds 3900 hot zones of cigarettes and cigars; it travels through the tobacco column and out of the mouthpiece. SS is formed between puffing and is emitted from the smouldering coal into the ambient air. The data presented throughout this review are derived from machine-smoking under standardized laboratory conditions (Brunnemann et al., 1976a; International Committee for Cigar Smoke Study, 1974). However, it has to be realized that machine-smoking parameters can differ substantially from the puff-drawing parameters of smokers, espe- cially in the case of cigarettes with low nicotine delivery (Herning et aI., 1981). About 30% of the total effluents of MS originate from the tobacco, the remainder comes from the air drawn through the cigarette. When leaving the mouthpiece, undiluted smoke from a nonfilter cigarette contains about 5 × 109 particles per millilitre, with a median particle size of about 0.4 pm (Keith & Tesh, 1965; Carter & Hasegawa, 1975). The pH of tobacco smoke is of major significance, since it influences the degree of protonation and, therefore, the proportion of nicotine and other basic components in the vapour phase. This determines the inhalability of MS (Armitage & Turner, 1970). At about pH 5.4, all nicotine in tobacco smoke is monoprotonated and resides in the particu- late phase (Fig. l). The pH of the MS of air-cured tobaccos and of cigars increases with ascending number of puffs. Consequently, the smoke of these products contains propor- tionately larger amounts of nicotine in the vapour phase. The smoke pH of cigarettes filled with flue-cured tobaccos or with tobacco blends, on the other hand, decreases slight!y or remains rather constant (Fig. 2; Brunnemann & Hoffmann, 1974). The total MS of a cigarette weighs about 400-500 rag. More than 92% of the total is made up of 400-500 individual gaseous components with nitrogen (-~58%), oxygen (--~12%), o o3 o

CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE 147 Fig. 1. Degree of protonation of nicotine in relation to pH (pH = pKa log 1 - c~/a (Henderson-Hasselbach)) pH Fig. 2. pH of total mainstream smoke of various tobacco products: (1) little cigar I; (2) little cigar II; (3) cigar; (4) Kentucky reference cigarette (84 mm); (6) blended cigarette without filter (84 ram) 8.6 LO No. of puffs 0 O~ O~ 0 O~ 0

148 HOFFMANN & WYNDER Fig. 3. Approximate chemical composition of mainstream smoke (from Norman, 1977) N2 58 STRONG ACIDS :~7.7 ACIDS 15.3 WATER NEUTRALS 162 carbon dioxide (-=13%) and carbon monoxide (-=3.5%) as major constituents. The remainder is comprised of other vapour phase components and of compounds constituting the particulate phase (Fig. 3; Norman, 1977). VAPOUR PHASE Bioassays with total smoke have indicated that the majority of the genotoxic and cocarcinogenic agents reside in the particulate phase (Dontenwill et al., 1973; Hoffmann et al., 1979). Thus, specific methods have been developed for the quantitative determination of smoke particulates. The most widely applied technique is the Cambridge filter method, utilizing a glass fibre filter pad which retains 99.7% of all particles with diameters of >0.1 pm (Dube & Green, 1982). This manner of trapping does not effect a strict separation of the solid and gaseous components in the physicochemical sense, nevertheless, it permits reproducible, quantitative determination of the particulate matter in the smoke of cigaret- tes, cigars and pipes and analysis of tile major vapour phase components by gas chromatog- raphy. In addition to nitrogen, oxygen, carbon dioxide and carbon monoxide, the vapour phase contains hydrogen, methane and other hydrocarbons, volatile aldehydes and ketones, nitrogen oxides, hydrogen cyanide and volatile nitriles and at least an additional 400-450 minor constituents (Keith & Tesh, 1965; Wynder & Hoffmann, 1967; Brun- nemann & Hoffmann, 1982).

CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE Table 2. Major toxic and tumorigenic agents in the vapour of freshly generated smoke of a nonfilter ciga, rette ~ Agent Ccnc./cigarette BioL effect Carbon monoxide 10-23 mg T Acetaldehyde 0.5-1.2 mg CT Nitrogen oxides (NO,) 50-600 ~g T Hydrogen cyanide 150-300/~g CT, T Ammonia 50-170 ,~g T Acrolein 50-100/xg CT Benzene 20-50/~ HC Formaldehyde 5-100/~g C 2-Nitropropane 0.2-2.2/,~g C Hydrazine24--43 ng C U rethane20-38 ng C Vinyl chloride 1.3-16 ng HC ~Ooea not include volatile N-nitrosam~nes ~ Abbreviat{or~s: T. toxic agent; CT. ci([atoxic agent; HC. human carcinogen; C. carcinogen Fig. 4. Hamburg II smoke inhalation device for Syrian golden hamsters 149 Fhe ing tnd 1 et on )d, L1 of its et- ~g- Llr ~al I1- 0 O~ O~ 0

150 HOFFMANN & WYNDER Table 2 presents a listing of the major known toxic and tumorigenic agents in the vapour phase of cigarette smoke. Each of the volatile smoke constituents was quantitatively assessed by analytical methods that had to be specifically developed for their determina- tion in the smoke of cigarettes or cigars. Despite the presence of volatile carcinogens in the vapour phase of tobacco smoke, currently available bioassays- and here mainly inhalation experiments with hamsters (Fig. 4) - have not been sensitive enough to induce tumours by administering the vapour phase as such, aside from the induction of lung adenoma in mice (Mohr & Reznik, 1978). PARTICULATE PHASE While the vapour phase by itself is not tumorigenic in most of the inhalation assays, and the total smoke induces benign and malignant tumours in the upper respiratory tract of rats and hamsters (Dontenwill et al., 1973; Hoffmann et al., 1979), evidence from contact carcinogenesis studies indicates that the particulate phase contains most of the known tumorigenic and carcinogenic agents of tobacco smoke. Tobacco smoke particulates Fig. 5. Fractionation of cigarette 'tar'; C, relative carcinogenic activity; P, relative tumour-promoting activity r o o", r.0 o

r e v e CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE 151 Fig. 6. Tumour-initiating activities of 80 end-fracti0ns from BI subfractions. Each end-fraction was tested on 20 mice, negative control, no initiator: 1% croton oil as promoter; line a, fractions with activities significantly above those in negative control group (p <0.05); line b, fractions with strong tumour-initiating activity (p <0.05 above those in line a) 2O 0 20 15 0, 0 80 10 20 30 &O 50 60 70 Subfractions BIh, 1-BIh, 80 ('tars') have consistently and in a dose-related response, induced benign and malignant tumours in the skin of mice and rabbits, and in the connective tissue and bronchial epithelium of rats (Wynder & Hoffmann, 1967; Mohr & Reznik, 1978; Hoffmann et al., 1979; IARC, 1986). Tumour initiators and cocarcinogens The findings from bioassays with tobacco 'tars' have led to more detailed and systematic testing on mouse skin of the various fractions and subfractions of the particulate phase (Fig. 5; Hoffmann & Wynder, 1971). The only fractions found to have significant activity as complete carcinogens were the neutral fraction and its subfractions B and BI. A further breakdown of subfraction BI, which amounted to 0.6% of the total particulate phase, led

152 HOFFMANN & WYNDER Table 3. Major compounds identified in neutral subfractions BIh 56-66 of the particulate phase of cigarette smoke Chlorinated hydrocarbon insecticides DDD o,p'-DDD DDT o,p'-DDT DDM (DDD-HCI) DDE (DDT-HCI) Trans-4 , 4 '-Dichlorostilbene N-Alkylcarbazoles 9- Methylcarbazole 9-Ethylcarbazole 1,9-, 2,9-, 3,9- and 4,9-Dimethylcarbazole Fluoranthenes Fluoranthene 1 -, 2-, 3-, 7- and 8-Methylfluoranthene X-Ethylfluoranthene(s) x,x'-Dimethylfluoranthenes Benzo[mno]fluoranthene Benzofluorenes 11 H-benzo[a]fluorene 11H-benzo [b]fluore ne 7H-benzo[c]fluorene 17 H-cyclopenta[a] phenanthrene 17H-cyclopenta[a]phenanthrene x-Methyl-17 H-cyclopenta[a]phenanthrene x-Ethyl-17 H-cyclopenta[ a]phenanthrene x-Phenylindene Pyrenes Pyrene 1 -, 3- and 4-Methylpyrene x,x'-Dimethylpyrene(s) Table 4. Major compounds identified in neutral subfractions BIh 71-78 of the particulate phase of cigarette smoke Chrysenes Chrysene 1 -, 2-, 3-, 4-, 5- and 6- Methylchrysene x,x'-Dimethylchrysene(s) x-Ethylchrysene(s) Benz[a]anthracenes Benz[a]anthracene x-Methylbenz[a]anthracene Benzo[c]phenanthrenes Benzo[c]phenanthrene x- Methylbenzo[c]phenanthrene Benzopyrenes Benzo[a]pyrene x- Methylbenzo[a]pyrenes Benzo[e]pyrene Benzofluoranthenes Benzo[b]fluoranthene 13enzo[j]fluoranthene Benzo[k]fluoranthene Ideno[1,2, 3,-cd]pyrene Dibenzopyrenes Dibenzo[a, h]pyrene (?) Anthanthrene Perylene Benzo[ghi]perylene to a highly carcinogenic concentrate, BIh (representing 0.09% of the 'tar') and this, in turn, was chromatographically separated to yield 80 end fractions. Upon testing as tumour initiators on mouse skin (Fig. 6), end fractions BIh 56-66 and BIh 71-78 were found to be highly active. Their chemical analysis revealed that they consisted primarily of polynuclear aromatic hydrocarbons, many of which are known carcinogens in laboratory animals (Tables 3 and 4; Hoffmann & Wynder, 1971). Application to mouse skin of these highly active end fractions in doses proportionate to their occurrence in the total particulate

CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE Table 5. Carcinogens and cocarcinogens in the smoke of a nonfiltercigarette Relative carcinogenic ng/cigarette Agent activity Carcinogens Benzo[a]pyrene + + + 10-50 5-Methylchrysene + + + 0.6 Dibenz[ a,h ]anthracene + + 40 Benzo[b]fluoranthene + + 30 Benzo[j]fluoranthene ++ 60 Dibenzo[a,i]pyrene ++ present Indeno[1,2, 3- cd]pyrene + 4 Benz[a]anthracene + 40-70 Chrysene + 40-60 Benzo[e]pyrene ? 5-40 Dibenz[a,j]acridine ++ 3-10 Dibenz[a,h]acridine + 0.1 Dibenzo[c,g]carbazole + 0.7 Cocarcinogens Pyrene 50-200 Fluoranthene 100-260 Benzo[g h i]perylene 60 4,4'-Dichlorostilbene 1 500 Catechol 25 000-360 000 3-Methylcatechol 11 000-20 000 4-Methylcatechol 15 000-21 000 4-Ethylcatechol 10 000-24 000 153 matter did not lead to tumour induction. Yet, co-application of the active neutral subfrac- tions with the inactive phenolic fraction of the particulate matter led to a tumour yield which accounted for approximately 65-?5% of that induced with the total 'tar'. This indicated that the phenolic fraction had cocarcinogenic activity, and further studies showed that catechols were the major cocarcinogens in the phenolic portion (Hecht et al., 1981). Catechol itself is the most abundant phenol in tobacco smoke, amounting to 26-360 /tg per cigarette (Wynder & Hoffmann, 1967; Brunnemann et al., 1976b). Table 5 lists the major epithelial carcinogens and cocarcinogens identified in the smoke of a non-filter cigarette. Organ-specific carcinogens Tobacco smoke contains, in addition to contact carcinogens and eocareinogens, several organ-specific carcinogens. This supports the epidemiologieal observation that cigarette smoking is an important factor in the etiology of cancer of the oesophagus, pancreas, renal pelvis and urinary bladder (IARC, 1986). Table 6 lists the known organ-specific carcino- gens in cigarette smoke. Polonium-210 (0.03-i.0 pCi/cigarette) has been incriminated as a possible contributing factor for the increased risk for cancer of the lung in cigarette smokers (Radford & Hunt, 1964; Harley et aL, 1980). The presence of aromatic amines in

154 HOFFMANN & WYNDER Table 6, Organ-specific carcinogens in cigarette smoke Carcinogen ng/cigarette N-Nitrosodimethylamine 1-180 N-Nitrosoethylmethylamine 1-40 N- Nitrosodiethylamine 0.1-28 N-Nitrosopyrrolidine 2-110 N-Nitrosopiperidine 0-9 N-Nitrosodiethanolamine 0-40 N'- Nitrosonornicotine 120-3700 4-(Methylnitrosamino)-I -(3-pyridyl)-I -butanone (NNK) 120-950 N'- Nitrosoanabasine 40-400 2-Toluidine 30-160 2-Naphthylamine 4.3-27 4-Aminobiphenyl 2.4-4.6 Nickel 20-3000 Polonium-210 0.03-1.0 pCi Table 7. Estimated exposure of US residents to nitrosaminesa Sou rce of Nitrosamines b Primary exposure Daily intake exposure route ~g/person) Beer Cosmetics Cured meat; cooked bacon Scotch whisky Cigarette smoking NDMA Ingestion 0.34 NDELA Dermal absorption 0.41 NPYR Ingestion 0.17 NDMA Ingestion 0.03 VNAc Inhalation 0.3 NDELA Inhalation 0.5 NNN Inhalation 6.1 ) NNK Inhalation 2.9 NAT+ NAB Inhalation 7.2 16.2d aFrom National Research Council (1981) bNDMA, N-nitrosodimethylamine; NDELA, N-nitrosodiethanolamine; NPYR, N-nitrosopyrrolidine; NNK, 4-(methylnitrosamino)-l-(3-pyridyl)-l-butanone; NAT, N'- nitrosoanatabine; NAB, N'- nitrosoanabasine ~VNA, volatile nitrosamines (NDMA+ N-nitrosomethylethyiamine+ N-nitrosodiethylamine+ NPYR) ~Tobacco-apecific nitrosamines smoke has been associated with the increased risk for bladder cancer in cigarette smokers (Doll, 1972). The N-nitrosamines are the major group of organ-specific carcinogens in tobacco products. They are formed during the processing of tobacco and during smoking by N-nitrosation of secondary and tertiary amines. Tobacco smoke contains volatile, non- volatile and tobacco-specific N-nitrosamines (TSNA; Table 6). It has been estimated that US residents receive the highest degree of exposure to nitrosamines from cigarette smok- ing (Table 7). In fact, the concentration of these compounds in tobacco smoke exceeds by oa o o~

CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE 155 Fig. 7. Tobacco alkaloids and nitrosamines which can be formed from them. With the exception of NNA, all of these compounds are present in tobacco and tobacco smoke - NICOTINE ~,~NORNICOTINE ANABASINE ANATABINE ~ N " CH3 4-(METHYLNITROSAMINO)-I" 4-(METHYLNITROSAMINO)'4" N....~'NITROSONORNICOTINE N_~"NITROSOANABASINE N_.~LN~TROSOANATABINE { 5- PYRIDYL)-I- BUTANONE (~,-PYRIDYL) BUTANAL (NNN) (NAB) (NAT) (NNK} (NNA} Table 8. N-Nitrosamines in cigarette smoke from different varieties of tobacco (ng/cigarette) a Burley Bright French black N-Nitrosamine tobacco tobacco tobacco N-Nitrosodimethylamine 11-180 N-Nitrosomethyiamine 9.1-13 N-Nitrosodiethylamine 4-25 N-Nitrosopyrrolidine 52-76 N'- Nitrosonornicotine 3700 4-(Methylnitrosamino)- 1 -(3-pyridyl)-I -butanone 320 N'- Nitrosoanatabine 4600 N'- Nitrosoanabasine 400 0.5-13.2 29-143 >0.1 2.7-12 nd-1.8 b 0.6-6 6.2 25-11 620 590 420 220 410 200 40 nd-150 From Hoffmann et al., 1984a rid, not detected at least two orders of magnitude the levels of nitrosamines reported in any other consumer product or respiratory environment, except for a few, very limited occupational settings (National Research Council, 1981). The most abundant nitrosamines in tobacco smoke are the TSNA. They are formed from nicotine and the minor tobacco alkaloids (Fig. 7). In the smoke, 25-45% of the TSNA originate by transfer from the tobacco, the remainder is formed by pyrosynthesis during smoking (Adams et al., 1983; Hoffmann & Hecht, 1985). The single most important factor for the smoke yields of nitrosamines is the nitrate content of tobacco (Adams et al., 1984), thus the smoke of air-cured tobacco is significantly richer in the nitrosamines (Table 8; Hoffmann et al., 1984a). Utilization of cigarette blends with stems and ribs, which are the portions of the tobacco leaf with the greatest abundance of nitrate, can substantially elevate the nitrosamine content of the smoke (Brunnemann et al., 1983). The nicotine-derived N-nitrosamines, N'-nitrosonornicotine (NNN) and 4-(methylni- trosamino)-l-(3-pyridyl)-l-butanone (NNK), are by far the most powerful carcinogens in 0 o3 I'~ 03 0 03

156 HOFFMANN & WYNDER Table 9. Carcinogenicityoftobacco-specific nitrosamines~ Route of Principal target Dose Nitrosamine~ Species and strain application organs NNN A/J mouse i.p. Lung F344 rat s.c. Nasal cavity, oesophagus oral Oesophagus, nasal cavity Sprague-Dawley rat oral Nasal cavity Syrian golden hamster s.c. Trachea, nasal cavity NNK A/J mouse i.p. Lung 0.12 mmol/mouse F344 rat s.c. Nasal cavity, 0.2-2.8 mmol/rat lung, liver Syrian golden hamster s.c. Trachea, lung, 0.9 mmol/hamster nasal cavity 0.005 mmol/hamster NAT F344 rat s.c. None 0.2-2.8 mmol/rat NAB F344 rat oral Oesophagus 3-12 mmoi/rat Syrian golden hamster s.c. None 2 mmol/hamster NNA A/J mouse i.p. None 0.12 mmol/mouse ~ From Hoffmann and Wynder, 1985 ~NNN, N'-nitrosonornicotine; NNK, 4-(methylnitrosamino)-l-(3-pyridyl)-l-butanone; NAT, N'-nitrosoanatabine; NAB, N'-nitrosoanabasine; NNA, 4-(methylnitrosamino)-4-(3-pyridyl)butanal 0.1 2 mmol/mouse 0.2-3.4 mmol/rat 1.0-3.6 mmol/rat 8.8 mmol/rat 0.9-2.1 mmol/hamster Fig. 8. Scheme linking nicotine, the major tobacco alkaloid and habituating factor in tobacco, to formation of the promutagenic DNA adduct O6-methylguanine TOBACCO P__.OCESS % .,..=O CIGAO:ETTE I~'~.,,I CH~5 NICOTINE SMOKING NNK METABOLIC ACTIVATION ECH3N=NOH3 ~ 7-METHYLGUANINE METHYLDIAZO- 06"METHYLGUANINE HYDROXIDE IN DNA tobacco smoke, inducing carcinoma in mice, rats and Syrian golden hamsters (Table 9). Perhaps the most important observation is that NNK induces benign and malignant tumours in laboratory animals not only in the upper respiratory tract but also in the lung. In hamsters, a single application of 1 mg of NNK suffices to induce lung tumours. In rats, NNK induces also liver tumours, nasal cavity tumours, and a high incidence of squamous- cell carcinoma and adenocarcinoma in the lungs of males and, at a significantly lower rate, in females (Hoffmann et al., 1984b; Hoffmann & Hecht, 1985). Although we are presently lacking definite evidence, it may be presumed that NNN and NNK are also formed endogenously when a smoker inhales the precursors, nitrogen oxides and nicotine, as smoke constituents. The inhalation of smoke from a single cigarette provides up to 600 pg of nitrogen oxides and up to 2mg of nicotine. The known catalytic effects of thiocyanate for N-nitrosation (Boyland et al., 1971) favour these reactions in smokers who have elevated ~,~ 0 O~ 0

CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE 157 levels of thiocyanate in the saliva and in blood (IARC, 1986), owing to the detoxificatign of hydrogen cyanide, inhaled as a smoke constituent in amounts of up to 500/~g per cigarette (Brunnemann et al., 1977). A most stimulating observation lies in the fact that NNK is metabolically activated by a-hydroxylation, yielding methyldiazohydroxide. This unstable compound is known to alkylate guanine in DNA to 7-methylguanine and O6-methylguanine in vitro as well as in vivo. Thus, we know today, that nicotine is not only the major habituating agent in tobacco but that it is also a precursor for the powerful carcinogen NNK. Figure 8 depicts the pathway of NNK formation from nicotine. Metabolic activation leads to a-hydroxylation of NNK which gives rise to methyldiazohydroxide. The latter methylates DNA to the pro- mutagenic DNA adduct, O6-methylguanine (Hoffmann & Hecht, 1985). ENVIRONMENTAL TOBACCO SMOKE Since 1981, a number of epidemiological studies have indicated a possible correlation between uptake of environmental tobacco smoke ('passive smoking') and an increased risk for cancer. The IARC concluded: 'The observations on nonsmokers that have been made so far are compatible with either an increased risk from "passive" smoking or an absence of risk. Knowledge of the nature of sidestream and mainstream smoke, of the materials absorbed during "passive" smoking, and of the quantitative relationships between dose and effect that are commonly observed from exposure to carcinogens leads to the conclu- sion that passive smoking gives rise to some risk of cancer.' (IARC, 1986). A comparison of the constituents of mainstream (MS) and sidestream (SS) smoke reveals that these combustion effluents are similar but not the same (Table 10). The differences become particularly apparent when one compares the chemical composition of undiluted MS and SS. Considering that 35-40% of the tobacco is burned during puff- drawing and the remainder during smouldering, one would expect, in the case of a non- filter cigarette, that the release of smoke compounds in the SS would be 50-100% greater Table 10. Comparisons of mainstream (MS) and sidestream (SS) smoke of cigarettes (physicochemical data) Parameters MS SS Peak temperature during formation (°C) -900 ~600 pH (total aerosol) a 6.0-6.2 6.4-6.6 Particle size (/zm) 0.1-1.0 0.01-0.1 Median diameter 0.4 Smoke dilution (vol. %) ~ Carbon monoxide 3-5 ___ 1 Carbon dioxide 8-11 Oxygen 12-16 16-20 Hydrogen 15-3 ~-0.5 85-ram nonfilter cigarette distance of 10 mm from the burning coal

158 HOFFMANN & WYNDER Table 11. Distribution of compounds in mainstream smoke (MS) and sidestream smoke (SS) of nonfilter cigarettes , Compound MS SS/MS Vapour phase Carbon monoxide 10-23 mg 2.5-4.7 Carbon dioxide 20-40 mg 8-11 Benzene 20-50 #g 10 Formaldehyde 5-100 ,ug 0.1 --50 Acrolein 50-100/~g 8-15 Acetone 100-250/a:j 2-5 Hydrogen cyanide 400-500 ,ug 0.1-0.25 Hydrazine 24-43 ng 3.0 Ammonia 50-170/~ 40-170 Methylamine 11.5-28.7/~ 4.2--6.4 Nitrogen oxides 50---600/~g 4-10 N-Nitrosodimethylamine 10-180 ng 20-100 N-NitrosopyrroHdine 2-110 ng 6-30 Particulate phase Particulate matter 15-40 mg 1.3-1.9 Nicotine 1-2.5 mg 2.6-3.3 Phenol 60-140 ,ug 1.6-3.0 Catechol 100-350,t~j 0.6-0.9 Hydroquinone 110-300/~g 0.7-0.9 Aniline 360 ng 30 2-Toluidine 30-160 ng 19 2-Naphthylamine 4.3-27 ng 30 4-Aminobiphenyl 2.4-4.6 ng 31 Benz[a]anthracene 40-70 ng 2-4 Benzo[a]pyrene 10-40 ng 2.5-3.5 N'-Nitrosonornicotine 120-3700 ng 0.5-3 4-(Methylnitrosamino)-l-(3-pyridyl)-l-butanone 120-950 ng 1-4 Cadmium 100 ng 7.2 Nickel 20-3000 ng 13-30 Polonium-210 0.03-1.0 pCi ? than in the MS. However, this is not the case. As seen in Table 11, compounds generated by reduction reactions are formed in significantly higher yields and those formed by oxidation occur in lower yields during smouldering (SS formation) than during puff- drawing (MS formation). These differences are primarily due to the depletion of oxygen inside the burning cone during smouldering as opposed to only a partial oxygen deficiency during puff-drawing. Excessive formation of SS compounds is greatest for ammonia, amines including aromatic amines and, especially, for the volatile carcinogenic N-nitro- samines (VNA). The high yields of VNA in SS explain the fact that they are detectable in smoke-polluted environments in spite of extensive dilution by air. The qualitative and quantitative differ- ences of MS and SS composition and the effects of ageing of SS constituents in the environment make it clear that smoke polluted indoor-air cannot be regarded as 'diluted mainstream smoke'. ~O 0 O~ O~ Oo 0

CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE 159 REDUCTION OF SMOKE CONSTITUENTS One of the earliest and yet most important observations in the association of cancer risk and smoking was that of a dose-response relationship (Wynder & Graham, 1950; Doll & Hill, 1954; Hammond & Horn, 1958). Therefore, during the last two to three decades, a reduced exposure to tobacco smoke by modifying the smoke yields of cigarettes was Fig. 9. US sales-weighted average tar and nicotine yields (adapted from Norman, 1982); RT, reconstituted tobacco; ET, expanded tobacco; F, cigarettes with filter tips; numbers, lengths of filter tips m.o.. ~ I .+o.- 85F 34°0°- 21.0 - - ~.0-- ~.0-. ~0-- 2~.0 - - ~0-- 25.0 - 24.0 - ~0- ~0- 21.0 - ~.0-~ 19.0 18.0 17.0~ 16.0 - lS.0 - 1~5 | I ! lC F I " Filters ~ Porous pa (. RT Nicotine I 100 F thin I I 1965 Year )ers Tar /~ Tip ventilation 3.1 - 3.0 - 2.9 -- 2.1 -- 2.7 -- 2.6 2.S 2.4 2.3 2.2 2.0 1.9 I.| 1.7 1.$ 1.5 1.4 1.3 1970 1.2 1.1 1.0 0.9 0.8 • 1975 1980

Table 12. Reductions of biological activity of smoke from experimental cigarettesa Methods~ Smoke constituents Selective reduction of biological activity c 'Tar' Nicotine Benzo[a]pyrene Carcinogenic[b/ Tumour promotion Remarks Agricultural aspects Tobacco type (Bright-Burley) '~ + + + + + New cultivars + + + ? Fertilization (nitrate) + + + + ? Tobacco processing Cut + -+ + +? ? Use of tobacco midribs + + + + + + + Reconstituted tobacco sheets (RTS) ~ + + + + + + RTS-pap,er process + + + + + + + Expanded tobacco laminae + + + + +? + Expanded tobacco midribs + + + + ++ ? Cigarette production Paper porosity + + + + ? Cellulose acetate filters + + + + + Charcoal filters~ + + + + + Perforated filters + + + + + + + + Some RTS give high CO Smoker's compensation z z Z m From Wynder and Hoffmann (1982) bMelhdology known to be applied to commercial US cigarettes. Reductions: ++, >50%; +, significant; -+, insignificant; __.?, questionable; ?, unknown CComparison of gram-to-gram 'tar' on mouse skin tests and/or hamster smoke inhalations Replacing Bright with Burley tobaccos Data given for RTS relate to those not made by the paper process Reductions of 'tar', nicotine, benzo[a]pyrene (and other nonvolatiles) and volatile N-nitrosamines are, in general, greater with cellulose acetate filters than with charcoal filters. S60~9890~

CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE 161 regarded as one significant step towards diminishing the cancer risks associated with smoking. Measures to reduce the smoke yields included changes in the culti~zation of tobacco, breeding and selection of new varieties, homogenized leaf curing, incorporation of stems and ribs into the tobacco blends, use of reconstituted and expanded tobaccos, and modification of wrappers and filter tips. The most obvious results of these changes in the make-up of cigarettes have been reflected in a trend of declining sales-weighted average 'tar' and nicotine levels in the smoke of cigarettes since 1955. This trend has been observed in many countries. In the USA, sales-weighted average 'tar' and nicotine values have dropped from 38 mg and 2.7 mg, respectively, in 1956 to 13 mg 'tar' and 1.0 mg of nicotine (Tobacco Institute, 1984). Figure 9 graphically documents the decline in 'tar' and nicotine while denoting the technical modifications that have contributed to the reduction of smoke yields of cigarettes (Norman, 1982). For our own studies (Wynder & Hoffmann, 1967; Hoffmann & Wynder, 1976) and for studies by the US National Cancer Institute (1980), experimental cigarettes were made in which specific parameters were changed. The smoke of these cigarettes was analysed and the resulting 'tars' were assayed for carcinogenicity and tumour-promoting activity on mouse skin. The most encouraging results in respect to a selective reduction of tumorigenicity were observed for cigarettes made entirely of reconstituted tobacco, of stems and ribs, of expanded tobacco and of expanded stems and ribs (Table 12). In smoke inhalation studies with modified cigarettes, significant declines in activity were also observed in respect of tumours in the larynx of hamsters (Dontenwill, 1974). We consider these changes in the make-up of cigarettes and a significant reduction of the tumorigenic potential of the resulting smoke as significant progress, although we need to acknowledge that the smoker of cigarettes with a low nicotine content tends to compensate by smoking more intensely (Herning et al., 1981). The IARC (1986) concluded that 'in a few countries, in which smoking has been established for many years, a substantial reduction in mortality from lung cancer has been observed in young and middle-aged men, which is greatest in the youngest age groups. This has occurred at a time when the number of cigarettes smoked by young men in these countries has remained approximately constant. No substantial cause (or cofactor) has so far been identified that offers a plausible explanation for the observed magnitude of the reduction of risk for lung cancer, other than changes in cigarette design which include reduction in tar content.' SUMMARY Tobacco smoke contains more than 3900 constituents. In this presentation we have summarized our present knowledge as to the physicochemical nature of tobacco smoke and specific agents therein. Emphasis has been placed on the discussion of formation and identification of toxic and, especially, of tumorigenic agents in tobacco smoke. In the concluding Table 13 we have listed those smoke constituents in the mainstream smoke of cigarettes that we regard as important contributors to the toxic and carcinogenic potential of tobacco smoke. This judgement is basea on extensive laboratory studies. Finally, data o o~ ~0 o ~o

162 HOFFMANN & WYNDER Table 13. Biologically active agents in mainstream smokea Smoke constituent Conc./cigarette Biological effect Total particulate matter 15-40 mg T, HC Carbon monoxide 10-23 mg T Nicotine 1.0-2.5 mg T Acetaldehyde 0.5-1.2 mg CT Acetone 10.0-250 ~ CT NO× 50-600/xg T Formic acid 80-600/~g CT Hydrogen cyanide 400-500/~g CT, T Catechol 140-500/~g CoC Ammonia 50-130/.zg T Benzene 20-50/~g HC Acrolein 50-100 ,ug CT Acrylonitrile 3.2-15.0 pg C Phenol 60-140/~j TP Formaldehyde 5-100,ug C Carbazole 1/4~ C? 2-Nitropropane 0.2-2.2/xg C N'- Nitrosonornicotine 120-3700 ng C 4-(Methyinitrosamino)-I -(3-pyridyl)-I -butanone 120-950 ng C N'- Nitrosoanabasine 120 ng C? N- Nitrosodiethanolamine 0-40 ng C N- Nitrosopyrrolidine 2-110 ng C N- Nitrosodimethylamine 2-180 ng C N-Nitrosomethylethylamine 0,1-40 ng C N- Nitrosodiethylamine 0.1-28 ng C N- Nitrosodi- n-propylamine 0-1 ng C N- Nitrosodi- n-butylamine 0-3 ng C N- Nitrosopiperidine 0--9 ng C N- Nitrosopyrrolidine 2-42 ng C Hydrazine 24-43 ng C Urethane 20-38 ng C Vinyl chloride 1.3-16 ng HC Benz[a]anthracene 40-60 ng C Benzo[a]pyrene 10-50 ng C 5-Methylchrysene 0.6 ng C Dibenz[a, j]acridine 3-10 ng C 2-Naphthylamine 4.3-27 ng HC 4-Aminobiphenyl 2.4-4.6 ng HC 2-Toluidine 30-160 ng C Polonium-210 0.03-1.0 pCi ~ Quantitative data refer to nonffiter cigarettes b Abbreviations: T, toxic agent; HC, human carcinogen; CT, ciliatoxic agent; CoC, cocarcinogen; TP, tu m~ur promote~; C, animal carcinogen are presented in support of the concept that product modification can reduce the car- cinogenic potential of cigarettes. However, it must be emphasized that the only safe way to avoid the cancer risks associated with smoking is to refrain from smoking.

CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE 163 ACKNOWLEDGEMENTS We greatly appreciate the extensive contributions of our colleagues J.D. Adams, K.D. Brunnemann, S.S. Hecht, E.J. LaVoie and A.S. Rivenson. We thank B. Stadler, D. Conroy and I. Hoffmann for their editorial assistance. Our studies in tobacco carcinogenesis are supported by Grants CA-17613, CA-29580, and CA-35667 from the National Cancer Institute, US Department of Health and Human Services. This is No. XXXIII of the series 'A Study of Tobacco Carcinogenesis'. REFERENCES Adams, J.D., Lee, S.J., Vinchkoski, N., Castonguay, A. & Hoffmann, D. (1983) Chemi- cal studies on tobacco smoke. 73. On the formation of the tobacco-specific carcinogen 4- (methylnitrosamino)-l-(3-pyridyl)-l-butanone during smoking. Cancer Lett., 17, 339-346 Adams, J.D., Lee, S.J. & Hoffmann, D. (1984) Carcinogenic agents in cigarette smoke and the influence of nitrate on their formation. Carcinogenesis, 5, 221-223 Armitage, A.K. & Turner, D.M. (1970) Absorption of nicotine in cigarette and cigar smoke through the oral mucosa. Nature, 226, 1231-1232 Boyland, E., Nice, E. & Williams, K. (1971) The catalysis of nitrosation by thiocyanate in saliva. Food Cosmet. Toxicol., 9, 639-643 Brunnemann, K.D. & Hoffmann, D. (1974) The pH of tobacco smoke. J. Food Cosmet. Toxicol., 12, 115-124 Brunnemann, K.D. & Hoffmann, D. (1982) Pyrolytic origins of major gas phase con- stituents of cigarette smoke. Recent Adv. Tob. Sci., 8, 103-140 Brunnemann, K.D., Hoffmann, D., Wynder, E.L. & Gori, G.B. (1976a) Determination of tar, nicotine, and carbon monoxide in cigarette smoke. A comparison of international smoking conditions. In: Wynder, E.L., Hoffmann, D. & Gori, G.B., eds, Smoking and Health. L Modifying the Risk for the Smoker (US Dept of Health, Education, and Welfare Publ. No. (NIH) 76-1221), Washington DC, pp. 441-449 Brunnemann, K.D., Lee, H.-C. & Hoffmann, D. (1976b) Chemical studies on tobacco smoke XLVII. On the quantitative analysis of catechols and their reduction. Anal. Lett., 9, 939-955 Brunnemann, K.D., Yu, L. & Hoffmann, D. (1977) Gas chromatographic determination of cyanide and cyanogen in tobacco smoke. J. anal. Toxicol., 1, 38-42 Brunnemann, K.D., Masaryk, J. & Hoffmann, D. (1983) Role of tobacco stems in the formation of N-nitrosamines in tobacco and cigarette mainstream and sidestream smoke. J. Agric. Food Chem., 31, 1221-1224 Carter, W.L. & Hasegawa, I. (1975) Fixation of tobacco smoke aerosols for size distribu- tion studies. J. Colloid Interface Sci., 53, 134-141 Doll, R. (1972) Cancers related to smoking. In: Proceedings of the 2nd World Conference on Smoking and Health, London, Pitman Medical, pp. 10-23 Doll, R. & Hill, A.B. (19~4) The mortality of doctors in relation to their smoking habits; a preliminary report. Br. med. J., i, 1451-1455 c, o ..43 0",

164 HOFFMANN & WYNDER Dontenwill, D.P. (1974) Tumorigenic smoke inhalation studies in inbred Syrian hamsters. In: Karbe, E. & Park, J.F., eds, Experimental Lung Cancer - Carcinogenesis and Bioassays, New York, Springer, pp. 331-359 Dontenwill, W., Chevalier, H.J., Harke, H.P., Lafrenz, U., Reckzeh, G. & Schneider, B. (1973) Investigations on the effects of chronic cigarette smoke inhalation in Syrian golden hamsters. J. natl Cancer Inst. 5, 1781-1832 Dube, M.F. & Green, C.R. (1982) Methods of collection of smoke for analytical purposes. Recent Adv. Tob. Sci., 8, 42-102 Hammond, E.C. & Horn, D. (1958) Smoking and death rates - report on forty-four months 0f follow-up in 187 783 men. II. Death rates by cause. J. Am. reed. Assoc., 166, 1294-1308 Harley, N.H., Cohen, B.S. & Tso, T.C. (1980) Polonium-210.'A questionable riskfactorin smoking-related carcinogenesis. In: Gori, G.B. & Bock, F.G., eds, A Safe Cigarette? (Banbury Report No. 3), Cold Spring Harbor, NY, Cold Spring Harbor Laboratory, pp. 93-104 Hecht, S., Carmella, S., Mori, H. & Hoffmann, D. (1981) A study of tobacco car- cinogenesis. XX. Role of catechol as a major cocarcinogen in the weakly acidic fraction of smoke condensate. J. natl Cancer Inst., 66, 163-169 Herning, R.I., Jones, R.T., Bachman, J. & Mines, A.H. (1981) Puff volume increases when low-nicotine cigarettes are smoked. Br. reed. J., 283~ 187-189 Hoffmann, D. & Hecht, S.S. (1985) Perspectives in cancer research. Nicotine-derived N-nitrosamines and tobacco-related cancer: Current status and future directions. Cancer Res., 45, 935-944 Hoffmann, D. & Wynder, E.L. (1971) A study of tobacco carcinogenesis. XI. Tumor initiators, tumor accelerators, and tumor promoting activity of condensate fractions. Cancer, 27~ 848-864 Hoffmann, D. & Wynder, E.L. (1976) Selective reduction of tumorigenicity of tobacco smoke. IlL The reduction of polynuclear aromatic hydrocarbons in cigarette smoke. In: Wynder, E.L., Hoffmann, D. & Gori, G.B., eds, Smoking and Health. L Modifying the Risk for the Smoker, (US Department of Health, Education and Welfare Publ. No. (N1H) 76-1221), Washington DC, pp. 495-504 Hoffmann, D., Rivenson, A., Hecht, S.S., Hilfrich, J., Kobayashi, N. & Wynder, E.L. (1979) Model studies in tobacco carcinogenesis with the Syrian golden hamster. Prog. exp. Tumor Res., 24~ 370-390. Hoffmann, D., Wynder, E.L., Rivenson, A., LaVoie, E.J. & Hecht,'S.S. (1983) Skin bioassays in tobacco carcinogenesis. Prog. exp. Tumor Res., 26, 43-457 Hoffmann, D., Brunnemann, K.D., Adams, J.D. & Hecht, S.S. (1984a) Formation and analysis of N-nitrosamines in tobacco products and their endogenous formation in tobacco consumers. In: O'Neill, I.K., yon Borstel, R.C., Miller, C.T., Long, J. & Bartsch, H., eds, N-Nitroso Compounds: Occurrence, Biological Effects and Relevance to Human Cancer (IARC Scientific Publications No. 57), Lyon, International Agency for Research on Cancer, pp. 743-762 Hoffmann, D., Rivenson, A., Amin, S. & Hecht, S.S. (1984b) Dose-response study of the carcinogenicity of tobacco-specific N-nitrosamines in F344 rats. J. Cancer Res. clin. Oncol., 108, 81-86 IARC (1985) IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals

CHEMISTRY AND BIOACTIVITY OF TOBACCO SMOKE 165 to Humans, Vol. 37, Tobacco Habits Other than Smoking; Betel-Quid and Areca-Nut Chewing; and "Some Related Nitrosamines, Lyon IARC (1986) IARC Monographs on the Evaluation of Carcinogenic Risk of Che~iddIk'to Humans, Vol. 38, Tobacco Smoking, Lyon International Committee for Cigar Smoke Study (1974) Machine smoking of cigars. Coresta Inf. Bull., 1, 31-34 Keith, C.H. & Tesh, P.G. (1965) Measurement of the total smoke issueing from a burning cigarette. Tob. Sci., 9, 61-64 Mohr, U. & Reznik, G. (1978) Tobacco carcinogenesis. Lung Biol. Health Dis., 10, 263-367. National Research Council (1981) The Health Effects of Nitrate, Nitrite and N-Nitroso Compounds, Washington DC, National Academy Press, p. 529 Norman, V. (1977) An overview of the vapor phase, semivolatile and nonvolatile compo- nents of cigarette smoke. Recent Adv. Tob. Sci., 3, 28-58 Norman, V. (1982) Changes in smoke chemistry of modern day cigarettes. Recent Adv. Tob. Sci., 8, 141-177 Radford, E.P. & Hunt, V.R. (1964) Polonium-210, a volatile radioelement in cigarettes. Science, 143, 247-249 Tobacco Institute (1984) US Sales-weighted Average Tar and Nicotine Yield, Washington, DC, p. 2 US National Cancer Institute (1980) Towards a Less Hazardous Cigarette (Report No. 5), Bethesda, MD, p. 29 Wynder, E.L. & Graham, E.A. (1950) Tobacco smoking as a possible etiologic factor in bronchiogenic carcinoma. A study of six hundred and eighty-four proved cases. J. Am. med. Assoc., 143, 329-336 Wynder, E.L. & Hoffmann, D. (1967) Tobacco and Tobacco Smoke. Studies in Experi- mental Carcinogenesis, New York, Academic Press, p. 730 Wynder, E.L. & Hoffmann, D. (1982) Tobacco. In: Schottenfeld, D. & Fraumeni, J. F., Jr, eds, Cancer Epidemiology and Prevention, Philadelphia, W.B. Saunders Co., pp, 277-292

---

THE L/kNCF~T, FEBRUAR~ 22,1986 ~ ' - ~= ~~~ ' 100"000 in 1983 (see fi~re an-d,table I). T~s difference Epidemiolo~, cohtrasts with average increases of 2" 5qo per year be~een ~Y .1973 and 1977 and 0-6% per year between 1978 and 1982. The rate of mcr~se has been slowing down during the past RATES OF L~G CANCER MEN THE UNITED STATES decade. The number of newly diagnosed cases of lung cancer JOHN W. HORM LARRY G. KESSLER Surveillance and Operations Re~earch Branch, Division of Cancer Prevention and Control, National Cancer Institut6 Bethesda, Marylanc~ USA Summary Lung-cancer incidence and mortality rates in the United States were calculated for the years 1973 to 1983. The historically increasing age-adjusted rates for white men levelled off in the late 1970s and fell between 1982 and 1983. These falls were seen for white men only. Both the incidence and mortality rates for women continued to rise with no hint of a reduction. INTRODUCTION LUNG-CANCER incidence and mortality rates in the United States have been increasing at up to 10% per year since at least the 1930s.I'2 During the 1970s the rates of increase in the age- adjusted figures were observed to be slowing down, at least for men.3'4 By at least one analysis of lung-cancer.incidence and mortality trends, the rates for younger men (a~ged <45 years) had been consistently falling during the 1970s.5 This finding led to speculation that as the population aged the age- specific risk of lung cancer in older age groups would fall, with a resultant levelling and finally a reduction in the overall ag.e-_adjusted incidence and mortality rates.6 Previous reports have not presented such an optimistic outlook for women. In fact, both the incidence of and the mortality from lung cancer among American women have continued to rise, with no hint of either a levelling offor a fall in the rates in the near future. We have analysed lung cance~ incidence and mortality trends in the USA for the years 1973 to 1983. SUBJECTS AND METHODS We used cancer incidence data as reported to the National Cancer Insiitute's Surveillance, Epidemiology, and End Results (SEER) programme.7 This ongoing programme, started in 1973, consists of ten geographically defined, population-based turnout registries. Participants in the SEER programme include five whole States, four metropolitan areas, and the Commonwealth of Puerto Rico. Exclusive of Puerto Rico, the SEER programme includes about 10% of the USA population. These registries report every case of cancer diagnosed among residents annually to the National Cancer Institute. The mortality data we used are from the National Center for Health Statistics; they are for the whole of the USA. User tapes of all deaths occurring in the USA are made available annually. Population estimates used as denominators in the calculation of incidence and mortality rate(are produced by the US Bureau of the Census, also on an annual basis. Since the mortality data for Puerto Rico for the later years cbvered here are not available from the National Center for H~alth Statistics, we have not used either incidence or mortalit~data from Puerto Rico in this analysis. The age-adjusted rates )v~rc adjusted to the age distribution of the USA determined by t.,h~ 1970 census by means of the direct method. RESULTS For the first time since at least 1937 there has been a s~gnificant fall in the historically increasing incidence rates of. lung cancer in white men: the age-adjusted incidence fell by 4" 1~ from 82-7 cases per 100 000 in 1982 to 79" 3 cases per among white men reported to SEER fell from 7042 in 1982 to 68"39 in 1983, despite the increasing-age of the population in the SEER registry areas. The incidence rates for black men do not show a.similar fall but the rate of increase does appear to be slowing down. The annual incidence rates for black men vary._greatly because they are based on less than 1000 cases per year. In 1983 the incidence rate for the black men was 12 5.3, about 58% higher than that for white men. The incidence of lung cancer for women ofeither.r~ce ha~ not shown even a hint of a fall during the decade we examined. In fact, the incidence rates for women were increasing at about 6% per year. " .... TABLE I--AGE-ADJUSTED MALE LUNG-CANCER INCIDENCE AND MORTALITY RATES (PER 100 000)BY RAC~E Incidence Mortality Year 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 White Black 72"3 103"9 73.6 I01.0 75"5 99.7 78"2 108"4 79"5 107"7 80"7 112'6 80"4 110"6 81"6 130"4 82"8 123.8 82'7 122"8 79"3 125.3 White Black 61.6 63.2 64"2 65"7 66"8 68"2 68"8 70.1 70" 1 71 "3 71 "2 74"6 78"0 79"3 81 "6 87"4 88'5 89 -4 92 "6 ..... 95"0 97"5 97"3 90- 8O 7O 1973 1975 1977 1979 1991 1983 W_FtR 13F DIFI(~DSI$ Age-adjusted lung-cancer incidence ra~s in white men~ 1973-83.

.I I;..i., )' ', .,.- , , - , _/' - .... 426 ~ ~ • ~ -" " / THELANCET,~BRU~Y22,1986 M~rt~i~ from lung cancer in white men rose ~y 2. 1% per men,.~hereas ~or older men (7~ and over), the morality rates y~r on~verage ~rom 1973 to 1977 a~ 1. I% per y~r ~rom ar~ Jbwer ~ black t~n in white men. A978 td 1982, but has also shown a very slight ~all ~rom 1982 to 1983. The 0.2% decrease is not statistically significant but is consistent with the incidence data and is the first decrease in overall age-adiusted mortality rates for lung cancer since at least 1950.8 However, we do expect to see a concomitant fall in the 1984 mortality rates when. they become available. The fall in the number of new cases was not accompanied by a fall in the number of deaths in the whole country (69 579 in 1982 and 70 519 in 1983). However, since the one-year survival rate for lung cancer is about 30%, a fall in incidence would not necessarily show up as a change of the same magn!tude in the mortality data for the same calendar year. The mortality rates for black men in the whole USA fell slightly but not significantly between 1982 and 1983 (table I). This rate is based on about 9000 black men dying annually in the USA from lung cancer. Mo.rtality from lung cancer among American women has continued to increase at about 6% per year; the rate in 1983 was 24.3 deaths per 100 0,00 women, about one-third the rate for men. The age-specific mortality rates (table If) have been continuously failing since at least 1973 for whirc men aged 35-44 years. The age-specific incidence pattern has been similar but with a little more fluctuation. In the next age group (45-54 years) both incidence and mortality increased from 1973 to 1978 and then began to fall. For white men aged 55 and over the incidence and mortality rates have been generally increasing with some evidence of levelling off in recent years and finally decreasing from 1982 to 1983. The only exception to this pattern was mortality in men of 75-84; this rate rose by 3% from 1982 to 1983. The age-specific trends for lung-cancer incidence and mortality among black men generally follow those for white men but at a level approximately 40% higher. However, the rates for black men are based on much smaller numbers of cases. What is mosi notable about the age-specific rates for black men is that for younger men (35-54 years) the incidence and mortality rates are about double those for white / TABLE II--AGE-SPECIFIC LUNG-CANCER INCIDENCE A'ND MORTALITY RATES (P'ER 100 000) IN WHITE MEN 1973 ~974 1975 1976 1977 1978 1979 1980 1981 1982 1983 Mortality 1973 1974 1975 1976 - -t977 1978 1979 1980 1981 1982 1983 / Age group 35-44 45-54 55764 65-74 75-84 85+ 17"2 76"3 221"0 415"8 442"3 314"8 14"1 83"1 222"8 419"6 456-1 320"7 13"9 89"0 226"8 427"2 468"1 330"9 13~0 85"6 238"4 427"2 531"0 352"8 15'2 86"9 231"6 452-0 522"1 390"5 14-0 90"4 242"3 449'5 531"4 379"3 11"5 87"6 240"3 460"6 518-2 419"0 12"6 88"7 243"7 454"8 553"4 422.2 14"2 83"9 ~F44.2 484"3 555"4 376"9 11"9 78"0 :'244"1 472"0 591"1 437"1 11'2 77"1 227"6 456"8 584"0 354"2 13"8 6~ 191"0 350"5 381-6 259.1 13"5 f0~9"0 195.6 351"9 403"3 272.4 12-6 /69"4 193"9 365"2 413"4 283"6 12-7 ~ 69"3 197"3 370"9 436"9 308'1 11"8': 70"3 197-1' 379"9 453"0 324"0 ~. '~ .. 71"2 203"1 387"3 459"2 346"6 71-0 202"8 389"5 473"6 360"4 '10"6 71"I 204"7 398"8 491"3 371"7 10"3 71-1 202"8 398"2 494"2 394"5 9"6 68"8 206"4 405"6 517"6 415-0 9"6 65"9 203"0 405"0 535"5 413"2 tn women of all races, age-specific incidence and mortality rates have been about level in the 35-44 age group. The incidence and mortality rates in each of the older age groups have.been increasing throughout the study decade. DISCUSSION Errors in the reporting of lung-cancer cases Qr in the population estimates for 1983 could affect the accuracy of the 1983 lung-cancer incidence rates in this report._I-Iowever, the observed fall in male lung cancer occurred in seven of the nine SEER registries. We have examined several other cancers and found no evidence of under-reporting of cases. An error inthe population estimates is unlikely because the trends M the rates were a)so seen in the numbers of cases. Our conclusion that this levelling off and fall in incidence and mortality in white men may signal the beginning of a downturn in lung cancer is consistent with several events in the USA in the past 20 years which have promoted the long- awaited decrease in lung cancer. These include the Surgeon General's report on smoking and health in 19649 coupled with,the requirement for warning labels on cigarette packages and in advertising since 1966; the introduction and accept.ance by smokers of filter-tipped cigarettes; and the development and use of cigarettes low in tar and nicotine,l° After the 1964 Surgeon General's report, the percentage of men reporting smoking decreased from slightly over 50% in 1965 and earlier to about 38% by 1980. This trend continued into 1983, with about 35% of men continuing to smoke.11 If there is a 20-25-year lag between patterns of smoking and lung cancer, we can ex.pect lung-cancer incidence and mortality in American men to continue to fall for the next 20-25 years. " Both the incidence of and the mortality from lung cancer among American women continue to increase at about 6% per year. The smoking prevalence in women remained at about 32% from at least 196 5 to 1976 then fell to about 29% in 1983. The decline in smoking prevalence among women, has not been as large or as rapid as that for men. Thus, the decrease in smoking by women is about 20 years behind that by men; we should not expect to see a fall in the age-adjusted lung-cancer rates for women until early in the next .century. Co~'respondence should be addressed to J. W. H., Room 532 Blair~'Na~fional Cancer Institute; DCPC/SORB/DAS, Bethesda, MD 20892-4200, USA. ff,]~E~.ENCES 1. Cutler S J, Devesa SS. Trends in cancer incidence and mortality in the USA. In: Doll R, Vodopija I, ed~. Host environment interactions in the etiology of cancer in man. IARCSciPub11973; 7: 15-43. 2. Devesa SS, Silverman DT. Cancer incidence and mortality trends in the United States: 1935-74. ff Natl Cancer lnst 1978; 60: 545-71. 3. Pollack ES Harm JW. Trends in cancer incidence and mortality in the Un ted~tates, 1969-76. J Natl Cancer lust 1980; 84: 1091-103. 4. Devesa SS, Harm JW, Connelly RR. Trends in lung cancer incidence and mortality in the United States. In: Mizell M, Correa P, eds. Lung cancer: cat!ses and prevention. Verlag Chemielnt 1984; 3: 33-45. 5. Harm JW, Asire AJ. Changes in lung cancer incidence and mortality rates among Americans: 1969-78. ff Natl Cancer lust 1982; 6~: 833-37. 6. Loeb LA, Ernster VL, Warner KE, Abbotts J~ Laszlo J. Smoking and lung cancer: an overview. Cancer Res 1984; 44: 5940-58. 7~..Young JL Jr~ Percy CL, Asire AJ, eds. Surveillance, epidemiology, and end results program: incidence and mortality data, 1973-77. Natl CancerInst Manogr 1981; I;7: 1-1082. 8. McKay FW, Hanson MR, Miller RW~ eds. Cancer mortality in the United States: 1950-1977. Natl Cance~ lust Monogr 1982; S~: 1-475. 9. Smoking and health. Report of the advisory committee to the Surgeon General of the Public Health Service. US Department of Health Education and Welfare: PHS Publication no 1103: 1964, 1-38% 10. The health consequences of smoking--cancer: A report of the Surgeon General. US Department of Health and Human Services, Publication no (PHS) 82-50179:1982~ 1-322. 11. Health United States 1985. US Department of Health and Human Services. National Center for Health Statistics (in.press). i|

---

omatick~ch ~v~inl st~le ~"D Praha adn~ stran~ ~asopisfi

STUDIA PNEUMOLOGICA ET PHTISEOLOGICA CECHOSLOVACA~ 46, 1985, ~. 3 Kou eni: situace ve Spolkov6 republice N6mecka F. SCHMIDT Forschungs~elle ftir. praventlve Onkologie Mannheim der Universit~it' Heidelberg, p~ednosta Prof. Dr. reed. F. Schmidt Der Aerztliche Arbeitskreis Rauchen und Gesundheit, p~edseda Prof. Dr. reed. F. Schmidt Kype~ne: C~yan;l~ B ~e~epaTn~HO~ Pec~ys~e Fep~aHn~ PE3~E: ~aeTc~ xapaKTepHCTHKa coBpeMeHnoro COCTOHHH~ O no~e~CTBH~X xype- ~s ~ nx oTpame~nn na COCTOSnnn 3~OpOBbS rpa~maa ~e~epaTn~no~ PecnySnn~n FepMa- HHH H coBpeMenHblX ~eponpnaT~fi no 6ops6e c ~ypeaneM. ~aeTca pa36op n xp~T~ecxas oue~a Tax na3~BaeM~x nerKnx cnrapew, xoTop~e Hec~owpa ~a c~n~ammymca npe~anenT- HOCT~ ~ype~na B ~e~epawnBHO~ PecnySnnxe Fep~annn, Be~yw He TOn~KO ~ 5once ~co~o- ~y e~e~HeBn0~y noTpeSnennm cnrapew xypnnb~nKaMn, ~o Ta~e n K 6o~ee CMepTHOCTH no nobody .paKa ~erxnx, npH~eM Ha6nm~aeTc~ yB~en~e.~TO~ CMepTnocwti y 5oaee MOaO~S~X BoapacTHbIX rpynm O~Hofi n3 ~op~ KaK. CHHaHTB ~pe~s~e cnrapeTHoro ~b~Ma Ha COCTOSHHe 3~OpOBb~ SB~SeTCS HpO~3BO~CTBO cHrapeT H3 HaTypaab- ~o~o ~axme 3TO ~ag~m%aeTc~ ~ y Xypnasm~OB cnr~peT }~ T~ygo~. H~CC~B~Oe Kype~e, B Ka~ec~Be o~noro ~3'ZCTO~KO~ 3arps~e~ ~3nea~ofi Cpe~b~, paccMa~pzBaaocs C TO~- K~ 3pean~ ~a~epore~Hoc~~ ~Tpo3aM~OB, ~onaen~paa~ EoTop~x B oKpymamme~ TO~e cHrapeTHoro ~b~Ma gB~BaeT 3Ha~HT~5HO 9B~COKO~ H Ha OCHOBe pe3y~hTaTOB ~X BnH~H~ Ha BO3HHKHOBeHHe EaH~epa y HeKyp~B~HKOB ~ B~H~HH~ ae~oSposoab~O~O Kypea~ ~a BO3HHKHOBeHHe 3agoneBaeMocw~ ~BIXaT~BHOFO au~apaTay Kam~e~m c~osa: nocaeacv~ ~ype~z~ ~ CPF -- Meponp~vn~, aanpa~ean~m npo- 'THB gypeH~fl -- TaK HaSb~BaeM~e ~>nerxae~¢ cnrapcTb] ~ naTypa~bnb~fi Tff6aK ~ naccaBn0e Kypenne KOM. S:tud. ~ne~moL pht~seol, eeehoslov., ~, ~986, No. 3, e. Smoking: the Posflion in the Federal German Republic SUMMARY: The author presents a detailed account 0n the ~o~temPo~arY position as regards sequelae of smoking on the health of citizens ~ of the FGR and on contempo- rary anti-smoking measures. The author analyzes and evaluates critically the problem o[ so-called light cigarettes, which despite the declining prevalence o[ smoking in the FGR led not only to a higher cigarette consumption by ~mokers but also to a higher mortality from lung cancer and its shift into younger age groups. One of the ways ho.w to reduce the harmful effects of cigarette, smoke is to produce cigarettes with nat~al tobacco which because o£ the~ alkaline reaction are not inhaled by smokers, similarly as it is the case in cigar and pipe smokers. PassD8 smoking as one of the important sources o~ contamination of the living environment is evaluated ~rom the aspect ot carcinogenity of nitrosamines, their concentration in the side stream of cigarette Smoke being very high, and based on results o[ investigations of their in- fluence on the development of carcinoma in non-smokers and the effect of smoking on the incidence of respiratory d~seases ~ children. Key works: sequelae ot smoking in the ~Bg -- anti-smoking activities -- so-vailed ,,light" vigareRes -- natural tobacco -- passive smoking O. Stud. pneumoI, pht~seol, eeehos~ov., ~, ~98~, No. 3, p. 136 r~ o z z k k 0 z: 0 a 0

6, 6. 3 sit[on ~mpo- oblem n the dgher ways with )kers, ~f the n the m of [r in- ntary :ailed KliSovd .s.lova: .d~sZedky I¢ou~ent v NSR --- protiku~dck~ 5innosti -- tzv. ,,lehk#. " cigarety. --.p~irodn~ tab~Ic -- pasivn~ kou~en~ Kou~enI a .pasivnl kou~enI jsou mezinarodnlm probl6mem. Nemaji p~e- k~ky stat~ch hranic, ani p~ek~ky ideologick6. To ]e d~vod, pro~ ]e nezbytn6 z~tenzivnit mez~arodnl spolupraci, aby bylo ~actvi potla~eno. Sv~tova zdra- votnicka o~ganizace j~ v roce ~975 konstatovala, ~e ~adn~m jin~m ]edia~m opat~enim ~y nebFlo mo~no zachr~nit vice lidsk~ch ~ivotfl a zabr~it mnoh~m nemocem, ne~ v~razn~m poklesem spot~eby cigaret. Podle Cancer Preventioa Study Americk~ spoleSnosti pro rakovinu zkracuji si t~ci ku~aci, kou~ici den- n~ vice ne~ 20 cigaret, Svfij ~ivot v prfim~ru o 8,3 roku [1}. P~i fiet~eni o p~ed- ~asn~ sm~i v NSR zjistil V~deck~ institut mistnich nemocenskych poji~foven U t~k~ch ku~akfi zkraceni ~Ivota dokonce o 12,3 roku {2}.. Podle tiskov~ho sd~leni Spolkov~ho m~isterstva zdravotnictvi {3) by bFlo mo~no ptedejit 40 v~ech rakov~ch onemocn~ni u mu~fi v NSR tim, kdyby nekou~ili, ]enom u bronchialniho karcinomu, kter~ byl sv~mi 25 00Ofimrtimi zattm posunut u mu- gfi v N~mecku na vrchol v~ech, organov~ch rakovin, by odpadlo t~m~ 30 v~ech ~rti .ha rakovinu u mu~fi. Rakov~a ,,horni kuP~ck~ cesty" (karcinom rt~ fis~i dutiny, hltanu a hrtanu], karcinom moSov~ho m~ch~e a ledvin, sli- nivky b~i~ni a pravd~podobn~ i karcinomu ~aludku k tomu patti t~2. Spolkova vlada odhadla ji~ v roce 1974, ge poSet obSanfi, kte~i ro~n~ p~edSasn~ umiraji, 'dosahuje 140 00O qsob..K tomu je v~ak nutno p~i~ist asi 100 00O p~ipadfi p~ed- ~asn~ ~validity ro~n~ v dfisledku kou~eni (4}; pozd~]i se .pom~r t~chto fidajfi zm~nil. P~esto jsem na jin~m mist~ prokazal (5}, ge tato vys~ka 5lsla ne]sou v ~adn~m ptipad~ ptili~ vysoka, ale prav~ naopak p~ili~ nizk& Pra~d~podobn~ je nutn~ toto 5asteSn~ odmitnuti d~iv~j~ich propo~tfi a nazorfi op~t p~ijmout. L~ka~sk~ sdru~eqi pro ko~eni a zdravi, kter6 se svymi !500 51eny je hrotem kopi boje proti kqu~eni.v NSR, toti~ vzalo .tyto fidaje z~e~ejn~n~ Sp~lkovou vladou za p0dn~t k tomu, aby prosazovalo energicka, opat~eni proti kou~eni. Bohu2el vlady v~ech zemi -- a neni to odvisl~ od spoleSensk~ho z~izeni :vykazuji malo ochoty k ~omu, aby za~edly energicka protiku~acka opat~eni; pouze skandinavsk~ zem~ tvo~i chv~lyhodnou v~ji~u. 05ividnfi necht~ji p~ed- Easn~ porazit kr~vu, kter~ ve form~ dani z tab~u tak pfln~ d~v~ ml~ko, a to jak u nas, tak kdekoli jinde. V NSR a v zemich s voln~m hospoda~stvim je si- tuace oproti zemfm se statnfm tabakovym mon0po!em ztigena dale rozsahlou reklamou, ktera v~e, co ~ini ~lvot cenn~m; ukazuje v souvislosti s k0uPenim. V mimul~m roce bylo jenom v NSR vydano na reklamu clgaret 300 mili6nfi rek. Varovn~ .oznameni na krabi~kach, cigaret: ,,Minlstr zdravotnictvi: Kou~eni ohro~uje Va~e zdravf' dosahlo zde prav~ tak mal~ho fisp~chu, jako zakaz tn- zerce cigaret wtelevizi a v rozhlase. Dries ka~d~ kuP~k vi, ge ]e kou~eni ~kodliv~- a p~esto kouPi date. Setrv~i pPi u~ivani drogy p~es poznanou ~kodli~ost je vyznamnym znamen~m zav~lost- niho chov~nL Mili6ny kuPakfi, kte~i by podle existujicich ~et~eni v NSR t~g radi zanechalt kou~eni, ale vlastni sflou toho nejsou schopni, jsou pr~v~ tak do- kladem n~vykov~ slo~ky nikotinu, jako v~tgina.ku~akfi, kte~i po do~asn~m ne- kuP~ctvi op~t propadli kou~enL Zku~enosti z cel~ho sv~ta ukazaly, ~e pouhou osv~tou nelze z uveden~ch dflvodfl po~statn~ snf~it spot~ebu clgaret. P~esto zkou~i Spolkova vlada ~e~it problem se Spolkovou centralou pro zdravotni osv~tu podobn~ jako dosud, tj. v~zvami k roz~u ,,sv~pravn~ho ob~ana", misto aby se p~ivala pravd~ do o~I, ge u dosp~l~ch navykov~ch kutakfi lze tImto zpfisobem dosahnout jen omezeni kou~enf, jen dil~ich v~sledkfi. L~ka~sk~ sdru- ~enI pro kou~eni a zdravi, kter~ m~o jin~ zorganizovalo v roce 1974 prvni n~- meck~ kongres o neku~actvi v Bad Neuenahr a v rote 1980 -- spole~n~ s N~- meckou centralou proti n~vykov~mu nebezpe~ -- Mezinarodni kongres ,,Kou- 137

~eni a zdravi'$v Bonnu, z toho vyvbdilo jig pied. ~asem zav~r,..~.e t~gi~t~"ve~ke- r~ho sna~.eni prod kou~eni musl b~t sm~rov~no t,a. ty,: kte~f "d'osud,,nel~0uf~;"tj. pfedev~im na d~tt a" mladist~osoby. 7. ntch motivovat.poku~d mob.no .nejv~t~i ~sttak, aby zOstaly neku~ky, ~ , .. ~ Tento c~l sleduje mimo jin~ n~mi vytvo~en~ audiovizu~lni program o zd~- v6tnIch ~kod~ch kou~e~f, kte~ byl |i.~ distribuov~n' asi ve 4000 .kopiich do ~l~Ol. Mfij let~k ,,B~t mlad~m a nek0u~it" (6) byl ji~.,ve 20 vyd~i/~iCh rezd~n p~ede- v~im ve ~kol~ch v. pNbli~n~ 900000 exempl~Nch~. ]e~t~ v~t~iho roz~i~en~ do- s~hl n'~ let~k ,,Fakta" 0 kou~eni", kter]7 jsme.-:- spole~n~ s. 6etn~mi.fi~ivo~ly k' odvyk~ni kou~eni " bezplatn~ rozeslali ku~a~k~m, kter~ by Mdy zanechaly k0u~eni, kdy~. n~m zaslaly frankovanou ob~lku s adresou. PNlo~.ili jsme: jim dal~i let~k s informacemi o ~kodlivosti'kou~eni pro ~.enst@ organizmUs 'a-pro plod, kou]ff-ll ~.ena v t~hotenstvZ Spolkov~ centr~la pro zdravotni osv~tu po- skyfuje z~larma bro~.uru ,,~5 sekund k zamy~leni", vedle dal~ich tiskovim .M.~ tak~ :-spolu S Lidov~mi vysok~mi ~kolami a Nemocensk~mi poji~fovnami ~ rozs0xh~ program pro odvyk~n~ kou~en~ na z~klad~~ metody sebekontroly, kte~ je organlz0v~n Brengelmannem. Vedle toho nabfzeji (etn~ l~zei~ska mista a.sa- nhtoria o'dvykaci kursy jako sou~st sv~ l~ebn~ p~e. Lehk~ k o u[" e:n i .. Cigaretov~ prf~mysl v NSR je obzvl~t~ 6i1~, aby sni2enim mn~.stv~ ~kod- liv~ch l~itek v cigaret~ch u~inil kou~eni pro zdravi m~p_~ ~kodliv~. Od roku 196.0 byl sni~.en obsah ~kodliv~ch l~tek u cigaret vyr~b~n~ch v NSR vice ne~ na polovinu. Nabizen~ cigarety'na trhu v NSR jsou pr~ podle toho ,,nejleh~i na sv~t~", v mnoha' publikacich (7--10} jsem prok~zal, ~e tente sm~r je scest- n~ a slou~.i p]~edevfiim zdravi tab~kov~ho prflmyslu. ]e to obzvl~t~ p~esv~d~iv~ z kr~tkg zpr~ivy komise Evropsk~ho spole~enstvi v Bruselu. Podle ni se snf~.il podil ku~ikfi z celkov~ho po~tu obyvatelstva v zemich EHS ze 43,8 0/0 v. roce 1960 na 39,2 o/0 v roce :[979. V NSR kou~i podle reprezentativniho vzorku oby- • vatel, podle ~et~eni Spolkovg centraly pro zdravotni osv~tu, stale je~t~ 36 °/0 dosp~l~ch ob~anfl. Sou~asn~ stoupla spot~eba cigaret v zemich EHS ze 378 mi- liard na 584 mlliard kusfi roan& M~n~ lidi tedy kouN vice. Prfim~rn~i spot~eba jednohO ku~ka v zemich EHS stoupla od roku 1960 do 1979 ze :[3,5 na 19,4 cigaret dennY. Ve stejn~m ~ase se vy~plhal podil ,,lehk~ch" cigaret s filtrem na trhu V NSR na vice ne~. 87 0/0. ZvI~t~ ,,tisp~n~m" snah~m n~meck~ho ciga- retov~ho prfimyslu o sni~.eni mno~stvi ~kodllv~ch l~tek, p~edev~im nikotinu, odpovidel zv~eni prfim~rn~ spot~eby cigaret ku~iky v NSR. Ta je toti~, nejvy~i ze v~ech zemi EHS: stoupla z 11,3 na 23,0 cigaret denn& Z toho vysvit~ nal~- hav~ zfiv~r: snt~eni p~edev~im obsahu nlkotinu pod ur~itou mez neni ku~kem akceptov~no bez pNslu~n~ odezvy. Iak zn~mo, tab~k se kouN pr~v~ pro obsah nikotinu a nekou~i se pro usu~en~ sal~t. Pokud neni z~visl~mu ku~kovi na- bidnuto p~islu~n~ mno~.stvi nikotinu podle jeho poY.adavku -- jako kupK tak zvan~mi cigaretami bez nikotinu, kter~ u n~s rychle zmizely z obchodu -- p~ejde buff na ,,slln~j~i" zna~ku cigaret, nebo zkusi svflj hlad po nikotinu uspo- kojit jin~m zpfisobem. PN tom se mu nabizej1 dv~ mo~nosti: zv~it spot~ebu cigaret nebo prohloubit inhalaci. Ob~ cesty byly ji~ nastoupeny. Ne naposledy vyglo najevo ze ~et~enl H. P. Harkeho [11} z V~zkumn~ho institutu n~meck~ho cigaretov~ho prfimyslu, ~e jednak podv~dom~ prohlubov~ni inhalace, kter~ mf,~e kolisat kolem dvou de- sitek, jednak zv~eni po~tu tahfi za ~elem vy~iho pNimu nikotinu~ je nesro- vnateln~ dfile~it~j~i ne~ obsah ~kodliv~ch l~tek. To znameh~: jestlige je hloub- ka inhalace pro p~ijem ~kodliv~ch l~tek p~inejmen~fm s:okr~t df~le~t~jg~ ne~ obsah ~kodlivin, mfi~e bgt sou~asn~ sn~enf koncentrace ~kodlivin na polovinu 138 [12 pe~ sni ton v~ zji~ d~ jak do tee sd ~er. ~m~ nil. od na 792 src ko~ gm ZIl~ m{ ko pr~ ip zd: zfi~ de to] oh"

~kbl. ~ede- : 'do- vody :haly jim po- :te.r~ ., kod- ~oku ne~ eh~i mst- ~iv~ ~oce ~by- 19,4 rein Iga- inu, al~- ~em ~sah na- tak ;po- ebu • Po ~e de- ;to- ~b- .~e~ komp,enz0'v~no, jen. 0 i % hlubgi inhalaci, nebo 'dokonce pP.ekompenzov~no. : ..pr~i.ve.m proto do~la Zpr~iva min..zdravotnictvi Spojen~ch st~tfi z roku 198:[ (12} .,k ja.sn~mu, z~v~ru: ,,.Neexistuje .~dn~i bezpe~n~i~ cigareta ani ~eidn~ bez- pe~mi hlad'ina:spot~eby. Kbu~ent ciga~et s nfzk~m .obsahem dehtu a nikotinu sni~.uje riziko rakoviny plic pouze tehdy, jestli~e .neni kompenzov~no. Ale i po- tom jd"g~k.~e srovn~ni s absolutni abstinenci minim~lnt... KompenZa~ni cho- Vaa~ ra.~"~ednosU iehk~c~{ 'ci~a~et, nebo mfi~.e doko~ rizik0 zv~it." Totb zji~t§~', b~lp "prok~#.~flo v~ledky epidemiol0gick~ch stddff, ne v neposledni ~a- d~ v~i~m ~mrt.n0~ti na.piicni karcinom v NSR. A~koli se obsah gkodlivin jak b~l~J zmin~no ~ u "c~garet nabizen:~ch na n~meck~m trl~u snf~.il od roku 196i ~"J' ' " ...... ~' ' " ''~ ' ' " 0 do 1975 wce nel na polovinu a od ti doby poklesl o dalsich 36 70; stoup~ neust~- ~e po~et fimrfi na rakovinu plic. V NSR byl 0d roku 1961 dosagen podle fidajfi .s'polk'o~,'~hp ~tatistick~ho.fif'adu 've Wiesbadenu~ v posledni~h hodnocen~ch le~ tech '.n.o~ ~bsolutni rek0rd. I kdy~ s~ U b~0nchogenn.ih0' karcinomu po'~itg s dlou~, od0bou latenci, r/em~ia by incidence stoupat, aleklesat, jestli~.e je snl- ~.er~. p~s'ah ~ehtu v. souyisiosti "s ni~.gl/fi rizil~em vzniku rakoviny. To plati ze- jm~.a 'p.ro ku~ky cigar.et .mladgich v~ko,v~ch skupin, kte~i od po~tku sv~ ku- ~ck~,.,kar.i~r~y k.ou~i p~ev~.l~ leh~I cigarety s filtrem. Skute~nost je v~ak opa~- n~. Ve v.~k0v~ skupin~ od 35 do 45 let se zv~gila fimrtnost na plicni rakovinu od roku 1968 obzvlagt~ v~znamn~, o vice ne~. 61%. T~. pr~m~rn~ v~k zem~el~ch na plicni r.akovinu se u ku~ikfi cigaret S filtrem, podle naglch ~et~eni. (13) se 792 n,emocn.~mi na bronchogenni karcinom v~znamn~ posunul o 1,6 roku ni~.e ve srovn.~n.i s ku2~iky cigaret bez filtru. Doglo k tomu i p~es to, ~.e prfim~rna dennl spot~eba oigaret byla u ku~ikfi s filtrem o 4,6 cigaret ni~.~i a jejich celkova doba kou~eni s..prfim~rn~m postern 40,5 let byla v~znamn~ krat~i ne~. u ku~k~ ci- gare.t bez. fi.ltru, kde byla 43,1 r0ku. Z toho jsem vyvodil po~adave.k (10] cilen~ na snI'geni gkodlivosti kou~enl na zd,ravL Je vhodn~, pokud mo~.no, sni~.it obsah dehtu, CO a jin~ch ~kodliv~ch latek, .ale nikoli obsah nik~.tinu. Ten by se m~l naopak zv~it, aby ,nlkotinov~i pot~eba" byla uspokojena ni~.~im postern vykou~en~ch cigaret a ~im se ale- spofi sni~il p~ijem i jin~ch gkodliv~ch l~tek. P~irodn~, tab~k t~g pro cigarety! ]egt~ vhodn~j~i by bylo uskute~n~ni mnou ji~. d~ive ~14) navr~en~ho P0- ~.adavkU, pou~.ivat t~. pro cigarety p~irodnl tab~k, kter~ nesv~idi k inhalaci. ]e zn~imo, ~.e ,,hlavni zabij~k" pf'i kou~eni je cigareta. Ku~cl cigar a d~mek jsou m~n~ ohro~.eni ne~ kuf'~ci cigaret, ~proto~.e kou~ d~mky a cig~ir svoji alkalic- kou reakci nenl zpravidla inhalov~n. Cigaretov~ ko.u~ je, jak zn~mo, um~le p~i- praven k inhalaci zvl~tnim zp~sobem sugeni a p~ipravy tab~ku a mimo jin~ i pf'id~inim jist~ch p~isad. Cigareta z pf'irodniho tab~ku s alkalickou reakcl kou~e ~ to je podle m~ho pf'esv~d~enl nejfi~inn~j~i cesta, jak alespofi trochu sni~.it zdravotnl gkody zpfisoben~ kou~enim. ]edin~ bezpe~n~ cigareta samoz~ejm~ z~st~iv~i nevykou~en~ cigareta. Na~izeni, kter~ by p~edepisovalo po#inn~ v~ini takov~chto tabak~ t~. pro v~robu cigaret, by pravd~podobn~ pomohlo p~e- dejit mnoha nemocem a ~ pf'inejmenglm z dlouhodob~ho pohledu -- by za- chr~nilo vlce lidsk~ch ~ivot~, ne~. vgechna dosavadnl opat~eni dohromady. P~i- tom by toto na~izenl nes¢~lo ant hal~ a bylo by realizovateln~ bez technick~ch obti~.L Pasivni kou~eni gkody zpfisoben~ na zdravi pasfvnim kou~enlm se dnes nedaji g~dn~ml v~- deck~mi argumenty pop~it z n~sledujicich dfivodfi: 1 V tab~kov~m kou~i se d~ prokazat vice ne~. 40 kancerogennich l~tek Nejv~tgi ~st t~chto l~tek odch~zi vedlej~im proudem kou~e do okolniho vzdu- 139

chu, odkud ho musi pasivnl ku~ik nucen~ vdechovat. Takov~ kulak v~ak inha- luje tab~kov~ kou~ vice ~l m~n~ z~ed~n~. V~znam tohoto z~edovacMo faktoru' je vgak vyrovn~n nesrovnateln~ vyg~i koncentraci karclnogenfi ve vedlejgfm proudu kou~e~ srovn~me-li jejich koncentraci v hlavnim proudu kou~e, kter~ vdechuje pouze aktivnl ku~ik. 2. Obzvl~gtnl pozornost si zasluhuje v t~to souvislosti vice nee tucet pro- kazateln~ch nitr~osaminfi v tab~kov~m kou~i, a to jak z l~valitativn~ho', tak z kvantitativnMo hlediska. V mno~.stvi 1 ppm jsou nitrosaminy potenci~ilnimi karcinogeny. V~echny -- vice ne~. dvacet -- dosud pot~it~ druhy experiment~l- nich zvi~at reagovaly na nitrosaminy bez v~jimky tvorbou tum0ru. Proto ne- mfi~.e bpt an£ ~lov~k vpjimkou. Pr~iv~ pro tento vysok~ obsah nitrosaminfi plati tab~kov~ kou~ za nejdfile~.it~j~I exogenni ,zdroj nitrosamlnfl v harem prost2edL P~it~.ujiclm jevem p~i tom je, ~.e obsah nitrosaminfi ve vedle]~im proudu koufe ]e podle Brunnemanna a .spol. (22)a~:. pades~tkr~t vyg~i ne~. v hlavnim proudu. Z toho dfivodu je z[edovaci efekt siln~ zpochybn~n, tak~.e p~ijem nitrosaminfi pas£vnimi ku~ky v siln~ zakou~en~ch mistnostech mfl~e dos~hnout hodnot, kte- r~ odpovida~I obsahu nitrosaminfi v hlavnim proudu asi 30 cigaret za hodinu~ Obsah nitrosaminfl ve vedle|~im.prou'du tab~kov~ho kou~e je kupf. nejm:~n~ 1000kr~lt vy~f ne~. v piv~ ~i ne~. v aminophenazonu -- v jinak osv~d~en~m l~ku -- kter9 byl pro nepatrnfi stopy nitrosaminu vy~azen z u~iv~ni. 3. Karclnogeny se odli~uji od jingch jed~ svgm vysloven~ suma~ntm p~so- benim. Jednotliv~ ~sti se s6itaji a~. ke kritick~ prahov~ hodnot~. Jig nepatrn~ d~ivky -- sta~i k tomu mfliontina gramu --zanech~vajf lreverzibilni zm~ny v bufice. Tyto tumor6zni z~irode~ng zm~ny mohou pak dal~im p~sobenim karci- nogennich 16tek nebo kokarcinogenfi p~erfist do kllnicky manifestn~ho karci- nomu. Proto nelze z teoretickgch d~vodfi podat ani jednu naprosto ne~k0dnou d~ivku l~tky vyvol~vajtci karcinom; to je t~. dfivod, ~.e pro ni neexistuje ~idn~ maxim~lni pgipustn6 koncentrace na pracovi~ti. 4. Zatim je k dlspozici nejm~n~ 8 publikaci, kter~ prokazaly vgznamn~ ~as- f~j~i vgskyt bronchogennMo karcinomu u neku~a~ek, kter~ byly provd~ny za ku~ky, ve srovn&ui s t~mi, kter~ byly provd~iny za neku~6ky (16--21]. 5. Karcinogenni pfisobeni pasivniho kou~eni bylo jednozna~n~ prok~z~no v poku.su na zviPeti [vlz 15}. 6. Existuje nejm~n~. 30 prac~ [15j, kter~ do~ly k Souhlasn~mu z~v~ru, 2e d~tl rodi~ ku~k~v~Tznamn~ 6ast~ji 0nemocnf na z6n~ty d~chacich cest ne~. d~ti neku~6k6. 7. Podle velk~ kalifornsk~: studie [23).o" Chronick~m pas~vnim kou~eni na pracovl~ti ovllv~uje nedobrovoln~ kouPenI funk~ni vlastnosti~ plic asi ve stejn~ mf~e jako kou~enf, p~i kter~m se neinhaluje. Tab6kov9 kou~ je tudI~. -- obzvl6~t~ v u~av~en~m prostoru -- nejd~le~.i- t~j~im ~initelem zne~i~fov~nf vzduchu a ~.ivotnMo prost~edL V dob~ stoupaj~cI- ho v~dom~ dfile~itosti ~istpty ~.ivotn~ho prostPedf pPedstavuje toti~, pasivni kou- ~eni pro neku~ky ~innou p~ku, kter~ by v sou~asn~ dob~ odsunula kou~eni c[garet na vedlejgi pozici z d~vodu zne~l~fov~ni ~.ivotnMo prost~edL VytvoPily by se tak podminky, aby byla vyko~en~na p~ita~livost kou~eni pro dorfistaj~ci ml~deL Podle ned~vn~ho ~v~dsk~ho soudn~ho rozsudku bylo onemocn~nt plicni rakovinou jedn~ paslvnl ku~a~ky, kter~i musela po dlouh~ l~ta spolupracovat ve velk~ kancel~sk~ mfstnostl se siln~mi ku~ky, uzn~no jako onemocn~n~ z povol~nL Je tudI~, ochrana neku~k~, zvl~t~ na pracovi~ti, nevyhnuteln& DoMo 29. 11. 1984 [87/84] L i t e r a t u r a u p~ekladatele. P~'elo~il: [. Kozdk F. S., Forsehungstelle fiir prfiventi~e Onkolog~e Mannh~m der Un~uersZt~t Heidelberg, Magbaehstr. 14--16, 68 Mannheim 1, NSE 140 STUDIA P Ku cl Poll Stud. SUMh 1702} com t[on on th friends. F~ and the c! ~ime 3.4% 5.7% of tl in boys w This pher children d child, in 1 among fat found onl o o~

: 2063628~0~

LOW-TAR CIGARETTES PUT To THE TEST /SIR;--A Commentary from Westminster report in The Lanced stated incorrectly that a smoking research project directed from St Thomas' Hospital had collapsed after its condemnation as "unethical" by the Chief Medical Officer for Wales. The feasibility of this project---a double-blind randomised study of the effect on - respiratory symptoms of changing from middle to low tar cigarettes--has~been investigated2 and a large-scale trial~ instigated by the Independent Scientific Committee on Smoking and Health, is underway. This committee strongly urges smokers to stop and non-smokers not to start. However, without a ban on cigarette sales or politically unacceptable tobacco duties many people will continue to smoke and others will start, despite all the warnings. So, while no cigarette can ever be regarded as safe, there is a need to assess "less hamfful" products. The tar, nicotine, and carbon monoxide yields of cigarettes have been falling* but the effect of these product modifications on health is not known. Lung cancer mortality in the UK in men under 70 and women under 55 years has been falling but the interpretation of this trend in terms of tar content is confounded by the accompanying fall in cigarette consumption? The relation betwee~ tar delivery and deaths from chronic bronchitis (on the decline in the UKs) arid coronary heart disease (static~) are even less well defined. Another difficulty is compensatory smoking: Peach et aV compared u~nary nicotine metabolite levels in people who had changed from high to lower tar cigarettes with those in smokers who had not changed and found similar levels. An international workshop in 1985s concluded that further evaluation of the lower tar policy was needed. The difficulties should not be underestimated. Smokers tend to report that they smoke lower tar brands than they actually do and are hazy about brands in earlier years.9 Thus questionnaires on smoking habits need objective validation. Colorimetric assays of urinary nicotine metabolites are cheap, simple, and, for large groups, give a reasonably good indication of amount smoked.1° Because smokers with incipient disease may switch to lower tar brandsTM an experimental design, with random allocation of cigarette type, is essential to test the effect of cigarette modification on health. In the feasibility study~ participants were invited to return an empty packet of their usual cigarette brand. Men aged 20 - 44 who smoked middle tar cigarettes were sent information on the dangers of smoking and those who had no intention of stopping were invited to take part in the trial in which they would be randomly allocated to middle tar/middle nicotine or low tar/low nicotine cigarettes in anonymous packets sold at discount to encourage participation. Participants were asked to collect their cigarette butts and provide urin.e spe .cimens at 0, 3, and 5 weeks. Drop-outs were similar in the two groups and cigarette price did not appear to affect consumption. Of some 19 000 individuals sent the questionnaire, only 1% satisfied the selection criteria and agreed to participate. About 270 000 individuals will need to be sent questionnaires to recruit sufficient participants for the full trial. Participants Wil~ be randomly allocated to middle tar/middle nicotine, low tar]low nicotine, or low tar/middie nicotine cigarettes, to be sold at wholesale price; cigarette uptake will be monitored in a 20% subsample. Smoking habits will be monitored for 6 months and the prevalence of and change in respiratory symptoms will be compared with a control group of non-smokers and ex-smokers. The ethical aspects havre been carefully considered. Those who smoke middle tar cigar~ ettes will receive postal advice from an independent source ~ncouraging them to stop smoking. 3 months later, only those who continue to smoke middle tar cigarettes will be invited to participate in the trial. An independent research institute will undertake the fieldwork. One problem for epidemiologists undertaking large-scale national studies is the need to approach a large number of ethical committees. Some central guidance or, even better, one authoritative ethical body to which epidemiologists involved in national studies could turn would be welcome. Despite some adverse publicity, of the 30 district health authorities approached, 8 . ac~pted the St Thomas' Hospital ethical committee's approval and 1~ had the study protocol approved by their local ethical ~ommittees; 4 ethical committees refused _permission and 3 THE LANCET, IULY 19, 1986 ./'authorities did not wish to take p~Lrt, for a variety of reasons. • . It is only by a well-designed randomised controlled trial that " effects on respiratory health of changing to lower tar cigarettes can be assessed. ° D~amnent of Community Medicine, United Medical and Dental Schools of Guy's and St Thomas' Hospitals, St Thomas' Hospital, London SE1 7EH; W.W. HOLLAND Department of Epidemiology end Community Medicine, University of Bristol Bristol, BS2 2PR J.R.T. COLLBY Deparm~etu of Community Medicine, UMDS of Guy's and St Thomas' Hospitals, St Thomas' Hospital, London SE1 FIONA ~N~ORTH 1. N[cKie D. Advertising and sponsorshil~ by the tobacco industry. Lamer 1986; i: 393. 2. Peach H, Ellard GA, Hayward D M, Morris RW~ Shah D. A double blind randomized controlled trial of the effect of a low versus a middle mr cigargtte on respiratory symptoms: Feasibility study. In: Proceedings of International Meeting ~n Cancer Control Prevention: Tobacco a Major Issue (Moscow, 1985). IARG SdMonogr (in press). 3. Jarvis M.D, Phil M, Russell A4AH. Tar and nicotine yields of IlK cigarettes 1972-1983: Sales-weighted estimates from non-industry~ources. Br ~ ~ddicdon 1985; S0: 429-34. 4. Wald NJ. Smoking. In: Vessey MP, Gray M, eds. Cancer prevention. Oxford: Oxford University Press, 1985: 44-67. 5. Holland WW, Gilderdale S. Epidemiology of chronic bronchitis. In: Scadding Cumming G, eds. Respiratory medicine. London: Heinemmm, 1981; 12-20. 6. Florey C du V, Melia RJW, Darby SC. Changing mortality from ischaemic ~e~rt disease in Great Britain 1968-76. Be Med~ 1978; i: 635-37. 7. Peach H, Hayward DM, Ellard DR, Morris RW, Shah D. Phlegm production and lung function among cigarette smokers changing mr groups during the 1970's. ~ Eplderm'ol ~onmgra Heahh 1986; 40:110-16. 8. Wald N, Stepney R, Haddow J. Is there a future for lower mr yield cigarettes? Lancel 1985; ii: 1111-14. 9. Peach H, Shah D, Morris RW. Validity of smokers' information about present and past cigarette brands: Implications for studies of the effects of falling ~ yield of" cigarettes on health. Thorax 1986; 41: 203-07. 10. Peach H, Ellard GA, Jenner PJ, Morris RW. A simple inexpensive urine test of smoking. Thorax 1985; 40: 351-57. 11. Alderson MR, Lee PN, Wang R. Risks of lung cancer, chro~tic bronchitis, ischaemic hea~ disease, and stroke in relation of type of cigarette smoked..~ Epiderrdol ~ommun Health 1985; 39: 286~3. PROSTAGLANDIN E~ DECREASES ACTIVATION OF ARTERIAL SMOOTH-MUSCLE CELLS SIR,--The antlaggregatory prostaglandins PGEz and PGI2 (epoprostenol, prostacyclin) have been used to tl~at patients with peripheral vascular disease (PVD) but the clinical benefit was much less than had been expected from the very potent in-vitro actions of these compounds. The mechanism was at first thought to be an antiplatelet action, but it now seems unlikely that prostaglandins exert, via their antiaggregatory properties, a clinical action that lasts much longer than the duration of the infusion. A fe'~'~years ago an antiprolfferative effect of prostaglandins and a beneficial effect on vascular wall smooth-muscle cell lipid metabolism was reported in animals,x; We wondered if the inhibitory effect on mitotic activity and proliferation of arterial smooth-muscle cells seen in laboratory animals~,~ is seen in vivo in man. Surgical specimens from the femoral and popliteal artery have been investigated in 33 patients (24 males, 9 females). As a last resort before surgery infusions of PGE~ (1 ng/kg/min) ('Prostarasin'; Sauol-Schwarz, West Germany) for 6 h a-day for 5 days had been given intra-arterially into the affected leg in 15 patients (11 males, 4 females). Tissue was immediately fixed in glutaraldehyde (,phosphate buffered to pH 7.4). Processing for semi-thin sections (.periodic-acid/Schiff and toluidine-blue staining*) and electron microscopy were done and the numbers of activated smooth-muscle cells in intima and media were expressed as the percentage of the total smooth-muscle cells on the basis of cell chromophilia and the predominance of unspecific cells organelles,s The percentage of activated smooth-muscle cells in human arteries is age dependent (table). Intra-arterial PGEI infusions for 5 days significantly decreased the numbers of activated smooth- muscle cells in intima and media (,p <0.01; t test). No apparent difference .between males and females was noted. These data reinforce findings in cultured cells~ and laboratory animals,~ demonstrating an andproliferative action of various prostaglandim. The demonstration of an antiproliferative effect of PGEt in man supports the interest in alternative routes for PGEt

---

O0 0 m >Z 3~8Z9~90~ ~' 0 0 m 0-~-< Z m K m

2063628113

WORLD HEALTH ORGANIZATION INTERNATIONAL AGENCY FOR RESEARCH ON CANCER TOBACCO: A MAJOR INTERNATIONAL HEALTH HAZARD Proceedings of an International Meeting organized by the IARC and co-sponsored by the All-Union Cancer Research Centre of the Academy of Medical Sciences of the USSR, Moscow, USSR held in Moscow, 4-6 June 1985 EDITORS c.;q " ',987 D. O. ZARIDZE R. PETO LIBRARY IARC Scientific Publications No. 74 INTERNATIONAL AGENCY FOR RESEARCH ON CANCER LYON 1986

OVERVIEW OF CANCER TIME-TREND STUDIES IN RELATION TO CHANGES IN CIGARETTE MANUFACTURE R. PETO Imperial Cancer Research Fund Reader in Cancer Studies, Nuf-field Department of Clinical Medicine, Radcliffe Infirmary, Oxford, UK SUMMARY The chief purpose of the present chapter is not to review lung cancer trends in general, but merely to consider the extent to which trends in national lung cancer rates can help assess any differences between the carcinogenic effects of different types of cigarette. For this limited purpose, the British data are uniquely informative, for (1) British male lung cancer rates were already high but stable before the cigarette tar levels were halved, and (2) British male cigarette consumption remained stable for some years thereafter. Against this apparently stable background, an otherwise unexplained decrease of about one-half in British male lung cancer mortality in early middle age has followed the decrease in cigarette tar deliveries, which is consistent with Stellman's conclusion (this volume1), based on review of the case-control and prospective studies, that cigarette-induced lung cancer risks are approximately proportional to machine-measured tar deliveries. Lung cancer trends are also reviewed for males from the USA (where cigarette tar deliveries have been greatly reduced) and from the USSR (where they have not). INTRODUCTION Changes in machine-measured tar yield per cigarette In many countries, cigarette manufacturing methods have undergone substantial changes over the past three decades. The most obvious alteration has been the progressive replacement of nonfilter cigarettes with filter-tipped cigarettes, but other changes have See p. 197. -211 -

212 PETO Fig. 1. US sales-weighted average tar and nicotine yields per manufactured cigarette (from American Cancer Society, 1981 ) 40,0 40 35,0 . 3 5 30.0 3 0 Tar 25 0 ~'~ 2.5 15.0 5.0 - 05 0.0 . 0.0 1950 1955 1960 1965 1970 1975 1980 Year involved the use of different varieties of tobacco plant, different (and generally more porous) types of paper, different methods of shredding and processing the tobacco, and different additives. The smoke from cigarettes yields a condensate that might typically contain several milligrams of 'tar'. This cigarette tar is a complex mixture of hundreds (or even thousands) of different chemicals, many of which can be used to cause cancer in laboratory animals. One of the chief purposes of changing the method of manufacture of cigarettes has been to reduce the amounts of'tar' they deliver when smoked in a standard way by a machine (Fig. 1), in the hope that this would decrease the net adverse effects of smoking. Before filter-tips began to be widely used, typical tar deliveries per cigarette might (depending on the country concerned) have been more than 30 rag. Even in countries where no systematic effort has yet been made to reduce tar deliveries, values in the range 20-30 mg might now be typical, whereas in countries where substantial reductions have been deliberately engendered the average tar delivery is likely now to be under 15 mg - indeed, in Finland an upper limit of 15 mg has recently been introduced. Compensatory smoking The extent to which such changes in cigarette tar deliveries will actually reduce risks is, however, not easy to predict, for smokers do not use cigarettes as predictably as machines

CANCER IN RELATION TO CIGARE'Iq-E-COMPOSITION 213 do, and the composition of a cigarette can influence the manner in which it is smoked. Many studies (e.g., Peach, this volumez) have shown that if cigarettes with a medium nicotine delivery (as measured by a standard machine) are replaced by cigarettes-with a lower one, then the amount of cigarette smoke that smokers choose to take into the periphery of their lungs will increase. At least for nicotine, this increase may suffice to compensate entirely, or almost entirely, for the change in the nature of the smoke, so there may be very little difference in the uptake per cigarette smoked of nicotine into the blood (except for cigarettes with yields so low that few smokers are currently prepared to use them). Moreover, many different components of smoke may vary approximately in parallel with any variations in nicotine, so this 'compensation' is likely to be of substantial relevance to the health hazards caused by the actual use of different types of cigarette by humans. First, it suggests that there may be no great difference between one type of cigarette and another in the amount of pharmacological satisfaction that smokers get from each cigarette (and hence no great difference in the daily number of cigarettes they choose to smoke, or in the ease of adoption or cessation of the habit). Second, it suggests that there may be no great difference between one type of cigarette and another in their production of those adverse health effects, such as heart disease or chronic obstructive lung disease, that are chiefly determined by the amounts of smoke products that reach the periphery of the lung. Third, for adverse health effects, such as bronchial carcinoma, that generally involve not the periphery of the lung but instead anatomic subsites that are higher up in the bronchial tree, it emphasizes the potential unreliability of theoretical predictions about the mag- nitude (or even the direction) of any differences in lung cancer risk between actual use of one type of cigarette and actual use of another. Difficulty in predicting carcinogenic effects on main airways The type of lung cancer usually produced by cigarettes is bronchial carcinoma, arising from the walls and, particularly, the bifurcations of the main airways. For the chief carcinogenic factors in the smoke, however, it is difficult to predict what the effective dose to the bronchi will be as the smoke streams past these wails and bifurcations, especially since it is not reliably known which of the many chemicals in cigarette smoke are chiefly responsible for its carcinogenic (or cocarcinogenic) activity in the bronchus. Indeed, it is not even known with certainty whether these agents are only in the particulate phase or whether some in the gas phase are also importantly relevant. (Cigarette smoke is a mixture of particles, which may swell rapidly in a moist environment, and gases.) Moreover, it is already known that differences in the chemical composition of the smoke substantially alter inhalation practices. Differences in, for example, the speed of inhalation would affect the length of time that the tissues of the main airways are exposed to the gas phase, and these (or other) differences in inhalation might also substantially affect the proportion of particulate matter that is deposited on the walls of the airways. See p. 251.

214 PETO The surprisingly substantial practical relevance of these speculations may be illustrated by two curious observations among men who are heavy smokers: (1) those who say they 'do not inhale'-seem to get almost as much smoke into the periphery of their lungs as other equally heavy smokers who say they 'do inhale', but (2) these 'noninhalers' get it there in a manner that in several studies has been found to give them significantly more lung cancer in their main airways than otherwise similar 'inhalers' get! This apparent anomaly has been nicely reviewed and discussed by Wald etal. (1983) (see also Wald, 1985), who suggest that it could arise chiefly because slow inhalation may expose the walls of the main airways to more of the chief cancer-causing substances than rapid inhalation does. But, although reasonably plausible explanations for the data can certainly be developed from such ideas, there obviously still remains great uncertainty about the quantitative determinants of exposure of the main airways both to gas-phase and, especially, to particulate-phase smoke components. Limitations of laboratory evidence Thus, when comparing the lung cancer hazards that are likely to be conferred by different types of cigarette, evidence from laboratory studies is, for the present, of limited practical value. It is known that differences between different types of cigarette can engender large differences in the extent (and hence, presumably, also the manner) of inhalation; it is known that differences in the manner of inhalation can engender large differences in risk whose magnitude (and even direction!) are difficult to predict; it is not known which the chief cancer-causing agents in cigarette smoke are, and even if it were there would at present be no reliable way of measuring the average extent to which actual smoking patterns would deposit particular agents onto the key target areas in the main airways. (In particular, although the extent of deposition of smoke products into the periphery of the lung can be measured by analyses of blood samples, this is not likely to be proportional to the extent of action of cancer-causing factors on the main airways: Wald et al., 1983.) Restriction of attention to epidemiological evidence From the foregoing, it appears that the only useful way to discover whether there are any important differences in the lung cancer risks caused by the habitual use of different types of cigarette is likely to be direct epidemiological observation. Two main types of epidemiological study may be considered: time-trend studies of an entire population (which will be dealt with below), and 'analytic' studies of individuals, i.e., studies that use standard case-control or prospective methods (Stellman, this volume3). Both 'analytic' and time-trend studies have their strengths and weaknesses, but when (as is actually the case) the conclusions of each are concordant then their strengths reinforce each other, and together they may provide ample evidence to justify practical action. See p. 197.

CANCER IN RELATION TO ClGARE-FI'E COMPOSITION 215 Time-trend studies Two of the main strengths of time-trend studies of an entire population are, first, that they may allow (at least in early middle age) direct comparison of prolonged high-tar usage with reasonably prolonged low-tar usage, and second, that they allow the study of extremely large numbers, so lung cancer rates can be studied meaningfully even for people as young as 30-34 or 35-39 years of age, among whom the disease is extremely rare but among whom the contrast in lifelong average tar delivery per cigarette may currently be greatest. The main disadvantage of time-trend studies is, of course, that other causes of change of lung cancer incidence may also have been operating, so it may be difficult to be certain exactly how much of any trend that is observed can be ascribed to changes in cigarette composition. If, however, a time-trend study is undertaken in a population in which the big increase in lung cancer mortality in those of middle age due to the delayed effect of a previous increase in cigarette use was largely completed before the tar levels began to undergo their main decrease, and in which no large change in current cigarette consump- tion is in progress, then it may usefullycomplement the analytic studies. Analytic studies The great strength of analytic studies (i.e., of studies comparing different individuals within the same population) is that they should be less subject to certain systematic biases than a time-trend study might be. Their weaknesses, however, are that it is difficult to use them to study the crucially informative period in early middle age sufficiently accurately (because the disease is so rare at these ages), and that it is generally impossible to use them to compare the prolonged use of high-tar cigarettes with the prolonged use of low-tar cigarettes, simply because as low-tar cigarettes become widely available in a particular country high-tar ones tend to disappear, so that the two do not coexist widely for long. Hence, we may expect a systematic tendency for analytic studies to underestimate any true differences in risk between prolonged use of high-tar cigarettes and prolonged use of medium-tar cigarettes. Despite this, the findings in analytic studies have actually been surprisingly substantial and consistent, and Stellman (this volume1), after reviewing them, concludes that 'relative risk for lung cancer is in rough proportion to tar yield' (i.e., to tar yield as measured by a standard smoking machine), adding that 'It is very likely that as successive cohorts of smokers are exposed to cigarettes of much lower yield for much greater proportions of their lives, the associated risks will decline even further.' (One further study - that of Alderson et al., 1985 - has recently yielded unpromising results, but inclusion of it would merely dilute, rather than reverse, Stellman's conclusions.) Review of some time-trend studies now follows, (1) to determine whether national trends, especially in people in early middle age, support Stellman's conclusion that substantial risk reductions have already been achieved, and (2) to determine how large the risk reductions in people in early middle age appear to be, especially in populations where the large increases in tobacco-induced lung cancer in people in early middle age had already been completed, or nearly completed, before substantial tar-level reductions emerged. Data will be presented from five developed populations, chosen to illustrate contrasting trends in patterns of cigarette usage and tar delivery.

216 PETO Two (Finnish and, especially, UK males) involve countries where cigarette smoking by young men appears to have become widespread in the first quarter of the century (Lee, 1975) and where large changes in cigarettes, which lowered tar yields, were implemented in the third quarter (Lee, 1976; Wald etal., 1981). Consequently, any moderate effects that these tar-level reductions may have on lung cancer can be assessed against a background rate of male lung cancer that had, at least in people in early middle age, already approxi- mately stabilized, albeit at a very high level (Doll & Peto, 1981). One (American males) involves a population where cigarette smoking by young adults increased Substantially in the second quarter of the century (Lee, 1975) and where tar-level reductions were also implemented in the third quarter (Fig. 1; US Surgeon General, 1982). Consequently, any moderate effects of these tar-level reductions on lung cancer rates have to be assessed against a background of the rapid rises in lung cancer produced by the delayed effects of the earlier increase in cigarette usage (Doll & Peto, 1981). The fourth example (French females) involves a population where smoking became common only in the third quarter of the century (Hill & Flamant, 1985), and because this increase in cigarette usage is so recent the large increase in lung cancer that it will eventually produce has not yet really begun to materialize. The fifth and final example (males in the USSR) differs not so much in timing but in tar trends. The USSR is a country where cigarette smoking by young men appears (although reliable data are not available) to have become widespread during the first half of the century, but where tar levels still remain much higher than they currently are in the first three countries. Perhaps because of this, the lung cancer rates in early middle-aged men in the USSR appear for the present to be remaining as high as those in the UK before British tar levels were reduced (Napalkov et aL, 1983). METHODS FOR TREND ASSESSMENT Sources of data on history of cigarette usage In developed countries, where cigarette sales are monitored quite closely, sales-weigh- ted data on actual cigarette sales per head (Lee, 1975) are usually reasonably reliable, and trends in these can generally be accepted as real and meaningful. Trends in data from questionnaires on the proportion of smokers or on the total numbers of cigarettes smoked may in contrast be systematically misleading, as antismoking propaganda may have a much larger effect on people's self-reported smoking than on their actual smoking. [Recent divergences between the trends in self-reported smoking and actual sales in the USA (Warner, 1978), France (Hill & Flamant, 1985) and elsewhere illustrate this point.] If data from questionnaires are to be used at all, they should therefore perhaps be used merely to apportion total sales between various sex- and age-specific categories. Another potentially misleading way of using questionnaires is to ask a sample of people what they used to smoke some decades previously, and then to extrapolate these answers back to well before 1950, when almost no direct information existed as to who smoked what. Such extrapolations may yield moderately useful (though not wholly reliable) estimates of the proportions of all cigarettes smoked previously by people of each sex, but may yield considerably less reliable estimates of age-specific, sex-specific past habits - as,

CANCER IN RELATION TO CIGARETTE COMPOSITION 217 perhaps, in the paper of Lee (1975) and in some of the US Surgeon-General's reports. For this reason, excessively detailed modelling of national lung cancer trends should generally be avoided (especially in old people, where the trends in lung cancer may depend particu- larly strongly on age-specific trends in smoking many decades previously). Use of mortality data on people in middle age Mortality data on people in middle age provide perhaps the most reliable source available for assessment of lung cancer trends, for reasons discussed in IARC (1986) and by Doll and Peto (1981). REAL EFFECTS OF HISTORY OF TOBACCO USAGE ON LUNG CANCER TRENDS The chief effects of tobacco on national lung cancer trends that need to be assessed are the long-delayed effects of nationwide adoption of cigarette usage, and the moderately delayed effects of nationwide decreases in the tar delivery per cigarette (Doll & Peto, 1981; Peto & Doll, 1984). Effects of nationwide adoption of cigarette usage Cigarettes cause a far greater risk of lung cancer than others forms of tobacco do (US Surgeon General, 1979, 1982). So, when a nation adopts widespread cigarette usage, large real increases in lung cancer will eventually follow, whether the switch is from nothing to cigarettes or whether it is from other forms of tobacco to cigarettes. These large increases in lung cancer may, however, appear many decades after the large increases in cigarette Table 1. 40 years of evolution of British annual respiratory~ cancer death certification rates per 100 000 men, emphasizing the high rates in men born about 1900 Age range (years) 1943 # 1953 ~ 1963 ~ 1973 # 1983 e 30-34 4 4 3 2 1 40-44 20 25 22 18 12 50-54 63 123 122 107 77 60-64 107 258 367 354 299 70-74 80 265 497 678 640 80-84 ~ 144 342 602 834 No. of cigarettes per man per day in preceding year 10.7 9.9 10.6 10.6 7.1 "All respiratory and intrathoracic in England and Wales, excluding nose, sinuses and larynx ~ Mean for five years centred on index year c Mean for 1982-1984 only (1985 not yet available, and 1981 subject to slight underestimation at older ages, due to temporary difficulties [n central records office; Ward, 1985) aData for people aged 80 and above were not subdivided, and anyway were subject to gross under-certification Data from Doll and Peto (1981 ; appendix E) and Office of Population Censuses and Surveys (1984a,b,c,d, 1985a,b)

218 PETO usage, simply because it is those who start to smoke in early adult life who will be at greatest risk in middle and old age (Doll & Peto, 1981), and 30 years separate the late teenage years from the age-range 45-49, while 60 years separate the late teenage years from the age- range 75-79. Hence, if other things are equal, it will probably not be until about 20, 30, 40, 50 and 60 years after cigarette smoking in people in their late teens or early twenties approaches a maximum that lung cancer at ages.35-39, 45-49, 55-59, 65-69 and 75-79 can be expected to do so. Thus, for 30 years after cigarette smoking among teenagers finally becomes nearly maximal, lung cancer rates in people at ages 45-49 may continue to rise. Thereafter they may become stable, while lung cancer rates in people in the age-range 65-69 may continue to rise for another 20 years before they too stabilize. Thus, lung cancer rates in people at ages 35-39, 45-49, 55-59, 65-69 and 75-79 may approximately reach their maxima 20, 30, 40, 50 and 60 years after a common starting point. This possibility is exemplified by the British male lung cancer death certification rates (Table 1). The underlined rates are those for men born in about 1900, and are approximately maximal. Modification of lung cancer hazard by changes in cigarette tar defivery The general pattern in each age group is one of sharp increases preceding this maximum, followed by an approximate stability that is disturbed only by the recent decreases that have begun to take place in people in early middle age following substantial changes (Wald et aL, 1981) in tar delivery per British cigarette. Similar decreases are beginning to emerge in Finland (Fig. 2), where tar levels have also decreased substantially, due in part to progressive abandonment of the 'Russian-style' papyrossi cigarettes that used to be favoured in Finland (Lee, 1975). It is interesting to contrast these figures from countries where tar levels have been reduced with the corresponding figures from a country such as the USSR, where they have not (Table 2). In the USSR, typical tar deliveries per cigarette are still running at about 20-30 mg (IARC, 1986; Zaridze et al., this volume4), with a mean of perhaps 25 mg. This is nearly as high as the tar deliveries of US or UK cigarettes in the 1950s, before their tar deliveries were halved (Fig. 1, Fig. 3), and it may be noteworthy that the USSR lung cancer incidence rates in people in middle age (Table 2) appear to be converging towards the old high UK lung cancer rates of the 1950s and not towards the lower rates that now obtain in both the UK (Table 1) and Finland (Fig. 2). If the hypothesis is true that tar-level decreases are important determinants of the recent decreases in lung cancer mortality in people in early middle age in the UK and Finland, then one may wonder why corresponding differences in risk are not seen in other countries. A possible answer might be that they are being seen but not recognized, simply because moderate (e.g., two-fold) differences such as these can easily be swamped by the vast increases in lung cancer that are being produced in many countries by the delayed effects of past changes in cigarette usage (Doll & Peto, 1981). On this view, the reason why men in the UK and Finland provide such a useful 'natural experiment' for observation of the effects of tar-level changes is that these are perhaps the only two countries in the world See p. 75.

CANCER IN RELATION TO CIGARETTE COMPOSITION 219 Fig. 2. 20 years of evolution of Finnish lung cancer incidence (from IARC, 1986) 5OO 200 100 10 5 ~75-79 65-69 60 -64 55-59 50-54 40 -44 35-39 1956-1960 1961-1965 1966-1970 1971-1975 1976-1980 Period Owing to the use of a 'log' scale, the decreases over the past 15 years (i.e., 1963-1978) at ages 45-49, 40-44 and 35-39 may not look important, yet they would represent, respectively, avoidance of about 31%, 41% and 53% of the 1963 lung.cancer deaths at these ages. As in the U K, changes in incidence are due chiefly to changes in the risk per cigarette rather than to changes in the number of cigar3ttes smoked. Ind~ed, except for a temporary decrease during the Second World War, cigarette consumption per Finnish adult has been fairly steady for more than 50 years, averaging about four/day and five/day in the second and third quarters of the present century, respectively (Lee, 1975).

220 PETO Table 2. 20 years of evolution of USSR annual lung cancer incidence registration rates per 1 O0 000 men, compared with the corresponding rates in the U K USSR incidence~ England & Wales Age range mortality x 1.1 ~ (years) 1960 1965 1970 1975 1980 1958 1983 30-39 3 5 6 7 6. 7 3 40-49 24 29 35 46 47 47 24 50-59 85 127 142 153 176 197 136 ~From IARC (1986) b By 1958, British mortality rates in people at ages 30-59 had reached their maxima and had stabilized. Rates for lO- year age groups are estimated as averages of the rates for the two corresponding five-year ages-groups in Table 4, and multiplication by 1.1 is intended to provide an approximate estimate of the ratio of incidence to mortality that "might be expected in parts of the UK, such as Birmingham, where registration of all incident cases has been in progress for several years (Waterhouse et aL, 1976). c USSR incidence data for 1960 are published for 60 yea~ and over, without subdivision. Fig. 3. Sales-weighted tar, nicotine and carbon monoxide yields of UK cigarettes (from Wald, 1985) 35 , 3.5 "~ar ~ 3.0 Nicotine ~ .......... ~-~ ~ 2O 15 ~ 10 5 1930 1940 1980 , , 0 1950 1960 1970 1990 Year of manufacture of cigarettes 0.5 (Lee, 1975) where cigarette smoking by young men became established so long ago that the lung cancer rates in early middle-aged men had stabilized (or, in the case of Finland, nearly stabilized) by the late 1950s, i.e., before the large tar-level reductions began. Conse- quently, these were the only two populations in which the effects of tar-level changes on lung cancer were monitored against a background of roughly constant lung cancer rates instead of against a background of rapidly rising lung cancer rates, as, for example, in the USA (Table 3; Doll & Peto, 1981). In the USA, cigarette sales increased between the two World Wars from one (in 1918) up to five (in 1939) cigarettes per adult per day and then during the Second World War they

CANCER IN RELATION TO CIGARETFE COMPOSITION Table 3. 40 years of US annual respiratory cancer death certification rates per 100 000 men, emphasizing the high rates in men born in the late 1920sa - Age range (years) 1940 1950 1960 1970 1980 30-34 1.5 1.7 2.4 2.1 1.3 40-44 7 11 15 22 19 50-54 24 47 67 87 102 60-64 41 97 166 225 261 70-74 38 103 211 355 444 80-84 30 77 152 291 467 Actual period studied 1938-1942 1948-1952 1958-1962 1968-1972 1978-1981 No. of cigarettes per adult per day in preceding year 5 10 11 11 11 ~ From IARC (1986) 221 quickly doubled, and have remained at about 10, 11 or 12 per adult per day ever since (Lee, 1976). As a delayed effect of this large increase before and, especially, during the Second World War in cigarette usage by young men, large increases in US male lung cancer death rates were taking place throughout the 1950s and 1960s, the maximal rate in any age group being seen among those who reached adulthood in the late 1940s (Doll & Peto, 1981). Thus, among US males aged 30-34, 35-39, 40-44 and 45-49 the maximum tung cancer rate has been reached, and in those at 50-54 it should recently have been reached (Table 3). Within each age group, the large increases before the maxima are clearly seen; in addition, however, there does appear to be a slight decrease after the maximum is attained (espe- cially in those at ages 30-34). This might reflect the effects of tar-level changes. Even if it does, however, and even if these effects spread to older age groups over the remainder of this century, they cannot be expected to outweigh the large increase in US lung cancer death rates in old age that will presumably continue to emerge throughout this century as a delayed effect of the large increase in cigarette usage by young adults before and during the Second World War. RECAPITULATION Rationale for study of UK male trends As already noted, large, but wholly artefactual, trends in lung cance~ death certification rates can result merely from improvements in the accuracy of diagnosis and/or certification of the disease. Moreover, large real trends in lung cancer rates can result from changes in patterns of tobacco use. (In particular, large increases in rates of the disease can be expected a few decades after the widespread adoption of cigarette smoking by young adults.) The key question is whether, in addition to these large artefactual changes and large real changes in lung cancer, any moderate decreases in the disease during the 1970s or early 1980s can confidently be attributed to the approximate halving of cigarette tar

222 PETO Table 4. Recent trends in England and Wales in lung" cancer death certification ratesb per million men in middle age (Note both the approximate constancy before the large decreases in tar delivery per cigarette in about 1960, and the large- and accelerating - decrease thereafter) Age range 1953 1958 1978 1983 (years) % Change from 1968 to 1978 1983 30-34 37 36 17 14 -54% -62% 35-39 100 94 56 44 -41% -53% 40-~44 250 253 139 122 -45% - 52% 45--49 584 594 402 321 -33% -46% 50-54 1232 1254 999 765 -20% -39% 55-59 2018 2326 1897 1705 -18% -27% "Data include those for pleura, etc. (1953---t 958:[CO6 and 7" 162-164; 1976-1978:ICD8 162-163; 1979-1984: ICD9 162-165), and hence the downward trend in bronchial carcinoma rates is slightly diluted by the upward trend in rates of pleural mesothelioma. ~ Each rate is for a five-year period centred on the index year (i.e., 1951-1955, 1956-1960, 1976-1980), except for the last one, which is for a three-year period (1982-1984). CSources of data: 1951-1955 and 1956-1960 numbers of deaths and populations are from Office of Population Censu sos and Su treys (1975). 1976-1980 and 1982-1984 nu mbe ra of d oaths are fro m the annual mortality retu ms of the Registrar-General (Office of Population Censuses and Surveys, 1978,1979, 1980a,b, 1982, 1983, 1984a,b, 1985a). 1978-1984 population estimates are from the revisions published after the 1981 census (Office of Population Censuses and Surveys. 1984c.d, 1985b): these differ slightly from the original (unrevised) estimates in the annual mortality returns. Wald (1985) has suggested that there may have been some underascertainment of lung cancer in 1981 compared with adjacent years, but in fact incTusion of the 1981 data would not materially alter the above 1982-1984 rates. deliveries that took place in some countries in the 1960s and late 1950s. The chief difficulty, of course, is that in general moderate decreases cannot confidently be identified against a background of large increases. Hence, very few national trends can yield really useful information about the effects of changes in cigarette tar deliveries. Obviously, populations in which the epidemic has not yet emerged (i.e., where lung cancer rates are not yet dominated by cigarette smoking) cannot do so, and nor can populations in which the epidemic was still emerging rapidly during the 1960s. Thus, the most informative popula- tions would be those for which, at least in some age/sex categories, the cigarette-induced lung cancer rates were high but stable during the late 1950s and early 1960s. The only two populations that really meet this criterion are UK and Finnish middle-aged males (Table 4, Fig. 2), and, in both countries, large changes in cigarette manufacture took place during the 1960s. In Finland, however, these changes involved, among other things, one rather unusual feature, viz., replacement of what were popularly called 'Russian-style' papyrossi cigarettes (i.e., cigarettes with a long hollow mouthpiece instead of a filter) by conventional manufactured cigarettes. In the UK, the change was merely from one conventional type of manufactured cigarette to another, and was accompanied by a decrease in sales-weighted tar delivery that has been reliably documented (see Fig. 3, from Wald, 1985). The fact that UK male lung cancer rates had already stabilized before the cigarette tar deliveries cb.anged substantially, together with the nature of the change in the cigarettes that was involved, makes the UK lung cancer trends uniquely informative about d:~fferences between one type of conventional cigarette and another. Moreover, male cigarette consumption in the UK (Table 5) was remarkably steady throughout the period

"ulty, nst a seful tions yet the mla- uced ~ged :ook rigs, [yle' ) by one ~y a rom the the ,out aale "iod CANCER IN RELATION TO CIGARE'FI'E COMPOSITION 223 Table 5. Daily consumption of manufactured cigarettes per UK male aged 15 or over, 1920-1982a Annual, 1975-1982, as percentage Quinquennial 1920-1979 of mean (10.4) during 1940-1974 1920--1924 5.8 1925--1929 6.6 1940--1974 100% 1930--1934 7.6 1975 98% 1935--1939 9.1 1976 93% 1940--1944 10.7 1977 87% 1945--1949 10.5 1978 88% 1950--1954 9.9 Mean 1940-1974= 10.4 t 979 87% 1955-1959 10.5 manufactured cigarettes 1980 85% 1960-1964 10.6 dally 1981 74% 1965-1969 10.3 1982 68% 1970-1974 10.6 1975-1979 9.4 From Wald (1985) Table 6. Approximate proportionofcigarette- induceda lung cancer likelyb to have been avoided by the percentage reductions in daily cigarette consumption in Table 5 (last column) Age range (years) 1978 1983 30--34 4.8% a .18.3% 35--39 ' 3.8% 15.6% 40--44 3.1% 13.5% 45--49 2.7% 11.9% 50-54 2.3% 10.6% 55-59 2.1% 9.6% aEstimated on the assumption that the ratio of the cigarette- induced lung cancer mortality in the age range Ato (A+4) years among regular smokers who have stopped for Y years is approxi- mately proportional to the fourth power of (A-Y-15): Doll & Peto (1976, 1978) ~Since a small proporhon of lung cancer is not ascribable to tobacco, the proportion ol all lung cancer thus avoidable might be slightly less. from 1940 to the mid-1970s, and the changes in it during the late 1970s were too small and too recent to have had any appreciable effect on UK lung cancer mortality in the late 1970s, which further simplifies matters. Indeed, the 13% decrease in cigarette consumption during the mid-1970s was, in 1978, still so recent that it would have been expected to produce a decrease of only 3 or 4% in male lung cancer mortality at ages 35-44 (Table 6). The percentage of UK men who described themselves, in reply to survey questionnaires, as 'smokers' had decreased during the 1970s rather more than male cigarette consumption

224 PETO had done, indicating that the average daily consumption per smoker was slightly higher in 1978 than it had been a few years earlier. If this finding is accepted, it suggests either that as some smokers gave up others smoked slightly more, or (perhaps more plausibly) that light smokers were more likely to give up than heavy smokers were. In neither case, however, would this change have been expected to diminish the lung cancer rate, since the excess lung cancer incidence per cigarette among heavy smokers is at least as great as that among light smokers (Doll & Peto, 1978). Results of study of UK male trends Hence, changes in cigarette smoking alone can account for a decrease of only a few per cent in the 1978 UK lung cancer death rates (e.g., 3 or 4% or so), and no large change in the curability of the disease has taken place since the 1950s. The actual decrease between 1958 and 1978 in mortality among men aged 35-44 between 1958 and 1978 was 40-50%, suggesting about a 40% reduction over and above any changes due to differences in the numbers smoked or in cancer therapy. These decreases, moreover, continued if anything to accelerate over the next few years, so that by 1983 the UK lung cancer death rate among men aged 35--44 was less than half what it had been in 1958 and was approximately half what it would have been expected to be just on the basis of changes in the numbers of cigarettes actually smoked (Table 6). It is too soon to know how great these decreases will ultimately become, but it certainly appears that the lung cancer incidence associated with a given history of cigarette usage will (at least in people in early middle age) be no more than half as great in the future as it was in the 1950s. This decrease by at least half in the lung cancer risks associated with a given habit has coincided with a halving of the tar delivery per cigarette, and although other factors may have influenced the lung cancer trends, none are known that, separately or together, would be expected to have had an effect even half as large as this. As already noted, the conclusion from Stellman's review of the case-control and pros- pective studies (this volumeS; see also, however, Alderson et al., 1986) was that 'relative risk for lung cancer is in rough proportion to tar yield', and that 'It is very likely that as successive cohorts of smokers are exposed to cigarettes of much lower yield for much greater proportions of their lives, the associated risks will decline even further'. Now, in the one country where an analysis of national trends would be expected to be most informative, a halving of cigarette tar deliveries appears to have been followed by an otherwise unexplained halving of the lung cancer risk, with further rapid decreases in progress. The study of national trends is usually a rather crude epidemiological tool, since it is always possible that some unsuspected factor has been overlooked, but in this instance it does appear to offer substantial support for the conclusions suggested by the 'analytic' studies. Eventual size of risk reduction suggested by UK male trends If these national trends are indeed due in substantial part to changes in the carcinogenic- ity of cigarettes, then it would be of considerable interest to know how large the decrease See p. 197.

CANCER IN RELATION TO CIGARETTE COMPOSITION 225 will eventually become. But, this cannot yet be answered for it is not possible to predict reliably whether the main effect of changing tar deliveries should be rapid or slow to emerge. If the timing of any effects of tar-level changes were analogous to the timing of the effects of cessation of smoking, then within only 10-15 years of tar-level decreases any changes in lung cancer would become apparent. In view, however, of the great importance of cigarette smoking in early adult life (Peto, this volume6), it is possible that tar levels experienced in early adult life might be importantly relevant to lung cancer risks many decades later, in which case the full effects of any changes in tar delivery might take several decades to emerge. If both effects applied, then one might expect to see (at least in a country such as the UK, where lung cancer rates had stabilized before tar-level reductions were introduced) decreases in lung cancer rates in adults of all ages as a result of the past 15 or 20 years of lowered tar levels, with the largest percentage decreases occurring, at least for the present, in people in the youngest age groups. This does indeed appear to be more or less what is currently being seen among UK males (Tables 2 and 4), so it is possible that the percentage change in those in early middle age may provide the first clear indication of what can ultimately be expected for adults of all ages, even though only a small minority of cancer deaths take place in people in early middle age. This would suggest that the introduction of cigarette tar-level reductions in countries where tar levels remain high might (unless it diverted attention from the much more important need to discourage smoking) ultimately avoid about half of all cigarette-induced lung cancer. In countries where lung cancer accounts for about one-third of all tobacco- related deaths, therefore, such changes might in turn avoid 10 or 20% of all tobacco- related deaths, if it is assumed that such changes have no comparable effect on other smoking-related diseases or on the extent to which people choose to smoke. REFERENCES Alderson, M.R., Lee, P.N. & Wang, R. (1985) Risk of lung cancer chronic bronchitis, ischaemic heart disease and stroke in relation to the type of cigarette smoked. J. Epidemiol. Community Health, 39, 286-293 American Cancer Society (1981) US tar/nicotine levels dropping. World Smok. Health, 62, 47 Doll, R. & Peto, R. (1976) Mortality in relation to smoking: 20 years' observations on male British doctors. Br. reed. J., ii, 1524-1535 Doll, R. & Peto, R. (1978) Cigarette smoking and bronchial carcinoma: dose and time relationships among regular smokers and lifelong non-smokers. J. Epidemiol. Com- munity Health, 32, 303-313 Doll, R. & Peto, R. (1981) The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J. natl Cancer Inst., 66, 1191-1308 Hill, C. & Flamant, R. (1985) A major cause of epidemic: the tobacco consumption increase in France (Fr.) Rev. Epidemiol. Santd publ., 33, 387-395 6 See p. 23.

226 PETO IARC (1986) IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Vol. 38, Tobacco Smoking, Lyon Lee, P.N. (1975) Tobacco Consumption in Various Countries (Research Paper 6, Fourth Edition), London, Tobacco Research Council Lee, P.N. (1976) Statistics of Smoking in the United Kingdom (Research Paper 1, Seventh Edition), London, Tobacco Research Council Napalkov, N.P., Tserkovny, G.F., Merabishvili, V.M., Parkin, D.M., Smans, M. & Muir, C., eds (1983) Cancer Incidence in the USSR, 2rid revised ed. (IARC Scientific Publica- tions No. 48), Lyon, International Agency for Research on Cancer Office of Population Censuses and Surveys (1975) Cancer Mortality, England and Wales, 1911-1970 (Studies on Medical and Population Subjects No. 29), London, Her Majesty's Stationery Office Office of Population Censuses and Surveys (1978, 1979, 1980a,b, 1982, 1983, 1984a,b, 1985a) Mortality Statistics, Cause, forthe Years1976, 1977, 1978, 1979, 1980, 1981, 1982, 1983, 1984 (Series DH2, Nos 3-11), London, Her Majesty's Stationery Office Office of Population Censuses and Surveys (1984c) Final Mid-1981 and Revised Mid-1961 to Mid-1980 Population Estimates for England and Wales (OPCS Monitor PP1 84/1 (10 January 1984)), London, Her Majesty's Stationery Office Office of Population Censuses and Surveys (1984d) Mid-1983, Final Mid-1981 and Mid- 1982 Population Estimates for England and Wales (OPCS Monitor PP1 84/3 (8 May 1984)), London, Her Majesty's Stationery Office Office of Population Censuses and Surveys (1985b) Mid-1984 Population Estimates for England and Wales (OPCS Monitor PP1 85/1 (14 May 1985)), London, Her Majesty's Stationery Office Peto, R. & Doll, R. (1984) Keynote address: The control of lung cancer. In: Mizell, M. & Correa, P., eds, Lung Cancer Causes and Prevention, Deerfield Beach, FL, Verlag Chemie International US Surgeon-General (1979) Smoking and Health - A Report of the Surgeon-General (Publication No. PHS 79-50066), Washington DC, US Government Printing Office US Surgeon-General (1982) Smoking and Health: Cancer - A Report of the Surgeon- General (Publication No. DHHS PHS 82-50179), Washington DC, US Government Printing Office US Surgeon-General (1984) The Health Consequences of Smoking: Chronic Obstructive Lung Disease - A Report of the Surgeon-General (Publication No. DHHS PHS 84-50205), Washington DC, US Government Printing Office Wald, N.J. (1985) Smoking. In: Vessey, M.P. & Gray, M.., eds, Cancer Risks and Prevention, Oxford, Oxford University Press, pp. 44-67 Wald, N.J., Doll, R. & Copeland, G. (1981) Trends in tar, nicotine and carbon monoxide levels of UK cigarettes manufactured since 1934. Br. reed. J., 282~ 763-765 Wald, N.J., Idle, M., Boreham, J. & Bailey, A. (1983) Inhaling and lung cancer: an anomaly explained. Br. reed. J., 287, 1273-1275 Warner, K.E. (1978) Possible increases in the under-reporting of cigarette consumption. J. Am. stat. Assoc., 73, 314-318 Waterhouse, J., Muir, C., Correa, P. & Powell, J., eds (1976) Cancer Incidence in Five Continents Vol. III (1ARC Scientific Publications No. 15), Lyon, International Agency for Research on Cancer

I 2063628131

AMERICAN JOURNAL OF Public Health May 1987 Established 1911 Volume 77, Number 5 EDITOR Alfred Yankauer, MD, MPH ASSISTANT EDITOR Kenneth Jo Rothman, DrPH EDITORIAL BOARD Mary F. Arnold, DrPH (1987) Chairperson Doris Bloch, RN, DrPH (1989) Irene H. Butter, PhD (1987) Joy G. Dryfoos, MA (1989) Martin S. F, avero, PhD (1988) Fre~t4c.,k C. Green, MD (1988) Mary Grace Kovar, DrPH, MS (1988) Jean Pakter, MD, MPH (1989) Kenneth D. Rogers, MD,MPH (1987) Zena Stein, MA, MB (1989) Fernando M. Trevino, PhD, MPH (1987) Julian A. Waller, MD (1989) Philip G. Weiler, MD (1988) .Joe David Wray, MD, MPH (1988) William H. McBeafh, MD, MPH Executive Director~Managing Editor . Adrienne Ash, PhD Publications Director • - Doyne Bailey Assistant Ma.naging Editor Marva Barnett Editorial Assistant /- Michelle Horton • Production Editor Less Hazardous Smoking and the Pursuit of Satisfaction Cigarettes deliver drugs; at root smoking is drug taking.) Scientific work has confirmed nicotine as a powerfully reinforcing, psychoactive drug.'- For the consum- er, then, ultra-low-yield cigarettes raise the simple issue of drug "cutting" or adulteration. The unsatisfactoriness of ultra-low-yield cigarettes is seen in the scarcity of customers for these products even in health-conscious California. As noted by Maron and Fortmann in this issue of the Journal,3 only 3.8 per cent of smokers in population-based sample smoke cigarettes in the range of 0.1 to 0.2 mg nicotine, 1 to 2 mg "tar". Despite the publicity about disease risks of smoking and the widespread belief that ultra-low-yield cigarettes are less hazardous, most smokers will not cross the street for these cigarettes, let alone walk the advertised mile. No doubt smokers have routinely tried ultra-low-yield cigarettes and just as routinely have rejected them as unsatisfying. A 1 mg "tar", 0.1 mg nicotine cigarette delivers about 80 per cent diluting air in each puff taken by official smoking machines.4 Those smokers who learn the tricks of compensatory smoking are more likely persist in smoking these cigarettes than are those smokers who do not develop satisfactory compensation techniques. For ultra-low-yield cigarettes, the main "'over- smoking" techniques are blocking the diluting air vents on filters with lips or fingers,5 taking larger puffs, and, as Maron and Fortmanns remind us, simply smoking more cigarettes per day. A smoker self-selection bias (compensators remain, non- compensators leave) may cause much of the discouraging pattern found in and reviewed in the currerit report.3 Experimental Evidence--Recent experiments show a more encouraging picture of what might be gained from the widespread use of ultra-low-yield cigarettes by smokers who refuse to quit smoking. West, et al,6 randomly assigned 14 smokers to remain with their own brand (average 1.3 mg nicotine, 14.2 mg "tar") and 12 smokers to switch to an ultra-low yield brand (0.1 mg nicotine, 1 mg "tar"). Over 10 days of smoking, the ultra-low yield group had plasma nicotine levels that were only 40 per cent of the own brand group (9.4 vs 22.8 ng/ml); carbon monoxide levels differed by 30 per cent (10.5 vs 33.2 ppm). (Although not noted in the Method, smokers were explicitly instructed not to block filter vents JR. West, Personal Communication]). Similarly, in the experimental component of their report, Benowitz, et al,7 found partial compensation in smokers who were forced to smoke ultra-low-yield cigarettes. (Behavioral blocking of filter vents was not forbidden in this study, but may have been discouraged by details of the procedure,s) The combined lesson of the cross-sectional surveys and the forced s~vitching experiments is that, if there is to be maximal progress with the current style ultra-low yield cigarette, its use needs to be encouraged more forcefully. One of the reasons Darlene Dobbs Production~Advertising Assistant I smokers don't put up with ultra-low-yields is that higher yield cigarettes are only an ' . ] easy put ;hast away, ~mag'~ne a chocolate 1 )v .'r f~ ted with : , ' I desert is and s~ ~ch a c~ ndy night seem a tre~ tt, but n a well-s .~. CO,NTRIBU~ING EDITORS ..I pseudoc locoh.tes wo fld g ither dust, espe~ :ia [ly ~' they ~vel " ~i ..~¢~,or, g.e J..p~., J.D: M.PH I the more satisfying ~hocolates. ~:i.~.~..~: ~umic~eattn a.na. tffe ~l~.w~ ..,. ]~.... ~.. If cigarette manufacturers '~ere requi :c( l to make on} "- tlroara ld, KosenKtll~iz, t-11.o : ~,~ " • • " ~ .. ,9 " ~i~ ~-~.~3,~/),,~ ~,~-,t~.-,-~..-~..~ .,-..-,-,~l- ~mg against more problemattc filter ~e ,tgn:, ), these easy purchase away, Imagine a chocolate lover faced with an ersatz chocolate: on a desert island such a candy might seem a treat, but in a weft-stocked candy-store_these pseudochocolates would gather dust, especially if they .were not even cheaper than " ~ ......... If cigarette manufacturers '~e,re required to make only }mg "ta{;' cigarettes ~O ..... against more probleaiattc filter designsg), th~se cigarettes might have a "- : to~dns than they do in the wide open market. Just ~ ~, --- tly~regulaU d, so should be drug yield.Of cig~et.t~ .~ ~.k~ .~hes,: cigarettes,,.b.y ,breaking. off. fd.t.~ .~,~ ~er numbers;.pcx day,, A!!.d s.~.~?~'~N

EDITOR, IAt.~ would roll their own cigarettes: a bothersome practice that might well serve to reduce the number of cigarettes smoked per day. The hope, however, would be that a higher percent- age of smokers would respond as the smokers in the forced- switching experiments; in other words, hopefully, on average the yields to smokers would go down and the number of smokers would go down as well. This decrease in customers and sales would no doubt be "painful" to the tobacco industry. Other responsible indus- tries have been faced with an outright banning of hazardous products, rather than such a compromise wounding of their product. (Indeed, a strong case can be made for a complete - ban on cigarettes!)-Remember that the cigarette industry is not a monopoly in North AmericaJ If any one company knew how to make a cigarette that sold well, but did not kill the customers, it would not hesitate to capture as much of the cigarette market as possible. In such a setting a cigarette that pushes the lower limits of satisfaction is at a serious com- petitive disadvantage--unless all manufacturers are com- pelled to meet the same standard. If a I mg "tar" requirement seems far-fetched, differ- ential taxation according to ultra-low yields (cf.10) and dif- ferential taxation according to pack size (now ranging from 15, 20, 25, and 30 cigarettes per pack in some Canadian markets) may be more practical and may foster less hazard- ous smoking (i.e., less smoking).I~ If medium-nicotine, low- "tar" cigarettes12 do turn out to be less hazardous, these could also be encouraged by similar means. Projected health gains, bower@r, would be at the cost of painfully lower tobacco sales "for. the industry. Other purveyors of toxic substances have had to deal with lower profits and shrinking markets. Should the tobacco industry be spared the restric- tions and regulations directed toward other hazardous chem- icals (e.g., asbestos, vinyl chloride, urea formaldehyde)? Simply recommending ultra-low-yield cigarettes to a nicotine-dependent smoker who is unwilling to stop smoking is like advising a suicidal person with a severe depressive disorder that he or she should "cheer up." Information and instruction are needed for the would-be less-hazardous smok- er to actually reduce smoke exposure. Consumers have had an unrealistic preoccupatiofi with low-yield cigarettes, as if it were unimportant to fix the number of cigarettes smoked. (Imagine a diet in which:one counted only calories per meal and paid no attention to the number of meals per day.) Dose per cigarette does depend more on smoking be- havior than on the standard tar and nicotine yield of com- mercially available cigarettes, but it should be noted that FTC (Federal Trade Commission) yields are not without value.4 The FTC's testing system is intended to estimate per ciga- rette yields to an idealized standard smoker. Maron and Fortmann3 show that, if one holds number of cigarettes smoked constant, the yield categories do relate to smoke exposure. A testing system that estimates doses per cigarette can not be expected to estimate the number of cigarettes the smoker will consume. Avoid Unnecessary Excesses in Smoking--Although key pharmacological rewards of smoking derive from nicotine, nicotine intake is overal/ cruddy regulated during smoking. Smokers avoid aversive high doses of nicotine, and they avoid prevent withdrawal syrup- , :between these "too high":and "too :low":~-doses,~~i drugtaking that occurs in cigarette smoking. If the smoker works with militant non-smokers or is "trying to cut down" (both psychosocial influences), cigarette intake is likely to be close to the "lower boundary" of biologically comfortable use. We found that smokers could decrease their daily nicotine intake (by an average of 33 per cent), by having the number of cigarettes available reduced from an average of 37 to 15 (60 per cent), and yet they reported no difficulty in doing so. 14 When we reduced the cigarette ration further to 10 and 5 cigarettes per day, problems.and suffering did occur (i.e., the lower boundary of intake was breached.) Restrictions and costs (social and financial) can contri- bute to less-hazardous smoking for continuing smokers, in spite ofthemsel'ves. The boundary model argues that reduced risk smoking means getting smokers to smoke as little as they comfortably can: this amount of smoking will necessarily be less than the amount that they would smoke if all envirbn- mental pressures encouraged smoking. Some recommendations for the continuing smoker: • smokers should be told to smoke as few cigarettes per day as possible (count out a daily ration, buy smaller packs). • smokers should try to minimize their dose per ciga- rette (pick the lowest-yield cigarettes [which will be venti- lated], don't block vent holes,5 try to minimize puffs per cigarette [leave longer butts, avoid producing darker "tar stains" on the filters'-~]). • understand that to the extent you don't miss your former higher yield smoke, it has probably not gone away (a "no pain, no gain" rule for less-hazardous smoking). Less hazardous smoking is a deprivation therapy. If the smoker's satisfaction has not diminished, there is good reason to doubt that maximal less-hazardous smoking has been achieved. And, finally, smokers should be told that they and those around them will be much better off if they stop smoking entirely. Limits of Persuasion--Like it or not, there will be ;millions of cigarette smokers for many years to come, despite the substantial progress that has been made in reducing the prevalence of smoking: less hazardous ways of tobacco use are needed.16 H. L. Mencken doubted that the so-called "drink problem" could ever be solved, because humankind could not be rescued from its own "incurable hoggish- hess")7 Solutions to the "cigarette smoke problem" are similarly impaired by a fundamental "'hoggishness." For those who will not stop smoking, less hazardous smpking practices may need to be forced upon the consumer much the way other public health measures (public sanitation services, mandatory immunizations) have been forced on people who will live in modern society. ACKNOWLEDGMENTS The opinions expressed here and those of the author f.~e'not hecessarily those of the Addiction Research Foundation. REFERENCES 1. Koztowski LT: The determinants of tobacco use: cigarette smoking in the context of other forms of tobacco use. Can J Public Health 1982; 73:236-241. 2. Henningfield JE: Behavioral pharmacology of cigarette smoking. In: , "Thompson T, Dews TB, Barrett JE (eds): Advances in Behavioral • ,-l~aarmacology, vol. 4. New York: Academic Press, 1984; 131-210. ~3.-Mar0n DJ Fortmann SP" Nicotine yield and measures of c~garette smoke ~sur¢ m a large population: are lower yield cigarettes safer. Am J ~K .pz:~o..w~.'.LT":~.Physical indicators 0_f actual tar and nicotine yields of ~..In:,Grabowski J, Bell CS (eds): Measurement in the Analysis ~_ ~re.~..t, ment "of Smoking Behavior. 'NIDA Research Monograph 48.

~ Kozlowski LT, Frecker RC, Khouw V. Pope MA: The misuse of • less-hazardous' cigarettes and its detection: hole blocking of ventilated filters. Am J Pub|ic Health 1980; 70:1202-1203. 6 West RJ, Russell MAH, .Iarvis M J, Feyerabend C: Does switching to an ultra-low nicotine cigarette induce nicotine withdrawal effects? Psychopharmacology 1984; 84:120-123. 7. Benowitz NE, Jacob P, Yu L, et al: Reduced tar, nicotine, and carbon monoxide ~xposure while smoking ultra-low- but not low-yield cigarettes. JAMA 1986; 256:241-246. 8. Kozlowski LT: Blocking the filter vents of cigarettes. JAMA 1986: 256:3214.--- 9. Consumer R0pons: The ultra-low-tar gimmick: How to turn a health hazard into a success. Consumer Reports January 1983; 26--27, 10. Harris arE: Public policy issues in the promotion of less hazardous cigarettes. In: Gori GB. Bock FG: A safe cigarette? (Banbury Report 3), Cold Spring Harbor, NY: Cold Spring Harbor, 1980; 333-3,10. II. Kozlowsld LT: Pack-size, reported cigarette smoking rates, and pob[ic health. Am J Public Health 1986; 76:1337-1338. [2. Russell MAH: Are cigardttes getting safer? Br J Addict 1984; 79:241-243. 13. Kozlowsld LT, Herman CP: The interaction of psychological and biolog- ical determinants of tobacco use: more on the boundary model. J App~ Soc EDITORIALS Psychol 1984: 14:244--~6. 14. Benowitz NE, Jacob P, Kozlowski LT, Yu L: Influence of smoking fewer cigarettes on exposure to tar, nicotine, and carbon monoxide. N Engl J Med 1986; 315:1310--1313. 15. Kozlowski LT, Rickert WS, Pope MA. Robinson JC: A color-matching technique for monitoring tar/nicotine yields to smokers. Am J Pubt;c Health 1982; 72:597-599. 16. KozlowskiLT: Less-hazardous tobacco use as a treatment forthe smoking and health problem, ln: Smart ILl, Cappell HD, Glaser F, et al, (eds): Research Advances in Alcohol and Drug Problems. New York, Plenum Publishing, 1984; 8:309--328. 17. Mencken HL: The cult of hope. In: Fan'ell JT (ed): H." L. Mencken "Prejudices: A Selection. New York: Vintage, 1958; 84-89, LYNN T. KozLowsKh PHD Address reprint requests to Lynn T. Koztowski, PhD, H6ad, Behavioral . Research on Tobacco Use, Addiction Reseach Foundation, 33 Russell Street, Toronto, Ontario, Canada M5S 2SI. He is also Associate Professor of Preventive Medicine and Biostatistics, University &-Toronto. ~ • .. ~ 1987 American Journal of Public Health 0090-0036/8751.50 INational Library of Medicine: The World's Link to Medical KnowledgeI The National Library of Medicine was first established in 1836. Located on the campus of the National Institutes of Health in Bethesda, Maryland, it serves as a hub of a national network that includes seven regional libraries, 125 medical school libraries, and 4000 hospital and other medical libraries. It is the world's largest medical research library, with holdings of more than 3.5 million books, jot~m~als, technical reports, theses, pamphlets, photographs, and audiovisual materials covering more .than 4"0 biomedical areas and related subjects. Included is a historical collection of rare medical texts and manuscripts.dating as far back as the eleventh century. With materials in 70 languages and information exchange capabilities internationally, the Library is a worldwide resource for all in the health care professions and related fields. Some 4 million journal articles, books, and computerized information searches were provided by NLM in 1985. The Library's Lister Hill National Center for Bi ~ojnedical Communications investigates ways to apply the most modern computer and communications technology in medical education, biomedical research, and public health care. A fundamental responsibility of the National Library of Medicine is to preserve permanently the content of biomedical literature for the nation. Residual acids in most paper made after the mid-nineteenth, century cause it to deteriorate, representing a major threat to the preservation of biomedical literature published since then. The use of acid-free paper and other permanent, archival materials which are now available could stop much of thepotential future preservation problem at its sourcc. ' To encourage the use of more permanent materials in the publication of biomedical literature, the Board of Regents of the National Library of Medicine sponsored a one-day hearing at the Library's Lister Hill Cemer auditorium on January 27, 1987, to explore the causes of paper deterioration, t.he aesthetics, economics, and availability of acid-free paper, and to foster further positive action in its Use, with emphasis on the use in journals. - "The address of the Library is: National Library of Medicine, Building #38, National Institutes of Health, 8600 Rockville Pike, Bethesda, MD 20209.

2063628135

:~m mm m ~Z 0 0 I11 • F" Z m 9~Ig39B903

2063628~37

© 1988 Elsevier Science Publishers B.V. (Biomedical Division) Smoking and health 1987. M. Aoki-et al, editors 67 TRENDS IN CIGARETTE CONSUMPTION IN THE U.S.A. RAYMOND L. WEISBERG, M.D. St. Mary's Hospital and Medical Center San Francisco, California 94117 Tobacco consumption per capita rose sharply in the United States in the 1940's and 1940's with some dips in reaction to medical reports on the dangers of i smoking. It reached a peak in 1963 before the publication of the first Surgeon General's report, when it dropped, then rose again. Since 1973, tobacco consumption per capita has dropped steadily until in 1985 it was at the lowest figure it had been since 1944.1 It is estimated that more than 40 million smokers in the United States have quit smoking. The percent cigarette smokers among men and women age 20 or over dropped from 1965 to 1985. 52% of men were smoking cigarettes in 1965. The percentage has decreased steadily to 32% in 1985. The percentage of smokers among women has 2 also dropped -- though not as rapidly -- from 34% in 1965 to 28% in 1985. A recent report from the Office of Smoking and Health puts the 1986 figures at 3 29.5% for men and 23.8% for women. The data for this paper comes from 2 sources, the two large prospective studies conducted by the American Cancer Society with the help of volunteers. The first of these is called Cancer Prevention Study I, conducted between 1959 and 1972. In this study 68,000 volunteers in 25 states enrolled 1,078,000 persons over the age of 30, who filled out a detailed questionnaire including questions on smoking. These persons were followed over ~ 13-year period with 98.4% traced (dead or alive). The surviving subjects also completed questionnaires, including questions on smoking, 1961, 1963, 1965 and 4 1972. In this study we examined mortality rates in men and women who smoked high, medium, and low tar yield cigarettes. In 1959, a cigarette classified as high tar had 25.8 mg. of tar and 2.0 mg~ or more of nidotine. A low tar yield cigarette in those days had 17.6 mg. of tar or less and I.i mg. or less of nicotine. Such a cigarette would be classified as a medium tar or even a high tar cigarette today. All cigarettes not meeting the above definitions 5 were classified as medium tar and nicotine cigarettes. Two time periods were studies: 1960-1966 and 1966-1972. Subjects who smoked high, medium and low tar yield cigarettes were matched on age and eight other variables. In each of the 2 study periods, total mortality for both men and women was highest in the high tar yield smokers, next_highest in the medium T/N smokers and lowest in the low T/N smokers. Low tar/nicotine smokers had

previoNslx; while 31.7% of those who smoked at the same T/N level, and 31.5% of:-~ 7 those who decreased their T/N level smoked more cigarettes than previously. Forty one-percent of women who switched to decreased T/N cigarettes over the 13-year period increased the number of cigarettes smoked per day. 37% of those who increased T/N and 39% of those who smoked the cigarettes at the same T/N level increased the number of cigarettes smoked. Another important aspect of smoking lower T/N cigarettes is that it helps smokers quit. Smokers at the start of the study in 1959 who said they quit smoking on the 1965 questionnaire were classified by the tar/nicotine level of the cigarettes they smoked in 1959. Thirty one percent of low T/N smokers quit; 26% of medium and 24% of high tar yield smokers quit.8 The same pattern was found in those who reported they quit smoking in 1972 compared to the tar yields of cigarettes smoked in 1965. A higher percentage of men and women quit in the last 6 years of the study than in the first six years, but again, a higher percent of low T/N smokers quit (41%) than medium (36%) or high (35%) T/N smokers.8 Preliminary data is now available from the second Cancer prevention Study of the American Cancer Society (CPS II). This study started in 1982 with 77,000 volunteers enrolling 1,200,000 subjects. As in the first study, it included men and women 30 years or older. Two follow-ups have now been completed and a third is planned for 1988. The volunteers report deaths of the people they enrolled, and death certificates are obtained from state health departments. Verification of primary site of cancer is made through writing to cancer 9 registries, physicians and hospitals. Smoking habits in CPS II w~re compared to CPS I data collected 23 years earlier. Vast changes in smoking could be seen. 0nly ½ as many men in CPS II currently smoked as in CPS I, and the percent of ex-smokers more than doubled. A greater percentage of women in CPS II had ever smoked when compared to CPS I. But the percent of current smokers was less than in CPS I, and the percent who quit was 4 times the percentage in CPS 1.9 - Another major change was the shift to lower tar/nicotine cigarettes. As can be seen in Table i, 25% of the men and 33% of the women smoked cigarettes with less than i0 mg. tar. Most of the cigarettes with 20 or more mg. tar were non- filter cigarettes and 16% of men and only 6% of women smoked these cigarettes.

- 69 TABLE I TAR CONTENT OF BRANDS SMOKED, CANCER PREVENTION STUDY II, 19829 Tar Category (rag) Male (%) Female (%) <5 8.2 14.5 5-9 16.9 18.6 10-15 14.9 46.3 16-19 17.9 14.4 20+ 16.1 6.2 Total i00.0 i00.0 Number of Subjects 80,230 119,918 An analysis was made of cigarette consumption in relation to the tar content the cigarettes smoked. More ultra-low tar cigarette smokers (cigarettes containing less than 5 mg.) smoked 40 or more cigarettes a day than other categories among both men and women but not enough to fully compensate for the lower tar levels. In men and women who smoked cigarettes with 20 mg. tar or more, there was a higher percentage who smoked 20 or more cigarettes a day than 9 in those who smoked cigarettes with less than 20 mg. tar. The same was not true for degree of inhalation. Ultra low and low tar smokers were less likely to inhale deeply than smokers of higher yield : cigarettes and the deepest inhalers where those who smoked cigarettes with 20 or more mg. tar.9 There is some evidence that lung cancer rates are decreasing in the United States. Lung cancer rates are decreasing for white men under the age of 55; and are leveling off in meninges 55-74. It is only in those 75 and over that the rates continue to rise. The National Cancer Institute has announced a 4% decrease in the incidence of lung cancer in white men. In women however, the lung cancer rate continues to increase. It still remains to be seen if smokers of ultra low tar cigarettes have lower lung cancer rates than was previously observed in what we called low tar cigarette smokers in the 1960's. Presumably this should occur. Current smokers of ultra low tar cigarettes inhale much less than did smokers of low tar cigarettes in CPS I. They cannot compensate fully by smoking more cigarettes. But there are unknown additives in these cigarettes, and we don't know the effects of such additives. The American Cancer Society's CPS II study will offer some evidence on this subject, but we will have to await completion of additional follow-up before data will be available. At the beginning of this presentation~ I reviewed the decrease in tobacco consumption in the United States as well as the decrease in the percentage of

?o moy%a%i~.._rates an. average of 16% lower than high T/N smokers.5 Mortality rates for ~coronary heart disease by tar/nicotine levels also showed that in eachof the two periods, for men and women, the coronary heart disease rate for smokers of low tar yield cigarettes was lower than for high tar yield cigarettes, an average of 14%. But the downward trend from high to low tar 5 yield smokers was not consistent in each group. Lung cancer mortality rates were even lower in low tar yield cigarette smokers compared to medium and high tar yield smokers. In men mortality rates in low tar yield cigarettes were 83% and 79% of the high tar yield smokers in the two 5 time periods. In women they were 57% and 62%. The question arises whether the number of cigarettes smoked had a greater effect than tar yield. Lung cancer mortality rates for persons who smoked 20 or more a day lo___~_w yield cigaKettes were compared to smokers of less than 20 a day high yield cigarettes._ The smokers of <20 cigarettes a day high yield cigarettes had far lower mortality rates than the smokers of 20 or more low yield cigarettes - 57% and 73% in men, and 57% and 33% in women in the two study periods.5 Persons who never smoked had far lower lung cancer death rates than cigarette smokers who smoke low T/N cigarettes. Among men the rate was only 9% that of the cigarette smokers. Among women it was about 32% as high.5 Several investigators have shown that persons who switch to cigarettes with lower tar and nicotine tend to "compensate" by smoking more cigarettes a day or inhaling more deeply. In a study by Benowitz and cow_orkers, two groups of smokers were identified and asked, under laboratory conditions, to smoketheir usual cigarette, a "high" yield cigarette; and then, in the first group, a "low" yield cigarette with 4.6 mg. tar, 0.4 mg. nicotine; and in Group 2, an ultra- low tar yield cigarette with 0.8 mgo tar, 0.i mg. nicotine. In both groups smokers of low or ultra low yield cigarettes smoked more cigarettes a day than when they smoked their usual cigarettes - the high yield cigarettes. The level of nicotine in their blood was about the same whether smoking their usual cigarette or the high or low tar yield cigarettes, probably because they compensated by inhaling more deeply and smoking more cigarettes. In the ultra low tar yield smokers, the nicotine level was cut nearly in half, which means if there were compensation, it was not enough to compensate for lower nicotine 6 yields. Over a long period of time compensation in the form of smoking more cigarettes was not the same for most smokers. Over the 13-year period of CPS I, men who switched to cigarettes with lower T/N levels did not tend to increase the number of cigarettes smoked per day. Twenty nine percent of those who increased their T/N level smoked more cigarettes than they did 13 years

71 bowed ease ield .okers low two 20 or day ette of th y or eir "low" ca- Lan ~vel adult smokers. I would like t~Q_c~Qmme~t_o.n_~hat_ma~ be-iactoms__contributing to these decreases. There are approximately 40~iillib~~k-smo~ers in the United States today. These individuals have responded to the increasing knowledge and awareness of the health risks of tobacco use to smokers. They have received encouragement to stop smoking and many have been helped through various programs designed to support and aid the smoker to become a non-smoker. Additionally, Americans generally have become more health conscious and more involved in personal "fitness. Studies have indicated that in the United States a majority of those who smoke begin the use of tobacco prior to the age of 21. The factors involved in adolescent behavior resulting in the use of cigarettes have been described as: i) peer pressure; 2) adopting adult behavior; 3) experimenting or risk taking; 4) role modeling. Studies have shown higher rates of smoking among adolescents whose parents smoke than among those whose parents are non-smokers. There has been an increasing effort to educate adolescents about the risk of tobacco use through school programs that deal with substance abuse, including tobacco. Additionally educational programs have been focused on grades one through six to reach children before a decision to start using tobacco may have occurred. The use of advertising by the tobacco companies to make cigarette use appear attractive to adolescents has been challenged and the kind of advertising used has been exposed to adolescents to reinforce their independence and ability to say no to tobacco use. Over the past ten years an increasingly larger percentage of adults have accepted the idea that smoking and exposure to second-hand smoke is harmful to an individual's health. 79% of Americans, including 76% of smokers, now believe that workplace smoking should be restricted to designated areas. 75% believe smokers should not smoke in the presence of others. We have seen an increasing number of laws at the city, state and federal level which restrict smoking in the workplace and indoor public places. We have gone from a normative smoking society to a non-smoking society. This has finally received significant attention in the media and has helped to further educate people. And there has been leadership from Surgeon General C. Everett Koop, the highest federal official, whose primary concern is health. Clearly, tobacco use has become socially less acceptable as adult behavior, and this has not gone unnoticed by adolescents. I would suggest to you that: i) The increased awareness of health risk of tobacco to tobacco users as well as non users;

?2 2)__Th~-soc±al acceptance of restricting tobacco use at work and in public places to protect the health of non-smokers; 3) Th~d~cational efforts to make children and adolescents aware of the negative social and health aspects of tobacco; 4) The decreased social acceptance of tobacco.use; and 5) The direct confrontation, in an organized manner, by the voluntary health agencies, as well as the non-smokers rights groups, to the tobacco industry, have all contributed significantly to the decrease in tobacco consumption in the United States since 1973. There is every reason to assume that tobacco consumption in the United States will continue to drop, and there is increasing reason to believe that the goal of a smoke free society by the year 2000 will be achieved. REFERENCES i) Department of Epidemiology and Statistics, 1987. 2) National Center for Health Statistics, 1980-85. 3) Cigarette smoking in the United States, 1986. Morbidity and Mortality Weekly Report. 36: 581-585, 1987. 4) Garfinkel L: Selection, follow-up, and analysis in the American Cancer Society prospective studies. NCI Monogr 67: 49-52, 1985. = 5) Hammond EC, Garfinkel L, Seidman H, Lew EA: "Tar" and nicotine content of cigarette smoke in relation to death rates. Environ Res 12:263-274, 1976. 6) Benowitz NL, Jacob P, Yu L, Talcott R, et al: Reduced tar, nicotine, and carbon monoxide exposure while smoking ultralow - but not low-yield cigarettes. JAMA 256: 241- 246, 1986. 7) Garfinkel L: Changes in number of cigarettes smoked compared to changes in tar and nicotine content over a 13-year period. In: Banbury Report 3: A Safe Cigarette? Cold Spring Harbor Laboratory, NY, pp. 19-28, 1980. 8) Hammond EC: The long-term benefits of reducing tar and nicotine in cigarettes. In: Banbury Report 3: A Safe Cigarette? Cold Spring Harbor Laboratory, N.Y. pp. 13-18, 1980. Stellman SD, Garfinkel L: Smoking habits and tar levels in a new American Cancer Society prospective study of 1.2 million men and women. JNCI 76: 1057-63, 1986. Department of Epidemiology and Statistics, 1987.

2063628144

~CE .ondon 311ection ~,nalysis Surveillance in Health and Disease Edited by W. J. EYLENBOSCH- Professor of Epidemiology and Community Medicine, University of Antwerp and N. D. NOAH Consultant Epidemiologist, PHLS Communicable Disease Surveillance Centre, London Published on behalf of the Commission of the European Communities by Oxford University Press Dissemination University of North ~i~h Sciences Library OXFORD NEW YORK TOKYO OXFORD UNIVERSITY PRESS 1988 o o.I

............ ........ II ,," . i liIll I I I lllllI]lill ten { 1985). ~cd strategy to • North Karelia ages I and 70: =pidemioL 6(2) '~',, and benefits 13 Surveillance of cancer D. PARKIN Introduction Surveillance systems of cancer have been mainly used for the development of aetiological hypotheses by the study of the occurrence of disease in different areas, subgroups of the population and over time. The planning of health services requires information on the size and distri- bution of the cancer problem and also on the effectiveness of preventive or therapeutic services. Because for most cancers there is a long and variable in- terval between exposure to carcinogens and clinical onset of disease, surveil- lance systems have only limited usefulness in the identification of discrete environmental hazards. In this chapter we review the different types of data on cancer morbidity and mortality, and their availability in the countries of the European Econ- omic Community (EEC). The uses to which such data have been put are also discussed. Measurement: Cancer Data Statistics on cancer occurrence The single most useful measure for cancer surveillance is the incidence rate, which is an indicator of the risk of disease, and can be used to estimate ser- vice need. Incidence defines the load of new cases arising per unit time, and hence priorities and requirements for treatment facilities. Measurement of incidence requires the identification of all new cases of disease in a defined population, and hence some kind of case-finding mechanism, and record- linkage to ensure that persons are not confused with events. The cancer registry is the usual method of collecting such data. However, cancer regis- tration is a relatively recent development, the oldest functioning registries having been in existence for at most fifty years.~ For surveillance purposes mortality rates have been more widely used. since these have been available for a much longer period, and usually for larger populations. In these cir- cumstances, the mortality rate is usually being used as a proxy measure of incidence. Death rates will be very similar to incidence for cancers with poor survival (for example, oesophagus, stomach, liver, lung). For cancers with a more favourable prognosis such as colon and breast, the incidence rate will

144 Surveillance in Health and Disease be considerably higher than mortality and any inferences which are made from the latter about variation in incidence of disease assume that survival rates are reasonably constant. However, when survival rates are changing substantially as a result of improved methods of treatment (for example, Hodgkin's disease, childhood leukaemia), then mortality rates will be a poor guide to incidence. Mortality, rather than incidence, is the appropriate measure in certain circumstances, however, particularly when the objective of surveillance is to estimate the effectiveness of treatment or early detection programmes. Mortalio, statistics Death rates are widely available as a result of the intro- duction of legislation requiring that the fact and cause of death in a commun- ity be certified, usually by a medical practitioner. The International Classification of Diseases (ICD) provides a uniform system of nomenclature and coding, and a recommended format for the death certificate. When mor- tality statistics are presented it is the underlying cause of death which forms the axis of classification: this may not equate with the presence of a particular tumour. Although the ICD contains a carefully defined set of rules and guidelines which allow underlying cause to be selected in a uniform manner, interpretation of the concept probably varies considerably: for example, when death occurs from pneumonia in a person previously diagnosed as having cancer. Comprehensive mortality statistics thus require that good diagnostic data are available on decedents, which are transferred in a logical, standardized fashion to death certificates which are then accurately and con- sistently coded, compiled, and analysed. There have been many studies of the validity of cause of death statements in vital statistics data. The methods used involved the comparison of the cause of death entered on the death certificate, with a reference diagnosis de- rived from autopsy reports,z detailed clinical records,3,4 or cancer registry data.s Such studies reveal that the degree of accuracy of the stated cause of death declines as the degree of precision in the diagnosis increases. Thus, although the total number of deaths from cancer of all types may be only slightly underestimated, the distribution by site of cancer may be incorrect. There is a tendency to over-record non-specific diagnoses instead of the cor- rect location (for example, large intestine instead of rectum), and accuracy is sometimes lower in those dying at older ages, or at home. The study of Puffer and Wynne-Griffith3 in twelve separate cities' (ten in Latin America plus Bristol and San Francisco) showed that much of the apparent difference in mortality rates between them could be explained by variation in certification practices. This study took no account of differences which would have been introduced if the certificates had also been coded in each centre. Percy and Dolman6 studied this by sending I246 death certifi- cates from the Third National Cancer Survey in the United States for coding of ~cause of d parisons at t~ coded the san Large variatit ences in the a iously affect c There are I mortality. A logical, and e will mean th: gin of the tm larly to canc( pancreatic ca trend studies successive re~ shows the re~ number of ca rules to ens~ category. The great age, and thei analyses of n Geneva on c national pop tion are requ are presentec Table 13.1 ICD-9) Year ICD-8 1977 1978 ICD-9 1979 1980 % Change 19" Source: Perc3--U 5 *Short tist: "A" li,q "l'na = not applica~

ch are made that survival are changing for example. rill be a poor appropriate the objective ~rl.~ detection t of the intro- n a commun- International ,omenclature e. When mor- : which forms ~fa particular of rules and form manner~ for example, diagnosed as ire ~1~ good ::d i.~.gical, ~tely and con- ~th statements ,arison of the : diagnosis de- :ancer registry ;tared cause of creases. Thus, s may be only y be incorrect. ~ad of the cor- nd accuracy is e cities (ten in much of the explained by of differences )een coded in death certifi- Ies for coding Surveillance of cancer 145 of 'cause of death" in seven countries in Europe and North America. Com- parisons at the three-digit level of the ICD showed that all seven centres coded the same underlying cause of death for only 53 per cent of certificates. Large variations existed for individual sites and it was concluded that differ- ences in the application of rules for selecting underlying cause of death ser- iously affect cancer mortality statistics. There are further potential sources of bias in the study of time trends in mortality. A wider availability of diagnostic resources (biochemical, radio- lo~ical, and endoscopic) or an increase in hospitalization of elderly patients will mean that fewer diagnoses of cancer are missed, and that the sites of ori- gin of the tumours are determined more accurately. This will apply particu- larly to cancers which are difficult to diagnose on clinical grounds alone (e.g. pancreatic cancer, lung cancer). In addition, for certain sites of cancer, time trend studies are complicated by changes in the classification system with successive revisions of the International Classification of Diseases. Table 13.1 shows the result of changes introduced with the ninth revision in 1978 on the number of cancers certified as 'bone" (ICD 170)--the new revision specified rules to ensure that 'metastatic bone cancer' was not included in this category. The great advantage of mortality statistics is their comprehensive cover- age, and their availability. All national vital statistics departments produce analyses of mortality rates by cause; in addition, the data are sent to WHO in Geneva on computer tape or by questionnaire. They relate only to entire national populations, however, and if detail on subdivisions of the popula- tion are required, national sources must be consulted. These mortality data are presented in tabular form in the World Health Statistics Annual. In addi- Table 13.1 Mortality from bone cancer: effect of ICD revision (ICD-8 vs. 1CD-9) No. of deaths certified to ICD-170 Year USA England & Wales ICD-8 1977 172i 417 1978 1737 465 ICD-9 1979 1190 316 1980 293 % Change 1978-1979 -32% -32°/3 Source: Percy--Unpublished data. 5 3-digit 1955-1982 *Short list: "A" list of ICD 6. 7.8. Basic Tabulation List of ICD-9 3-digit: ICD 7-9. ~'na = not applicable.

146 Surveillance in Health and Disease tion, in recent years WHO has set up a system of standardized computer tapes which allow extraction of the data in more detail or in different formats. There are three files relevant to cancer: a general mortality file (deaths by age, sex. and cause), a special cancer mortality file (deaths by age. sex, and detailed site), and a denominator file on population by sex and age. The availability (mid-1985) of data from the countries of the EEC is shown in Table 13.2. For the majority, there is information by three-digit ICD categor- ies since 1955, the seventh revision being in use until 1968 (1971 in Luxem- burg and Portugal). and since 1979 the 9th revision has become standard. The table also shows two indicators which are often used to judge quality of data. It is known, for example, that about one-fifth of the deaths coded to 'Senility and ill defined conditions' in Europe are likely to be due to cancer,v Unfortunately, it is not possible to adjust the cancer deaths by reassigning deaths from the senility category, since it is known that many countries already perform routine, mechanical 'corrections" to reduce the numbers in this group. Similarly, the number of cancer deaths without specification of site of primary, or where it is given in only vague terms, is shown. In general, as far as can be judged from such figures, quality of data is fairly good. Morbidity statistics A direct estimate of cancer incidence must be obtained by collecting data on cases of cancer in the population. Statistics based on utilization of health services (for example, clinic attendances or hospital dis- charges) are often available but since these are event-based it is usually not possible to relate them to incidence rates. Incidence rates of cancer are de- rived from population-based cancer registrations which involves the collec- tion of information on all new cases of cancer in a defined population. The first such registry in Europe which is still functioning, the Danish Cancer Registry, was founded in 1942 and there has been a steady growth in the number of cancer registries, and in the population covered, since that time. A registry must collect information on cancer cases from diverse sources, and link together the documents pertaining to a single individual (or more correctly a single tumour), so that as far as possible no new case of cancer is missed, and no case is recorded twice. Sources of information may be special notification forms completed by physicians and sent to the registry--in some countries there is a legal basis for this provided by compulsory notification. However, most registries rely in addition, or as an alternative, upon the use of documents completed for other purposes. Hospital discharge abstracts, treatment records (especially from oncology or radiotherapy units), and pathology reports referring to cancer are the most common source docu- ments. Most registries also make arrangements to obtain from vital statistics offices copies of death certificates which mention cancer as a causative or contributory factor. Registries will usually attempt to elicit further informa- tion on cases of cancer which first come to their attention in this way, but in

:d computer in different -~ortality file :ath's by age, sex and age. ~ is shown in CD categor- 1 in Luxem- ne standard. ge quality of ths ~oded to e to cancer.~ • reassigning n.v countries , numbers in _-cihcation of ~. In general, • good. ~ be obtained _ics based on hospital dis- not de- es the collec- ,ulation. The ~nish Cancer rowth in the : that time. ,erse sources, ual (or more :e of cancer is lay be special ;try--in some ' notification. upon the use rge abstracts, v units), and source docu- vital statistics causative or ther informa- is way, but in 147

148 Surveillance in Health and Disease its absence practice varies as to whether to reject such cases, or record them as 'Death Certificate Only" (DCO) cases. The proportion of cases of cancer first coming to the attention of the registry in this way is a useful indicator of completeness of registration.9 If it is high, this implies that recorded inci- dence rates are likely to be too low. If a proportion of fatal cancers come to light only as a result of death certificates, then presumably a corresponding percentage of non-fatal cases is also being missed. Estimates of completeness of registration among non-fatal cases can be produced u~ing information on time of diagnosis and reporting of cases first notified to the registry by death certificate.~° In some countries (for example, France and Belgium) the med- ical part of the death certificate (which includes the diagnostic information) is anonymous, so that it is impossible to systematically link death certificates with registered cancers. A further index of completeness of registration is a comparison of registra- tions with deaths in the same period and population, and for the same cause. Unless incidence is declining at a very rapid rate, then incidence should exceed mortality, the ratio being determined by the survival rate. Some care is needed in interpretation, however, since the registry diagnosis is likely to be more precise than that recorded at death certification (see above). Caution is also necessary when interpreting time trends in cancer incidence from a registry in which the completeness of registration is changing, since better ascertainment will give rise to apparent increases in the absence of any true change. The quality of information recorded at cancer registration is likely to be superior to that at death registration and considerably more information about the patient and his/her tumour (including histological type and extent) can be recorded. A crude index of'quality' of registry data is the calculation, for each site, of the percentage of diagnoses based up6n histological examination--a high figure being taken to imply that precise information was available to allow correct identification of tumour site and histology. However, it should be noted that a high percentage of histologically verified neoplasms might equally imply that case ascertainment by other than pathol- ogy reports was defective. A few studies have attempted to assess accuracy of information recorded in a registry. West~ ~ found, for example, that according to his interpretation of the clinical records, 6-3 per cent of registrations in Wales had been allo- cated an incorrect three-digit ICD code. A review of lung cancer registration in the Swedish Cancer Registry12 found that 1.7 per cent of registrations were incorrect. However, for histological diagnoses there is usually a much larger discrepancy between registered diagnosis and that of the independent reviewer, depending upon the precision of histological type specified, l~-x a Cancer registries record all newly diagnosed cases as 'incident', thus avoid- ing a problem of death certification--namely the decision as to whether a cant betx Hox elde caFlt aim duc cha pro,,, nun scre case the trea riot h rest has wh~ has EE~ obt,, ide~ tra t tire Salt ca r( Sur usel ! OthI Mo! gra:! . fact

,r record them ases of cancer ul indicator of recorded inci- ~ncers come to corresponding I" completeness ~formation on dstry by death ium) the med- nformation) is ath certificates ;on of registra- he same cause. ,idence should ate. Some care • sis is likely to _,ove). Caution :idence from a g. since better ,ce of any true to be .,pe and extent) he calculation, ,n histological ~e information and histology. ,gically verified er than pathol- ation recorded ; interpretation had been allo- :er registration ff registrations lsually a much ae independent eci fied. ~ ~-13 nt', thus avoid- s to whether a Surveillance of cancer 149 cancer has 'caused" a particular death. Some problems of comparisons between countries or due to revisions of ICD are thus much less evident. However, a new difficulty arises in deciding what constitutes an incident case. Prostate cancer, for example, is extremely common in subclinical form in the elderly--in the USA 25 per cent of males autopsied at age 70 have a prostate cancer.~4 If registries record autopsy-detected cancers as incident, then rates will vary according to the autopsy rate. When screening programmes which aim to detect early cancers (for example, breast or colo'n cancer) are intro- duced, the total number of new invasive cancers should not, in theory, change. There is some doubt whether this is true; in the lung cancer screening programme in the Mayo Clinic, screened groups continue to show an excess number of cancers over the unscreened,~s and it has been. suggested that screening brings to light cases of disease which would never have become clinically apparent. In addition, screening permits the diagnosis of in situ cases which should always be reported separately from invasive cancers. In the USA, incidence rates for breast cancer show a rise, whereas mortality is more or less constant.16 Since there is no evidence of improving results of treatment, this may well represent the diagnosis and registration of pre- viously inapparent lesions. Incidence rates derived from cancer registries are also considerably more restricted in availability than mortality. The establishment of cancer registra- tion in Europe has been a very haphazard process, in some countries there has been a (more or less) official policy to support and fund registries, else- where individual initiative of research orientated clinicians and pathologists has often been a major factor. The current status of cancer registration in the EEC countries is summarized in Table 13.3. As an alternative to cancer registration, data on incidence of cancer can be obtained from morbidity surveys. These may be ad hoc studies limited to identifying specific tumours, which are essentially the same as cancer regis- tration, except that the time-scale is limited, and the survey purely retrospec- tive. General morbidity surveys record all cases of disease appearing in a sample of the community, for example, the surveys of morbidity at primary care level undertaken in The Netherlands and Great Britain (see chapters on gurveillance in primary care). The problem with such community level sur- veys is that for comparatively rare causes of morbidity, such as cancer, there are relatively few cases among the large number of contacts recorded by pri- mary care workers, so that the populations studied are too small to yield very useful information. Other statistics Mortality data may be used to evaluate effectiveness of early detection pro- grammes, and decrease in mortality is also the goal of treatment services. In fact, treatment is usually evaluated in terms of survival rates. Computation

Table 13.3 Population-based cancer registries in EEC countries, 1985 REGISTRY* Data Approx. Dala in Notes colleclion population cancer incidence slatted covered in I'tve conlinenls BELGIUM A National Cancer Registry was established in 1983. This is a conlinualion and extension oflhe annual re- ports on canccr prodnced by the Ministry of Heallh/()cnvre Beige dn ('miter since 1947. which are based on claill|s itlild e to the lJnioas Nulionalcs de M nttndil~s 11lye lalge confederalioas of sickl~eSs fnllds). DENMARK D~ni,;h Cuncer Registry 1942 5.1 million Volsl IV since 1953 FRANCE Regislries h~ve also been stnrlc,.I in Tam (Albi). Heraull (Montpellierl. Vaucluse (Avignon) and North B~s-I~.hin (Strnsbourg) 19'75 8831100 Vol. 1V 11975 77) Ard&'he IAnnonay). 'l'he~c is a digestive ttac| cnncer registry in ~'6te d'~)r II)iion) ~nd registries of e|;ild- I)ouhs (Besan,ion) 1976 472111)0 Vol. IV (1977) ht~od callcer ill I orraillc (Nimcy) ill|d Alpcs-('6te d'A/'tlr ( Mnrscille)~ Is~re (Grenoble) 1979 95111100 The Enqut3te Permanente Cancer collects dala on el|sos treated in the 22 Cancer ('enlres ill France. and ('nlvados (Caen) 1981 561) 1]00 publishes regntar st~ltislics, nolably on snrviv~d. These centres cover about 25'!~, of all calncer cases: =ecruil- (nil sites) n'ient patterns wiry by inslitulion and by c~ncer site. FEDERAl. ]here is a spccialised registry for childhood cancec in Mainz, had for bone tumonrs in Heidelberg. REPtJBI.IC OF GERMANY Legal restrictions preserving confidentinlity of medical records have gready hindered extensioa ~ff cancer Itamhurg 1954 1.7 million Vols, 1 IV registralion, (since 1060) Saarland 1966 I.I nlillionVols. I11. IV 11968 77"1 ~ GR[:-I".CI~ No population based rcgislry. An Anntml Slalistical Survey of Cancer has been performed since 1967. organized by the Stalistical Service of the Mmislry of Sncial Service,;. which relics apo~'~ notificnlion of eaoeer cases Ireated in hospitals. ,t~over;Ige rlllher variable by region. IRELAND St)uthern ] amor Registry 1977 511':1000 (Cork) Table 13.3 cont.

IREI.ANI) SoulbernI uul(,r Registry (('ork) 1977 Table 13.3 cont. ITALY hi rcceni years, pllplilltlilm-hli.~..d rcgi~llie~ Pi~mollle (Torino) 1965 I I(ll) OIh'l I]~lhlgllll and =1 iiesle, lu iOllll, il [iilllld ill millhln I I I ",, ,~1 Ihe i~opuhilhqll ;11%. co~cred Lomhardy (Varese) 1974 778000 Vol. IV 1976-77 A childhood rnll~.'er rcgistr)' is prcsenl in I Parma 1976 400 Sicily (Ragusa) 1980 270 LI.JXEM BCJIJ RG NETttERLANI)S A regislry was established iu Rollerdanl in 1982. a~'~d 5 further rcgi'ilries slarled dala collection in 1985 ill SOOZ. Eindhoverl 1955 I million Leiden, Niimegen. t Itrcchl. Lcidcrdorp, and I dburg with lilt' ilia1 o1" providing nali.nal ctwerage. PORTU(iAL Viana do Castelo 1975 200 (ill0 No pnhlishcd report available. SPAI N New registries hiive beeu slatted in Mllrcia and (hlipu/cllll (S~III Sel'laSl iilnJ ~llltl Ior digestive cancers only in Zaragoza 1960 8021100 Vol. III, IV Malhlrca (Palnla). 1967 77 Nawlrru IPlimplonal 1970 484 Ofl0 Viii. IV 1973 77 Tarragona 1980 513 UNITED KINGDOM There is a tuitional nelwork of p~lpulalion hased registries in each region v. hich snbmit abstracts to li cen- England & Wales National Certain regional tral bureau responsible Itlr natioaal cancer incidence slalistics. In England and Wales the Office of Popula- coverage 49.2 million registers e.g. in lion Censuses Surveys collects data frnm I I regional regislrics, lind in Scolland the Information Services 1962 Vol. IV: Division of Ihe Heahh Dcpartmenl receives data frma live regioas. The qualily of dala varies somewhat National 5.1 million Scotland 1973 77 l~lwl.~ll Ihese regional registries (ref. 651. coverage England: 1958 N. Western 1973 77 W. Midlands 1973 76 Mersey 1975-77 S. Metropolitan 1973 77 Oxford 1974-77 Trenl 1974-76 Scotland N. Ireland 1.5 million *Only registries which began data collection prior in 1982 are listed here.

152 Surveillance in Health and Disease of survival depends upon the follow-up of a group of cancer patients, and the calculation of the numbers su~,iving after different intervals of time. The usual method is the actuarial or life table method, and there are different ways of allowing for 'normal" or non-cancer mortality in the followed-up patients. The most familiar is the 'Relative Survival" which computes the observed mortality rate in the cancer patients as a ratio of that expected in the population from which they come.I v Prevalence of cancer is often advanced as a useful measure in cancer sur- veillance, l 8 indicating the number of patients alive who require medical care. However, there is no standard definition of a prevalent case of cancer. In theory, it should refer to someone once diagnosed as having cancer who is still alive, but then long survivors who are 'cured' are included, and it is not a useful measure of need for service. A compromise might be to regard only patients alive between 0 and 5 (say) years after diagnosis as 'prevalent" cancers, but the estimation of this figure would require good data on inci- dence and survival. The use of prevalence to denote cases still receiving treat- ment, or undergoing follow-up clearly has no value as an indicator of service need, since it will be largely dependent upon the availability of facilities and personnel. Uses of cancer statistics Exposure to aetiological factors One of the goals of continuous measurement of cancer occurrence, or risk, is to assess the importance of differences in environment and individual be- haviour in cancer causation. In theory, surveillance may reveal new or unex- pected hazards, but in practice this has rarely happened. Variation in incidence rates over time, identification of regions of high or low incidence, and abnormal rates of disease in occupational or other groups provoke a search for corresponding differences in exposure to possible aetiological agents. Proof of a causative link. however, requires ad hoc individual-based studies. Time trends The continuous evaluation of time trends in incidence and mortality has been advanced as a means of identifying the emergence of new environmental hazards. The problems involved in such surveillance have been well reviewed.19 The first is that, since most individual cancer types are relatively rare (especially if they are defined in terms of histology as well as site), then large populations, long time-periods, or very big changes in risk are needed if the change in incidence is to be statistically significant. As noted above, apparent changes in incidence, especially over long time-periods, can be due to changing diagnostic ability, or coding rules. Conversely, for common cancers, the introduction of a new carcinogenic factor which affects on15 the ~ to graF will, A invo astu by c liver besti the ~ obse fifty Ft conc appa cific. creas there the Fig. Til lung UK ~ and ~ cigar, lung, levek, data over I Geogt and r aetiol value ation practi to url~ aetioh Ar admil:

• er patients, and the :rvals of time. The there are different in the followed-up hich computes the of that expected in sure in cancer sur- quire medical care. case of cancer. In ving cancer who is aded. and it is not a t be to regard only ~osis as 'prevalent' good data on inci- ~till receiving treat- ndicator of service ity of facilities and or risk, is tnd individual be- .~veal new or unex- ned. Variation in or low incidence, groups provoke a ssible aetiological ,.. individual-based Surveillance of cancer 153 only limited subgroups of the population will lead to only a small increase in the overall incidence in the entire population. Yet if surveillance is extended to the frequent examination of changes in incidence or mortality by geo- graphic area. sex, age-group, etc, then many statistically 'significant' changes will emerge by chance. Although passive surveillance is most likely to be successful when changes involve rare cancers, in fact most such °epidemies" have been identified by astute clinicians, and the reality and magnitude of the increase later validated by cancer registration. Examples are the occurrence of angiosarcoma of the liver in vinyl-chloride workers,-'° and mesothelioma in persons exposed to as- bestos,z~ The clinical suspicion of the role of post-menopausal oestrogens in the aetiology of endometrial cancer was strongly supported by concurrent observations of a very rapid increase in incidence in white females aged over fifty in the early years of the 1970s.zz For the epidemiologist, time-trend data more usually present hypotheses concerning aetiology which require testing by other means. Examples are the apparent rapid increases in certain tumours which may be confined to spe- cific age-groups such as testicular cancer,23'2'* or represent a generalized in- crease in successive birth cohorts, as in malignant melanoma,as Sometimes there is no evident explanation even for quite marked time changes, such as the progressive decline in stomach cancer, observed everywhere (see Fig. 13.1). Time-trend data may also help to substantiate aetiology. Recent trends in lung cancer mortality show declines amongst males in early middle age in UK and Finland which can only be explained by changes in cigarette design and tar yields in addition to consumption patterns. In these two countries cigarette smoking by young adults became established so long ago that their lung cancer rates had stabilized by the late 1950s, i.e. before reductions in tar levels began,z° These studies show how important it is to study time trend data in different birth cohorts; however, this requires that data be available over long time-periods--as a minimum for fifteen years. in incidence and emergence of new surveillance have d cancer types are stology as well as ig cl~,anges in risk ~nificant. As noted time-periods, can • Conversely, for ctor which affects Geographic comparisons The study of geographic differences in incidence and mortality from cancer have been very important in the generation of aetiological leads. Although international comparisons are of enormous value in this respect, cancer surveillance is usually thought of in terms of vari- ation in disease frequency within the same country. It has become standard practice to examine variations in mortality by administrative areas in order to uncover the existence of particular environmental factors important in aetiology, for example, dietary items, pollutants, radiation. A particularly effective way of presenting data on cancer occurrence in administrative subunits of the population is the Cancer Atlas. Atlases show-

154 Surveillance in Health and Disease EUR 10 40 1950 60 70 80 GERMANY 60 40 20 0 1950 60 70 80 FRANCE 8O 40 1950 60 70 80 BELGIUM 60 4O 1950 60 70 80 IRELAND t_._.~L__.l.~l_ 1950 60 70 80 SPAIN 60 40 " ~ 1950 60 70 80 JAPAN 0L~ 1950 60 70 80 ITALY 80 0 195O 60 70 80 THE NETHERLANDS 60 40 "-~..~. ~ .0 t-t~t__ L .t_ 1950 60 70 80 LUXEMBOURG 60 40 1950 60 70 80 UNITED KINGDOM 40 20 0 t'- t~___t____.~t .1_. 1950 60 70 80 DENMARK 60 60 40 40 20 ~ 20 0 L L-.--._--L a__. 0 ~_ 1950 60 70 80 1950 GREECE L 1 J-- 60 70 80 PORTUGAL 60 40 .~.~ 20 ~ 0 t l t ,[_ 1950 60 70 80 males ---- -- females USA 60 40 2O 0 ~ 1950 60 ~0 Fig. I3.1. Stomach cancer death rates over time. X-axis=calendar year: }'-axis= age-adjusted (world standard) mortality per I0s. ing vari Belgiun Spain.,~ prepare maps i~ tions, n which r (for ex~ NevertL study a tire tra~ rence3: publica" especial Refin marked comm u suggest~ techniq graphic units or is usual graphic only ge standar cal fact closely : percent, cerned. cancer t with va~ ity or in tigadon suggest~ ally in d vincing. between Englanc The e localitie tering e aetiolog variety PO 0 O~

FRANCE 50 60 70 80 BELGIUM 50 60 70 80 IRELAND 70 80 SPAIN ) 60 70 80 JAPAN 60 70 80 d~r year; Y-axis= Surveillance of cancer 155 ing variation in mortality rates from different cancers have been produced for Belgium?~ Federal Republic of Germany,28 Italy29 The Netherlands,3° Spain?I and England and Wales.3z If data are available, similar maps can be prepared using incidence rates, as has been done for Scotland.~3 Cancer maps inevitably produce a large number of apparently significant associa- tions, man3' of which are due to chance. Other associations shown are those which might have been anticipated from known distribution of risk factors (for example, higher incidence of smoking related cancers in urban areas). Nevertheless, there are often sufficient interesting leads to prompt further study at a local level. In Belgium. the differences in mortality rates of diges- tive tract cancers3~ were later confirmed by differences in frequency of occur- rence3-' which could in part be related to different dietary patterns. The publication of cancer atlases also generates a great deal of public interest, especially from those living in areas of apparently increased risk! Refining the unit of analysis to very small local areas shows that very marked variations in incidence and mortality can exist, as between adjacent communes for carcinoma of the oesophagus in Brittany (Fig. 13.2) and this suggests the presence of important environmental factors. A widely used technique is to correlate disease rates and various socio-economic, demo- graphic, or environmental variables, using for analysis small geographic units or aggregates with similar characteristics. The limitation of such studies is usually the non-availability of data on exposure of interest for the geo- graphic units concerned. Often, as in the ecological analysis in Finland,3~ only generalized associations with cancer risk such as urbanization and standard of living emerge, and it is not possible to define the precise aetiologi- cal factors. This is not surprising in the absence of an exposure which is closely associated with location. Even for industrial exposures, unless a large percentage of the population of the areas is involved with the industry con- cerned, quite large relative risks will be missed.37 Nevertheless, the study of cancer mortality at county level in the USA has lead to clear correlations with various industries,~8, 39. and in Italy suspicion of excess cancer mortal- ity or incidence in some small but highly industrialized cities led to the inves- tigation of possible chemical exposures.*° Although several studies have suggested a link between gastric cancer and environmental nitrate, specific- ally in drinking water, the evidence that such an association is real is uncon- vincing.*~ Chilvers and Conway*z showed that there was no association between cancer risk and the concentration of fluoride in drinking water in England and Wales. The evaluation of cancer occurrence by small areas frequently reveals localities with excess risk, and investigation may uncover an apparent clus- tering of cases in space and time, suggestive particularly of an infectious aetiology. Most interest has been in the leukaemias and lymphomas, and a variety of statistical methods proposed to identify statistically significant

6~9~90~ Ii Age-standardized annual mortality rates per 100 000 males 1958-1966 0 10 20 30 40kin

Surveillance of cancer 157 aggregates;43 however, only for Burkitt's lymphoma is there sufficiently con- vincing evidence of such clustering.*~ Personal variables Several personal variables are recorded sufficiently fre- quently in routine statistical sources as to permit examination of their asso- ciation with the occurrence of cancer. Occupation has commanded most attention, since it presumably relates to a set of relatively clearly defined environmental exposures, The recording of information concerning occupation on death certificates, and on census schedules, allow the calculation of mortality rate (or proportional mortality ratios) for different occupations and cancer sites. The most extensive analyses of this kind have been in England and Wales: Logan4s reviewed the decen- nial occupational analyses between 1851 and 197l for cancer mortality, and at the some time presented comparable international data. There are several problems involved in these analyses, particularly in relation to the accuracy with which the statement of occupation given by the next of kin after death accurately reflects the major occupation during the lifetime of the deceased (who will often have retired, or changed jobs, prior to death); nevertheless, such studies provide a relatively simple method of monitoring potential occu- pational hazards. The possibility for cancer registries to perform such routine surveillance has been suggested, but quality of information on occupation and the statistical problem of multiple comparisons present difficulties.~6 On the other hand, the possibility of identifying all incident cancers, in the ab- sence of selection bias, allows estimation of the aetiologic fraction of given exposures, such as occupation for the general population..7 A different approach is to use routine surveillance systems (cancer registra- tion, death registration) as the case-finding mechanism for prospective cohort studies. If the exposure variable is also available from some routine statistical source, then the exercise of follow-up is essentially one of record linkage. An example of this is the follow-up of 1 per cent sample of workers identified at the 1971 census in England and Wales, using the national cancer registry to study cancer incidence in relation to occupation.'~ However, data on exposure may have to be derived from different sources--sometimes the cohort members provide the necessary information, but other records may be available allowing the exposure groups to be defined on the basis of past ex- perience (for example, the studies of exposure to radiation for ankylosing spondylitis),.9 or the potential carcinogenicity of previous drug therapy,s°-s'- A particular example which uses the cancer registry to both de- fine the cohort and outcome was the study of the effects of radiation in the treatment of cancer of the cervix,s3 In special circumstances, the need for long:term surveillance of populations exposed to high levels of potential car- cinogens has led to the establishment of registers, for example for the 220 000 residents of Seveso. exposed to dioxin in 1976.s~ L

158 Surveillance in Health and D&ease Place of birth is frequently recorded on death certificates and at cancer registration and can be exploited to investigate incidence and mortality rates in relation to migration. Studies of migrants, between or within countries, are of particular interest to epidemiologists in that they provide clues as to the relative importance of genetic factors and environment, and the latent period of environmental "exposures'. in cancer aetiology. Studies of other subgroups which differ in terms of life-style and/or genetic makeup, for example ethnic or religious groups, have been immensely important in the generation of aetiological hypotheses, but such studies are of little potential in Europe where surveillance systems rarely collect the relevant data. Evaluation of cancer control Although statistics on disease occurrence have been widely exploited by epi- demiologists to elucidate causes of cancer, most surveillance systems would justify their existence in terms of their role in the planning and evaluation of health care. Preventive programmes are frequently monitored in terms of reductions in exposure in the target population (for example, the reduction in the preval- ence of cigarette smoking). However, surveillance systems should be able to identify whether the objective of such programmes, i.e. reductions in the inci- dence of cancer, have been achieved. There have been relatively few con- trolled trials of preventive strategies, and those reported have not been designed to detect effects on cancer incidence. Thus most evaluation will re- quire before--after comparisons within the population into which the inter- vention has been introduced. There will usually be a long interval between the institution of preventive measures, and the results in terms of changed incidence. Figure 13.3 shows the results to be expected, based on a computer simulation, from the introduction of a programme to reduce smoking uptake by young people,ss Nevertheless, as described above, the results of reduced prevalence of smoking and toxicity of cigarettes are now evident as a lower risk of lung cancer in young males in Finland and UK. Evaluation of the efficacy of cancer control measures in the occupational setting is more difficult due to the small size of the exposed populations. Simi- larly, slight modifications in the general environment (drinking water, air pollution, background radioactivity), even if they have an impact on cancer occurrence, cannot be detected by routine epidemiological surveillance. Most screening programmes for cancer aim for early diagnosis, and sub- sequent reduction of mortality: an exception, however, is screening for cervix cancer where discovery and treatment of a precursor condition can also pre- vent the onset (incidence) of clinical disease. Although there have never been any controlled trials of cervix-cancer screening programmes, their effective- hess is now reasonably certain, thanks to a large number of descriptive, co- hort, and case control studies (for review see ref. 56). In theory, then, the 110~ 100I 60 40 20 0 1953 Fig. 13. in Fint, simulat in 1976 of the of thos,, 30, 7=2.5: 1976-81 3, and vals). effecti~ lowing rather trends appare the prc lying Bret~ reduci~ mentat rates Europ~ change pretin~ mentic The ally re~

and at cancer mortality rates n countries, are clues as to the ~e latent period ther subgroups example ethnic generation of tial in Europe ploited by epi- systems would ~ evaluation of "reductions in in the preval- uld be able to ns in the inci- ,ely few con- ~ve not been ation will re- s of changed l a computer )king uptake s of reduced ~t as a lower ,ccupational ttions. Simi- g water, air :t on cancer lance. .s, and sub- g for cervix m also pre- never been ir effective- riptive, co- . then, the /103 100 8O 6O Surveillance of cancer 4O 20 2 3 I J 2_. 159 1953 1975 2000 2050 Fig, 13.3. Age-adjusted incidence rates (/10s person-years) for lung cancer in males in Finland 1953-75. and forecasts for the rates in 1980-2050 derived by means of a simulation model with the following assumptions. In each consecutive 5-year period in 1976-2050, 10 per cent of the smokers in each smoking category will stop, and one of the following alternatives holds true: (I) ct% of non-smokers aged 1 0-14 years,/3% of those aged 15-19. and 7" % of those aged 20-24 years will start smoking, ct = 60,/3 = 30, y= I0; (2) Same as (1) but a=30,/3= 15, y=5; (3) Same as (1) but a= 15,/3=7.5, 7" = 2.5; (4) Same as ( 1 ) but the values of a for consecutive 5-year periods starting from 1976-80 are: 30, 24, 18, 12. 6, and 0 (remaining intervals), those for/3 are: 15, 12, 9, 6, 3, and 0 (remaining intervals), and those for 7" are: 5, 4, 3, 2, 1, and 0 (remaining inter- vals). (From Hakulinen and PukkulaSS.) effectiveness of cervical cytology screening could be easily evaluated by fol- lowing trends in incidence (or mortality) in the community. In practice, this is rather more difficult than apparent since there are quite marked underlying trends in incidence which may complicate interpretation--for example, the apparent lack of effect of screening in England and Waless7 is misleading-- the programme has had a moderate success, but in the face of a strong under- lying increase in risk of disease in young women, s8 Breast-cancer screening by mammography also appears to be effective in reducing mortality,s~ The coming years will probably see increased imple- mentation of such programmes, which will require monitoring by mortality rates in the screened areas. The mortality rate from breast cancer in most European countries shows a progressive increase (Fig. 13.4). any apparent change must be judged against this underlying trend; the problem of inter- preting incidence data in the face of active screening programmes has been mentioned. The evaluation of the effectiveness of treatment programmes has tradition- ally relied upon the measurement of survival. Survival rates are the normal

160 Surveillance in Health and Disease \ u 25t- ~ .............. ~~~/? z,'.~/-~,f ~-~ u..."'~" ....... " .... • • ! "" b ~20- . / " ~0 ........ ......~,* 50 55 60 65 70 75 80 85 Fig. 13.4. Breast cancer mortality: Europe 1950-85. Key: b=Belgium; d=Den- mark; e= Ireland; f= France; g= Fed. Republic of Germany; h = Greece; i= Italy; n = The Netherlands; p = Portugal; s = Spain; u = England and Wales. endpoints in clinical trials of cancer therapy, and it seems natural to use population survival to evaluate the results of therapy on a population basis-- that is for all cancers of a particular type. Thus it is possible to examine time trends in survival, as well as differences within various subgroups of the population,s9 However, great care is needed in interpreting the survival rate; it is a composite index which reflects several turnout factors (stage at diagno- sis, histologic type) and patient factors (age, race, socio-economic circum- stances) as well as treatment efficacy. Enstrom and Austin#° question whether, given these problems and the difficulties involved in estimating population-based survival rates (see above), evaluation might not be more easily performed by a simple comparison of incidence and mortality rates. The most impressive improvements in survival have been in a rather limited number of cancers--Wilm's tumour, childhood leukaemia and Hodgkin's disease--whilst for the major solid turnouts of adults (lung cancer, stomach cancer, breast cancer) there appears to have been practically no change, o a Although the treatment and aftercare of cancer patients aim for more than a simple improvement in survival, and various measures have been devised to quantify health, or quality of life,oz it has to be admitted that none is capable of application on a wide enough basis to qualify as surveillance techniques. Planning of services Provision of health-care services should logically depend upon an indication of need, which can be regarded as a measure of the amount or burden of par- ticula~ variot servic; often latent dence betwe~ dence relati~ for set Conclu Once t inform. source statisti exploit health- tries, tl trast tc disease ment ir will inc compo vide da jection: treatm~ epidem in the I: tions is questio systems for this ciently registr?' accordi~ data co measure existing Bibliogr I. Wag

80 85 = Belgium; d = Den- h = Greece; i = Italy; "ales. ms natural to use p~l~ation basis-- ,ic~xamine time subgroups of the g the survival rate; -s (stage at diagno- -economic circum- Austin6° question lved in estimating night not be more ad mortality rates. in a rather limited ~ia and Hodgkin's ~g cancer, stomach ly no change.~1 aim for more than ~ve been devised to aat none is capable ance techniques. apon an indication t or burden of par- Surveillance of cancer 161 ticular conditions, and their amenability to interventive measures. Although various ways in which information on incidence of cancer have been used in service planning have been discussed by Wrighton63 it is not known how often provision is in fact based upon objective criteria. Because of the long latent interval between exposure and clinical disease, trends in cancer inci- dence tend to be relatively stable, with regular increases (or decreases) in risk between successive cohorts. This means that projections of past cancer inci- dence (or mortality) trends to provide estimates of future rates of disease are relatively accurate.64 The use of routine cancer data to predict future need for services should, in theory at least, be a useful planning too/. Conclusions Once the organization for delivery of health care passes a rudimentary level, information for planning and evaluation purposes becomes essential. The sources are usually vital statistics systems, or activity analysis (utilization statistics) from the health care system itself. Both sources have been widely exploited by epidemiologists interested in disease causation, as well as by health-care planners. Statistics on cancer are also available from cancer regis- tries, the existence of which is due in part to the fact that for cancer, in con- trast to many chronic diseases, the definition of onset and existence of the disease state is relatively clear. With the mounting pressure for cost contain- ment in heaIth services, information systems have come under scrutiny, and will increasingly need to justify their role. Surveillance systems form a vital component of all cancer control programmes. At the simplest level they pro- vide data on cases, deaths, life years lost, etc. which together with future pro- jections are important in establishing priority areas for prevention, treatment, and rehabilitation. Surveillance systems are a useful resource in epidemiological research, and a knowledge of aetiology is obviously essential in the planning of prevention strategies. Evaluation of health care interven- tions is ideally carried out by controlled trial designed to answer specific questions. This is rarely feasible, however, and usually routine data-collection systems have to be adapted for this purpose. Their general overall suitability for this must be kept constantly under review, and those which are suffi- ciently flexible to respond to changing demands, for example the cancer registry, should be prepared to adapt their data collection and presentation accordingly. These needs must be set against the frequent" requirement for data covering relatively long time-periods (effectiveness of cancer control measures will rarely be evident in a short interval), so that major changes to existing systems should not be undertaken lightly. Bibliography 1. Wagner G., Cancer registration: Historical aspects. In Parkin, D. M., Wagner, G.

162 Surveillance in Health and Disease and Muir, C. S. The role of the registry in cancer control. IARC Scientific Publica- tions 66, Lyon (1985). 2. Heasman. M. A. and Lipworth, L., Accuracy of certification of cause of death. Studies in Medical and Population Subjects 20. HMSO, London (1966). 3. Puffer, R. R. and Wynne-Griffith, G., Patterns of urban mortality. Scientific Pub- lication t 5 I, Pan American Health Organization, Washington (1967). 4. Bosch, F. X., Garcia, A., Orta, J., Juvanet, J., Camproden, A., and Pumarola, A., The accuracy of medical certifications of canrer deaths and of cancer diagnosis in the municipal area of Barcelona in 1979. Rev. Esp. Oncologia 30, 17-24 (1983). 5. Percy, C., Stanek. E., and Gtoeckner, L., Accuracy of cancer death certificates and its effect on cancer mortality statistics. Amer. J. publ. Hlth 71,242-50 (1981). 6. Percy, C. and Dolman, A., Comparison of the coding of death certificates related to cancer in seven countries. Publ. Hlth Rep. 93, 335-50 (1978). 7. Hansluwka, H., Cancer mortality in Europe, 1970-74. Worm Hlth Stat. Quart. 31, 159-94 (1978). 8. McLennan, R., Muir, C. S., Steinitz, R., and Winker, A., Cancer registration and its techniques. IARC Scientific Publications 2 I. Lyon (1978). 9. Waterhouse, J. A. H., Muir, C. S., Shanmugaratnam, K., and Powell, J., Cancer incidence infive continents, Vol. 4. IARC Scientific Publications 42. Lyon (1982). I0. Benn, R. T., Leck, I., and Nwene, U. P., Estimation of completeness of cancer registration, lnt. J. Epidemiol. 103, 362-7 (1982). I 1. West, R. R., Accuracy of cancer registration. Br. J. prevent, social. Med. 30, 187- 92 (1976). 12. Larsson, S., Completeness and reliability of lung cancer registration in the Swedish Cancer Registry. A cta path. microbiol, scand. 79A, 389-98 (1971). 13. Stalsberg, H., Lymphoreticular tumours in Norway and in other European coun- tries. J. Natl. Cancer Inst. 50, 1685-702 (1973). 14. Breslow, N., Chan, C. W., Dhom, G. et al., Latent carcinoma of prostate at autopsy in seven areas. Int. J. Cancer 20, 680-8 (1977). 15. Fontana, R. S., Early detection of tung cancer: The Mayo lung project. In Screen- ingfor cancer (eds P. C. Prorok and A. B. Miller). UICC Technical Report Ser- vices, Vot. 78, Geneva (1984). 16. Dolt, R. and Peto, R., The causes of cancer. Oxford University Press (1981). 17. American joint committee on cancer staging and end results reporting. Reporting of cancer survival and end results 1982. American Cancer Society, Chicago (1982). 18. Hakama, M., Hakulinen, T., Teppo, L., and Saxen, E., Incidence, mortality and prevalence as indicators of the cancer problem. Cancer 36, 2227-31 (I 975). 19. Muir, C. S., McLennan, R., Waterhouse, J. A. H., and Magnus, K., Feasibility of monitoring populations to detect environmental carcinogens. In Environmental pollution and carcinogenic risks. IARC Scientific Publications 13, INSERM Sym- posia Series 52 (1976). 20. Creech, J. L. and Johnson, M. N., Angiosarcoma in the manufacture ofpolyvinyl chloride. J. occup. Med. 16, 150-1 (1974). 21. Wagner, J. C., Sleggs, C. A., and Marchand, P., Diffuse pleural mesothelioma and asbestos exposure in the north western Cape province. Br. J. indust. Med. 17, 260-71 (1960). 22. Austin, D. F. and Roe. K. M., Increase in cancer of the corpus uteri in the San Francisco~Oakland standard metropolitan statistical area, 1960-1975. J. Natl. Cancer Inst. 62, 13-16 (1979).

Scientific Pub'lica- ,~ o.f cause of death. on (1966), diO'. Scientific Pub- (1967}. • and Pumarola, A., cancer diagnosis in ~0, 17-24 (1983). :r death certificates 71,242-50 (1981). n certificates related I. 'd Hlth Stat. Quart. 'eer registration and Powell, J.. Cancer • 42. Lyon (1982). ,~leteness of cancer cial. Med. 30, 187- ";tration in the 971). ,pean coun- ma of prostate at project. In Screen- mical Report Ser- Press (1981). 9orting. Reporting ', Chicago (1982). ace, mortality and -31 (1975). • K., Feasibility of In Environmental 3, INSERM Sym- cture of polyvinyl ral mesothelioma • indust. Med. 17, s uteri in the San 60-1975. J. Natl. Surveillance of cancer 163 23. Clemmensen, J., Trends and risks Denmark 1943-1972. Acta path. microbiol. stand., Suppl. 261, 1-286 (1977). 24. Davies, J. M., Testicular cancer in England and Wales: Some epidemiological as- pects. Lancet i, 928-31 (1981 ). 25. Muir. C. S. and Nectoux, J.. Time trends in malignant melanoma of the skin. In Trends in cancer incidence (ed. K. Magnus). Hemisphere, Washington (1982). 26. International Agency for Research on Cancer, Tdbacco smoking. IARC mono- graphs, Vol. 38. Lyon (1986). 27. Rijckeboet, R.. Janssens, G., and Thiers, G., Atlas of cancer mortality in Belgium, 1969-1976. Ministry of Public Health and the Family, Brussels (1983). 28. Becket. N., Frentzel-Beyme. R., and Wagner, G., Krebsatlas der Bundesrepublik Deutschland. Springer-Verlag, Berlin (1984). 29. Cislaghi, C.. Decarli, A.. Morosini, P. L., and Puntoni, R., Atlante della mortalitfi per tumori in Italia 1970-1972. Lega Italiana per la Lotta contro i Tumori, Milano (1978). 30. Netherlands Central Bureau of Statistics, Atlas of cancer mortaHO' in the Nether- lands 1969-1978. Dept. of Health Statistics, The Hague (1980). 31. Lopez-Abente. G., Escolar. A.. and Errezda, M. (eds), Atlas del cancer en Espana. Victoria-Gasteiz (1984). 32. Gardner, M. J., Winter, P. D., Taylor, C. P., and Acheson, E. D., Atlas of cancer mortality in England and Wales 1968-1978. John Wiley, Chichester (1983). 33. Kemp, I., Boyle, P., Smans, M., and Muir, C. S., Atlas of cancer in Scotland1975- 1980. Incidence and epidemiological perspectives. IARC Scientific Publications 72 (1985). 34. Ramioul, L. and Tuyns, A. J., La distribution g~ographique des cancers du tube digestif en Belgique. In,gale r~partition de la mortalit6 par cancer de l'estomac et par cancer du rectum. A eta gastro-ent, belg. 40, 129-49 (1977). 35. Tuyns, A. J., Reginster, G. and Ravet, L., Une 6tude ~pid~miologique sur les cancers du tube digestif en Belgique. Etude interuniversitaire: Alimentation et sant& Rev. Med. LiOge 40, X, 1-7 (1985). 36. Teppo, L., Pukkala, E.. Hakama, M., Hakulinen, T., Herva, A., and Saxen, E., Way of life and cancer incidence in Finland. Scand. J. Soc. Med., Suppl. 19, 5-84 (1980). 37. Goldsmith, J. R., Geographical pathology as a method for detecting occupa- tional cancer. J. occup. Med. 19, 533-9 (1977). 38. Hoover, R. and Fraumeni, J. F. Cancer mortality in US counties with chemical industries. Environ. Res. 9, 196-207 (1975). 39. Blot, W. J. and Fraumeni, J. F. Geographic patterns of lung cancer: Industrial correlations. Am. J. Epidemiol. 103, 539-50 (1976). 40. Riboli, E., Bai, E., Berrino, F., and Merisi, A. Mortality from lung cancer in an acetylene and phthalic anhydride plant: A case reference study. Scand. J. Work Environ. Hth 9, 455-62 (1983). 41, Fraser, P. Nitrates: Epidemiological evidence, In Interpretation of negative epi- demiologieal evidence for earcinogeniciO, (eds N. J. Wald and R. Doll). IARC pp. 183-94. Scientific Publications 65, Lyon (1985). 42, Chilvers, C., and Conway. D. Cancer mortality in England in relation to levels of naturally occuring fluoride in water suppiies. J. Epidemiol. Community Hth 39, ~ ~7 (1985),

164 Surveillance in Health and Disease 43. Smith, P. G. Spatial and temporal clustering. In Cancer epidemiology andpreven- tion (eds L. D. Schottenfeld and J. F. Fraumeni). Saunders, Philadelphia (1985). 44. Day. N. E., Smith, P. G., and Lachet, B. The latent period of Burkitt's lym- phoma: The evidence from epidemiological clustering. In Burkitt's lymphoma. A human cancer model (eds G. Lenoir et aL). IARC Scientific Publications 60, Lyon (1985). 45. Logan, W. P. D. Cancer mortality by occupation~, and social class 1851-1971. IARC Scientific Publications 36. Studies on Med. and Pop. Subjects 44. Lyon and HMSO, London (1982). 46. Jensen, O. M. The cancer registry was a tool for detecting industrial cancer risks. In The role of the registry in cancer control (eds D. M. Parkin, G. Wagner, and C. S. Muir). IARC Scientific Publications 66. Lyon (1985). 47. Pastorino, U., Berrino, F., Gervasio, A., Pesenti, V., Riboli, E., and Crosignani, P. Proportion of lung cancers due to occupational exposure. Int. J. Cancer 33, 231-7 (1984). Fox, A. J. The role of OPCS in occupational epidemiology: some examples. Ann. occup. Hyg. 21,393-403 (1979). Court-Brown, W. M. and Doll, R. Mortality from cancer and other causes after radiotherapy for ankylosing spondylitis. Br. reed. J. ii, 1327-32 (1965). Clemmensen, J. and Hjalgrim-Jensen, S. Does phenobarbital cause intracranial tumours? A follow-up through 35 years. Ecotoxicol. Environ. Quality 1, 255-60 (1981). 5l. Reimer, R. R., Hoover, R., Fraumeni, J. F., and Young, R. C. Acute leukaemia after alkylating-agent therapy of ovarian cancer. New Engl. J. Med. 297, 177-81 (1977). 52. Howe, G. R., Lindsay, J., Coppock, E., and Miller, A. B. Izoniazid exposure in relation to cancer incidence and mortality in a cohort of tuberculosis patients. Int. J. EpidemioL 8, 305-12 (1979). 53. Day, N. E. and Boice, J. R. (eds). Second cancer in relation to radiation treatment for cervicalcancer. IARC Scientific Publications 52, Lyon.(1983). 54. Fini, A.. Puntoni, R., Gennaro, V., Rosa, A. M., and Vercelli, M. L'enregistre- ment du cancer dans la zone de Seveso. Proc. VIII rrunion du groupe pour l'Epi- drmiologie et l'enregistrement du cancer dans les pays de langue latine, Raguse, Italy, 12-13 May. Int. Agency for Research on Cancer, Lyon (1983). 55. Hakulinen, T. and Pukkala, E. Future incidence of lung cancer: Forecasts based on hypothetical changes in smoking habits of males. Int. J. EpidemioL 10, 233-40 (1981). 56. Parkin, D. M. and Day, N. E. Evaluating and planning screening programmes. In The role of the registry in cancer control (eds D. M. Parkin, G. Wagner, and C. S. Muir). IARC Scientific Publications 66, Lyon (1985). 57. Editorial. Cancer of the cervix: death by incompetence. Lancet ii, 363-64 (1985). 58. Parkin. D. M.. Nguyen-Dinh. X., and Day, N. E. The impact of screening on the incidence of cervix cancer in England and Wales. Br. J. Obstet. GynaecoL 92, 150- 7 0985). 59. Hanai, A. and Fujimoto, I. Survival rate as an index in cancer control. In The role ofthe registry in cancer control (eds D. M. Parkin, G. Wagner, and C. S. Muir). IARC Scientific Publications 66, Lyon (1985). Enstrom, J. E. and Austin, D. F. Interpreting cancer survival rates. Science 195, 847-51 (1977). 48. 49. 50. 60. 61. t 62. ~' 63." l 64.

:v and preven- phia (1985)• ~urkitt's lym- lymphoma. A ons 60. Lyon ~ 1851-1971. 44. Lyon and cancer risks. Wagner. and d Crosignani. J. Cancer 33, ,amples. Ann. - causes after intracranial ty 1, 255-60 te leukaemia 297, 177-81 in F Int. on treatment L'enregistre- pour l'Epi- ine, Raguse, ecasts based r. 10, 233-40 grammes. In .~r, and C. S. -64 (1985). .~ning on the .~ol. 92, 150- • In The role 2. S. Muir). ~cience 195, Surveillance of cancer 165 61. Page, H. S. and Asire, A. J. Cancer rates and risks, 3rd ed. NIH Publication 85- 691. US Department of Health and Human Services (I 985). 62. Spitzer, W. O., Dobson, A. J., Hail, J. et aL Measuring the quality of life of cancer patients. J. chron. Dis. 34, 585-97 (1981). 63. Wrighton, R. J. Planning services for the cancer patient. In The role of the registry in cancer control(eds D. M. Parkin, G. Wagner, and C. S. Muir). IARC Scientific Publications 66, Lyon (1985). 64. Hakama, M. Projection of cancer incidence: experience and some results in Finland. World Hlth Stat. Quart. 33, 228-40 (1980). 65. OPCS. A report of the advisor), committee on cancer registration. Series MB1 6. London, HMSO (1981). 66. Tuyns, A. J. and Mass6, L. M. F. Mortality from cancer of the oesophagus in Britanny. Int. J. Epidemiol. 2, 241-5 (1973).

2063628169

6. ZARBL ]'], SUKUMAR S, ARTHUR AV, et ~. Direct mutagenesis of Ha-r'as-1 oncogenes by N-nitro-N-methy~urea during initiation of mammary carcinogenesis in rats. Nature 1985;315:382-386. 7. QUINTANILLA M, BROWN K, RAMSDEN M, et al. Carcinogen-specific mutation and amplification of Ha-ras during m. ouse skin carcinogenesis. Nature 1986;322:78-80. 8. FUIITA I, YOSr~IO,~ O, YUASA Y, et al. Ha-ras oncogenes are activated by somatic alterations in human urinary tract mmours. Nature 1984; -~ 309:464-466. 9. FUnTA J, SRIVASTAVA SK, KRAUS MH, et al. Frequency of molecular alterations affecting ras protooneogenes in human urinary tract tumors. Proc Natl Acad Sci USA 1985;82:3849-3853. 10. ITO N, ARAI M, SUGIHARA S, et al. Experimental urinary bladder tumors induced by N-butyl-N-(4-hydroxybutyl)nitrosamine. Gann Monogr Cancer Res 1975;17:367-38 I. 11. CH~,NG EH, FURTH ME, SCOLNICK EM, et al. Tumorigenic transforma- tion of mammalian cells induced by a normal human gene homologous to the oneogene of Harvey murine sarcoma virus. Nature 1982;297: 479-483. 12. Tr~oR A, HORAN HAND P, WUSt~ERLICH D, et al. Monoclonal antibodies define differential ras gene expression in malignant and benign colonic diseases. Nature 1984;311:562-565. 13. FUR?H ME, DAv~S LJ, FLEtmD~L'tS B, et al. Monoclonal antibodies to the p21 products of the transforming gene of Harvey murine sarcoma virus and of the cellular ras gene family. J Virol 1982;43:294-304. 14. SRtVASTAVA SK, YUASA Y, R~VNOLOS SH, et al. Effects of two major activating lesions on the structure and conformation of human ras oncogene products. Proc Natl Acad Sci USA 1985;82:38-42. 15. REYNOLDS SH, STOWERS SJ, MARONPOT RR, et al, Detection and identi- fication of activated oncogenes in spontaneously occurring benign and malignant hepatocellular tumors of the B6C3F 1 mouse. Proc Natl Acad Sci USA 1986;83:33-37. 16. PA~'AGEOgGV. A, DEFEO-JOrq~.S D, ROB~NSO~ P, et al. Saccharomyces cer- vis/ae synthesizes proteins related to the p21 gene product of ras genes found in mammals. Mol Cell Biol 1984;4:23-29. 17. OHUC~II ,N, THOR A, PACE DL, et al. Expression of the 21,000 molecular weight ras protein in a spectrum of benign and malignant human mammary tissues. Cancer Res 1986;46:2511-2519: 18. FUJITA J, NAKh~t~,MA H, OSOl~ H, et al. Frequency of ac~iv~ ras oncogenes in human bladder cancers associated with schistosomiasis. Jpn J Cancer Res (Gann) 1987;78:915-920. 19. "HICKS RM. Nitrosamines as possible etiological agent~" in bilharzi~l bladder cancer. In: Magee PN, ed. Nitrosamines and human cancer. New York: Cold Spring Harbor Laboratory, 1982:455-471. 20. GALLICK GE, KURZROCK R, KLOETZER WS, et al. Expression of p21 ras in fresh primary and metastatic human colorectal tumors. Proc Natl Aead Sci USA 1985;82:1795-1799. 21. VIOLa MV_,. FROMOW~TZ F, ORAREZ S, et al. Ras oncogene p21 expres- sion is increased in premalignant lesions and high grade bladder carci- noma. J Exp Med 1985;161:1213-1218. "- 22. BRows K, QUINTANILLA M, RAMSDEN M, et al. v-ras genes from Har- vey and BALB routine sarcoma viruses can act as initiators of two- stage mouse skin carcinogenesis. Cell 1986;46:447-456. . . Projections of Lung Cancer-Mortality in the United States: 1985-20251 Charles C. Brown, Larry G. Kessler3,4 Lung. cancer has been the leading cause of cancer death in the United States for the larger part of this century. Increases in smoking prevalence from the 1900s through the 1950s have resulted in more than 100,000 deaths annually. Because of the changes dur- ing the last three decades in smoking prevalence, the decreasing tar content of cigarettes, and the increasing popularity of low-tar cigarettes, trends in lung cancer are difficult to predict. This article presents an analysis of smoking and lung cancer data using an age-period-cohort model for projecting lung cancer mortality through the year 2025. The projections are based on the initial parameterization of the model and on prevention objectives related to smoking behavior established by the National Cancer Institute. It is con- duded that the recent trends in lung cancer are unlikely to be affected by changes in cigarette com- position and consumption in the near term, but in- creasing the effectiveness of anti-smoking campaigns can have a considerable effect on lung cancer rates in the more distant future. [J Natl Cancer Inst 1988; 80:43-51] Trends in lung cancer mortality during this century have been among the most remarkable phenomena in health sta- tistics. From a relatively rare killer of both men and women in the early 1900s, lung cancer has developed into one of the biggest public health tragedies of the century, now claiming over 100,000 lives each year (1). Cancers of the lung and bronchus have been the most frequent cause of cancer death Am~I~W~,T~ONS USE~: HIS = Health Interview Survey; ICD = International Classification of Diseases; NCI = National Cancer Institute. 1 Received August 24, 1987; accepted September 3, 1987. ~ Biometry Branch, Division of Cancer Prevention and Control, National Cancer Institute, Bethesda, MD 20892-4200. 3 Surveillance and Operations Research Branch, Division of Cancer Pre- vention and Control, National Cancer Institute. 4 We thank Dr. Ronald Wilson, National Center for Health Statistics, for providing unpublished data from the U.S. Health Interview Study. We alsc thank Dr. David Byar, Dr. David Levin, Dr. Ed Sondik, Dr. Ben Hankey, Dr Mitchell Gall, Dr. Steven Piantadosi, and Dr. Julian Peto for their comment: and suggestions. In addition, we express our gratitude to Ms. Helen Triok for tireless computer assistance and to Ms. Caroline Ball for typing an~ editorial assistance. Vol. 80, No. 1, March 2, 198[ 2063628170

44 of mal~s in the United States for the past 4~0 years and were the second most frequent cause of cancer death among U.S. females for the past two decades. Recently, lung cancer has nearly surpassed breast cancer in age-adjusted mortality rates for women (2). Because lung cancer is such a major contributor to the overall U.S. cancer mortality picture, its reduction ~s of the greatest priority among all the various groups involved in cancer control. However, the implementation of effective cancer control strategies and their evaluation will depend to some degree on what is likely to happen to lung cancer trends in the absence of any new major cancer control activi- ties. The purpose of this article is to present projections of lung cancer mortality rates through the remainder of this century and the first three decades of the next. These projec- tions are of considerable interest to the public health com- munity not only because they will suggest the coming mor- tality for one of the major causes of death, but also because they reflect the general health burden for a variety of dis- eases related to smoking. Although studies showing that cigarette smoking is the major determinant of lung cancer were presented to the scientific.community as early as 1950, it was not until the first Surgeon General's report on smoking and cancer in 1964 that this relationship was established in a widely accepted public forum (3). Therefore, it is not surprising in reviewing the evidence concerning cigarette smoking pat- terns in the United States that the decades between the end of the First World War and the Surgeon General's report were marked by a substantial increase in the per capita consump- tion of cigarettes, with a particularly rapid growth during the Second World War (4). Because there is a considerable lag time between beginning smoking and the development of cancer [e.g., Doll and Peto (5)], the effects of the tremen- dous increase in smoking have been seen in the lung cancer mortality statistics in the decade of the 1970s. Two phenomena of recent decades may have led to the beginning of changes in the lung cancer picture for the 1980s and beyond (4). Recent data have shown considerable declines in the smoking prevalence among U.S. males, but slower declines for U.S. females (6, 7). These declines among men began with the release of the 1964 Surgeon General's report and continued with the removal of cigarette advertis- ing from television. Reduction in tar content of all brands of cigarettes and the introduction of low-tar brands beginning in the late 1960s should also lead to lower lung cancer risks even without prevalence changes. Changes in the prevalence of smoking and tar content of cigarettes have already affected U.S. incidence and mortality rates for men under age 55 (8,9). However, no such decline is yet seen in women, which no doubt reflects their dramatic increases in smoking prevalence during the 1960s and 1970s. These complex changes in smoking behavior over time have lead to morbidity and mortality patterns that have engendered much research (10-16). These articles have generally attempted to explain the distinct patterns in lung cancer mortality by various models, sometimes taking ex- Journal of the National Cancer Institute plicit account of smoking trends. Despite the use of a variety of models fitting available mortality data, there has been lit- tle published to date on projections of those trends. One such projection by Janerich (17) using a simple model suggests that lung cancer mortality will continue to rise at a rapid rate and will dominate mortality trends for the near future. For- tunately, this rapid increase is not likely to be realized, as recent reports (8) continue to show the decline in lung cancer incidence among U.S. males. Clearly, a sophisticated modeling approach to these trends is needed and, preferably, one that takes into account available data on past smoking behavior. Subjects and Methods ~ The analyses presented in this article are based on ~he numbers of deaths from malignant neoplasms of the trachea, bronchus, and lung (ICD 162-163.0, eighth revision) occur- ring among white males and females in the United States during 1958-1982. We have not used data prior to 1958 because the ICD coding changed at this time. The data, numbers of deaths and person-years at risk, have been aggregated into 12 age groups (30-34, 35-39 ..... 80-84, >--85) and 5 calendar year periods (1958-1962 ..... 1978- 1982). Person-years at risk are approximated by mid-year population estimates. Our estimates of past smoking prevalence are derived from cigarette smoking histories from the 1978-1980 HIS conducted by the U.S. National Center for Health Statistics. Details of the HIS, a stratified, household-based, personal interview survey, are reported elsewhere (18). The respon- dents in this survey numbered 22,990 males and 26,725 females age 17 or older at the time of the interview. Less than 1% of the interviews were proxies and were not included in our analysis. For each individual, a history of smoking status was derived from answers to the questions of current smoking status, age started smoking regularly, and the time since last smoked regularly. Those who never smoked cigarettes regularly were classified as nonsmokers for their life. Former smokers were classified as nonsmoking from the age they reported cessation of regular smoking. Those reported as ever smoking with partially missing information were classified according to the following assumptions: 1) When the age at initiation of regular smok- ing was unknown (<0.5%), the modal age of 18 was used; 2) when a self-reported former smoker's time since cessation was unknown (1.6%), the time was assumed to be zero; 3) individuals who were coded as having unknown current smoking status but did report an age of initiation (0.2%) were treated as current smokers. Because cigarette smokers have higher mortality rates than nonsmokers, estimates of past cigarette smoking prev- alence based on currently living persons will understate the actual prevalence. We therefore followed the approach of Harris (6) to correct for this bias. Using his notation, we let Ptu and Qt~ denote the proportions of current and former smokers at age t among respondents alive at age u > L An

• estimat.e of the prevalence of cigaretteSmoking at age ~ u - t years in the past, is given by Ptu/Stu P,(u) = Ptu/Slu + Qm/Ft~ + (1 - P~ - Qlu)/Nlu ' [1] where Stu, Flu, and Nlu represent the probabilities of surviv- -~ng from age t to age u for current smokers, former smokers, and never smokers, respectively. Estimates of St, and Ntu are from the American Cancer Society study (Garfinkel L: per- sonal communication). We did not have estimates of Ftu, so, assuming former smokers would experience mortality more similar to that of smokers than never-smokers, we used Slu to represent their survival. The 3 survey years provided three estimates of the age- specific smoking prevalence for each year in the past. These were combined into a single estimate weighted by the number of persons interviewed. For example, individuals aged 70 and interviewed in 1978, those aged 71 and inter- viewed in 1979, and those aged 72 and interviewed in 1980 contributed to the estimate of the proportion of smokers aged 42 in 1950. Because the trends over time in lung cancer morbidity and mortality are-strongly related to changes in cigarette smok- ing and the composition of cigarettes (4,19,20) and because the latent period for lung cancer occurrence may be 20-40 years, past changes in smoking behavior and cigarette com- position should be an important aspect of any projection of future lung cancer rates. The prevalence of cigarette smok- ing has been changing from one birth cohort to the next (6), and the introduction of low-tar cigarettes has greatly re- duced the average tar content of cigarettes sold in the United States (fig. 1). Whereas changes in smoking prevalence should produce birth cohort effects in the trends of lung cancer mortality, changes in cigarette composition affecting smokers across different birth cohorts at the same calendar time should produce calendar period effects in the lung cancer mortality trend. ~955 ~960 1965 ~igUre |. ~verage tar content of cigarettes sold and age-a~ust~d smoking p~eva|ence of adu|ts in the United States, ~ 9~4- ]9~0. "~" "°" "°' 1970 1975 For these reasons we: have based our statistical analysis of lung cancer mortality on an age-period-cohort model (21,22). This type of statistical model has been used previously for cohort analyses of lung cancer (23,24) but has not been used to project future disease occurrence. This model assumes that the number of cancer deaths Dij observed in age group i during calrndar periodj follows a Poisson probability distri- bution with mean I~ij = Nqri.i, where Nij denotes the size of the population at risk and rU denotes the rate of cancer m0r: tality. The rij are modeled as a function of age, calendar period, and birth cohort. More specifically, it is assumed that log (rij) = Ai -[- I~ q- C1-i +j, for i = 1,2, I - andj = 1,2 ..... J, .... [2] where Ai (i = 1,2 ..... 12) denotes the age effect for ages 30-34, 35-39 ..... 80-84, >--85; Pj (j = 1,2 ..... 5) denotes the period effect for calendar periods 1958-1962, 1963- 1967 ..... 1978-1982; and C~ (k = 1,2 ..... 16) denotes the cohort effects for birth cohorts 1869-1877, 1874-1882, .. :., 1944-1952. The model is then fit to the data by maximum likelihood methods. A linear dependency (k = I - i + j) exists among the three factors age, calendar period, and birth cohort, which induces a noriidentifiability of the linear components of the three sets of parameter estimates (25-27). This non- identifiability means that we cannot estimate the linear component of trends over time for the period and cohort effect parameters. Therefore, we cannot determine whether the period effect parameters are increasing while the cohort effect parameters are decreasing or vice versa. The nonlinear components of each factor, however, are estimable (26). Unfortunately, the increasing or decreasing linear trends of the individual period and cohort factors are often of primary interest. A number of solutions have been proposed for this non- identifiability prbblem. Most of these involve constraints to be placed on the parameters; however, the parameter esti- mates have been shown to be sensitive to the choice of con- straint (27). The change in parameter estimates from use of one constraint to another can be so extreme as to preclude meaningful interpretation. Therefore, unless there is a very compelling reason for choosing a particular constraint, this approach will not provide a satisfactory solution. Another method, proposed by Day and Charnay (23), requires two or more populations, one of which can be adequately fit by a two-factor model. We found this approach was not applicable because we could not satisfactorily describe either the male or female deaths by a two-factor model. Other proposed solutions involve finding the single three-factor model that is "closest" to the best fitting two-factor model (25,28). How- ever, these approaches are ad-hoc statistical solutions thai have no biological justification. Rodgers (29) suggested that a valid solution could be obtained by replacing one of the factors with a more directl) relevant variable for which the factor is thought to be ar indirect indicator. This is the approach we have used here As suggested by Day and Charnay (23) in their analysis ol lung cancer in Slovenia and Finland, one would expect the trend of changing cigarette tar content to be reflected in lung Vol. 80, No. 1, March 2, 198~ 2063628172

cancer mortality as a calendar period effect acting across all a~e groups. Therefore, we followed Rodgers' approach by replacing the period parameters in equation 2 with a regres- sion variable related to the average tar content and number of cigarettes sold. Thus our model is -, log (rij) = Ai q- BXj + Cl-i+j, [3] where the period parameter of equation 1 is replaced by a regression on Xj which denotes a measure of the population's exposure to cigarette tar during thejth calendar period. Results To fit the model in equation 3 to the observed lung cancer mortality data, we evaluated two exposure measures for the variable Xj: (1) the average tar content of cigarettes sold in the United States and (2) the product of average tar and the number of cigarettes sold per capita (age -->20) as a measure of the entire population's average exposure level. Figure 1 shows the changing average tar content of cigarettes sold (30,31) along with estimates of the number of cigarettes sold per capita to males and females for the period 1954-1980. Our estimates of the sex-specific numbers of cigarettes sold are derived from age-adjusted smoking prevalence estimates applied to the total annual cigarette sales. Our smoking prevalence estimates are based on applying the estimator in equation 1 to the HIS data to derive age-specific prevalence rates aggregated into 5-year age groups. Table 1 shows that the age-specific prevalences of past smoking as estimated from the 1978-1980 HIS generally agree (an average differ- ence of 3.6%) with prevalences estimated from actual sur- veys conducted in 1966, 1970, and 1975. We then used the 1970 standard U.S. population age distribution to obtain direct age-adjusted smoking prevalence rates back through 1954. Since the HIS data did not provide _age-specific prevalence rate estimates for the age groups 80-84 and -->85 during the earliest time periods, we used the rates based on our estimates for 1975. The error induced by this should have little impact, since few people were in these age groups. Table 1. Estimated prevalence of current smokers according to age and sex in the United States, 1966-1975 Sex Age (yr) 1966 1970 1975 Survey* HIS'~ Survey HISS" Survey HISS" Males Females 20-24 61.9 56.2 49.8 52.4 41.3 46.5 25-34 59.9 56.2 46.7 52.6 43.9 50.4 35-44 59.0 57.4 48.6 53.0 47.1 47.2 45-54 53.8 54.3 43.1 51.1 41.1 47.3 55-64 47.7 44.2 37.4 42.0 33.7 41.6 ~65 27.8 25.9 22.8 28.1 24.2 27.9 20-24 49.2 43.1 32.3 38.7 34.0 37.5 25-34 45.1 43.2 40.3 43.0 35.4 39.8 35-44 40.6 40.9 38.8 39.6 36.4 39.9 45-54 42.0 36.5 36.1 36.9 32.8 36.4 55-64 20.6 21.6 24.1 26.9 25.9 30.1 ~65 7.6 8.0 10.2 11.4 10.2 15.1 *From (41); separate surveys were done in 1966, ~'Estimated from 1978-1980 HIS as described in Journal of the National Cancer Institute 1970, and 1975. text. Figure 1 shows a steady decrease in the sales-weighted average cigarette tar content from 37.5 mg in the mid-1950s to 14 mg in 1980, a drop of over 60%. The estimated number of cigarettes sold to males peaked between 1960 and 1965 and has declined steadily since then, whereas the estimated number sold to females has risen over this period. When fitting the model in equation 3, we used 5-year aver- ages of the average tar content or of the product of average tar and number sold. Because changes in carcinogenic expo- sure are not reflected immediately in changing cancer mor- tality, we examined possible lag periods between our expo- sure measures and mortality when fitting the regression mo.del in equation 3. Information on the average tar content of cigarettes sold before 1954 is not available. Since filter cigarettes were originally introduced in the mid-1950s, one assumption is that tar content underwent little change before this time (Warner K: personal communication). Therefore, we assumed the pre-1954 average cigarette tar content to be the average of the 1954-1958 levels, 36.6 mg. On the basis of minimizing the deviance as a measure of goodness-of-fit, we concluded that the best exposure regressor variable is the product of average tar content and number of cigarettes sold, lagged for 24 years. Peace (32) found a 21-year lag when correlating overall lung cancer mortality with cigarette tobacco sales by weight in England and Wales during 1880-1983. However, his analysis was not adjusted for age and birth cohort. In addition to changes in tar content of cigarettes, there have been other changes that probably have affected lung cancer patterns. Changes in air pollution and occupational exposure head the list. We judge these likely to be very small in comparison to cigarette smoking. The relative risk of occupational exposures that might serve as an upper bound for the effect of environmental carcinogens is on the order of 1.4-3.2 (33), and these apply to small proportions of the population, leaving a small attributable risk. Others have suggested a minor role for pollutants (4), and the evidence from time trends in nonsmokers does not substantiate any temporal effect in environmental carcinogens (34). Finally, recent evidence from a large case-control study in Western Europe shows substantial reductions in lung cancer inci- dence attributable to lower tar cigarettes (35-37). To assess whether separate regression coefficients were needed for males and females and for different age groups, we compared the fit of variohs models using an analysis of deviance (38). Our analysis indicated that the slope of the average tar content × number cigarettes sold exposure vari- able differs by both sex and age. A statistical test of equal slopes for males and females yields a one degree of freedom chi-square value of 37.6, while a test for equal slopes for under age 50 and ->50 yields a value of 155, both highly significant (P <.001). We examined other age group cate- gorizations and found the under/over 50 to give the best fit. As shown in tables 2 and 3, females have a larger slope than males and the under-50 group has a larger slope than the over-50 group. These differences are consistent with surveys beginning in the late 1960s that have shown that males and older persons are more likely to continue to smoke higher tar cigarettes (39). Therefore, decreases in tar levels would be

expected to have a greater effect on Ring cancer among feinales arid younger persons. Thus our final model that we fit to male and female lung cancer mortality is I Ai + B1Xj + Cl-i+j i<--4 = , [41 log (rU) {Ai + BzXj + Ct- i +j i > 5 where B1 is the regression coefficient for ages 30-49 and B2 is the coefficient for ages -->50. Parameter estimates of the age, period, and birth cohort effects for males and females are given in tables 2 and 3, Table 2. Parameter estimates from fitting model in equation 4 to male lung cancer mortality Age (yr) Birth cohort 30-34 -- 12.92 35-39 -- 11.77 40-44 - 10.79 45-49 --9.95 50-54 --9.01 55-59 --8.43 60-64 --7.91 65-6.9- ......... --7.54 • 70-74 --7.25 75-79 --7.08 80-84 -6.98 -->85 --7.01 Slope of average tar X No. sold 1869-1877 0.13 1874-1882 0.37 1879-1887 0.64 1884-1892 096 1889-1897 1.22 1894-1902 1.42 1899-1907 1.56 1904-1912 1.64 1909-1917 1.72 1914-1922 1.79 1919-1927 1.85 1924-1932 1.92 1929-1937 1.87 1934-1942 1.75 Age 30-49 yr 2.24 X 10-3 1939-1947 1.52 Age _>50 yr 0.92 × 10-3 1944-1952 1.21 Table 3. Parameter estimates from fitting model in equation 4 to female lung cancer mortality Age (yr) Birth cohort cohort becomes habituated to smoking when young. To examine this interpretation, figures 2 and 3 compare the time patterns of the sex-specific cohort parameter estimates with the age-specific cigarette smoking prevalence estimates among males aged 20-24 and females aged 30-34 for dif- ferent birth cohorts. Because the prevalence of smoking for a cohort peaks around these ages, this age-specific prevalence is hypothesized to represent a measure of smoking habits by birth cohort, which becomes translated into the cohort parameters in our mortality model. The age at which this prevalence reaches a peak has changed over the years. For -- cohorts born around 1900, the peak prevalence was reached around age 30 for men and age 45 for women. More recent cohorts have shown a peak at 20-24 for males and 25-29 or 30-34 for ferhales. Both figures show that each set of estimated cohort parameters for lung cancer mortality exhibits a pattern, quite similar to that of our cohort smoking behavior index. The time patterns of the cohort parameters and cohort smoking index exhibited by the females are very similar to one another (correlation coefficient of 0.99), while the pattern of the male cohort smoking index appears less regular than the female pattern (correlation of 0.83). We hypothesize that the male pattern has been affected by the Depression, reducing the smoking index for cohorts born from 1905 to 1920. These figures indicate that our proposed cohort smoking index provides a good representation of the cohort parameters in our age-period-cohort model, and this correlation will be used for the projections in the following section. 30-34 --13.42 1869-1877 --1.20 35-39 --12.20 1874-1882 --1.04 40-44 --11.19 1879-1887 --0.92 45-49 --10.36 1884-1892 --0.77 50-54 --9.55 1889-1897 --0.59 55-59 --8.95 1894-1902 --0.33 60-64 --8.40 1899-1907 --0.01 65-69 --7.93 1904-1912 0.36 70-74 --7.52 1909-1917 0.75 75-79 --7.21 1914-1922 1.07 80-84 --6.96 1919-1927 1.29 ->85 --6.77 1924-1932 1.50 Slope ofaverage tar× No. sold 1929-1937 1.60 1934-1942 1.58 Age 30-49 yr 5.38X 10-3 1939-1947 1.49 Age_>50yr 3.28X 10-3 1944-1952 1.29 respectively. The male lung cancer mortality rate peaks for the cohort born around 1928, while the rate for females attains a peak for the cohort born arour.d 1933. Since the age and cohort parameters are unique up to an additive con- stant, we adjusted the age parameters to reflect the age- adjusted mortality rates among nonsmokers (34). This was done so the model would predict what the effect of eliminat- ing smoking would be to attain nonsmoker mortality rates. Day and Charnay (23) suggested interpreting the cohort parameters as reflecting the number and type of cigarettes a 0.5 0.4 15~$ 1~95 1905 1918 19~ 1930 1940 Figure 2. Comparison of estimated cohort parameters and prevalence of smoking at ages 20-24 for U.S. white males. Basis for projections. To make projections of lung cancer mortality, our age-period-cohort model requires projections of the parameters of the period and cohort factors. Our assumption that the age parameters remain fixed at their estimated values is consistent with our interpretation that they represent the background level of lung cancer risk in a nonsmoking population. Because lung cancer risk is primar- ily the result of cigarette smoking, our estimates of the future period and cohort factors are based on projections of future Vol. 80, No. 1, March 2, 1988 2063628174

cigarette composition and consumption. From the model in equation 4, the period parameters are linear functions of the product of average cigarette tar content with the number of cigarettes sold, while figures 2 and 3 show that the cohort parameters can be estimated as functions of the prevalence of smoking among young adults. Therefore, we need to pro- jeer;..(1) the average tar content of cigarettes sold in the United States, (2) the number of cigarettes sold in the United States, and (3) the prevalence of smoking among young adults. 0.5 Figure 3, Comparison of estimated cohbrt parameters and prevalence of smoking at ages 30-34 for U.S. white females. Projecting tar content in the future is somewhat difficult because the sales-weighted averages shown in figure 1 are a combination of changes in the production of virtually all cigarettes made in the United States, the development of fil- ters and of new low-tar brands in the last three decades, and cl~anges over time in the proportion smoking various types. In addition, the trends in these data are quite .strong and almost linear; however, a linear decrease cannot continue unabated, and the point at which one might choose to "level off" these projections is speculative. Surveys have shown that smokers of high-tar (especially unfiltered) cigarettes are generally concentrated among the elderly (I9), and these individuals will soon die off, creating further declines in the sales-weighted average tar level. In addition, as the propor- tion of women in the smoking population increases, the market share of high-tar cigarettes will likely further de- crease. For the projections to follow, we assume sales-weighted tar will continue to decrease in a linear fashion until reach- ing the optimistic level of 5 mg per cigarette and then level off. This linear trend is estimated from the 1972-1981 aver- age tar content values, when the estimated yearly decrease was 0.74 mg. In a second scenario we took the conservative view of leaving the sales-weighted tar constant at 13.22 mg per cigarette, the 1981 value. A projection of the future number of cigarettes sold in the United States requires projections of future smoking preva- Journal of the National Cancer Institute lence among all adults and. the average number of cigarettes purchased by each smoker.' Developing projections for smok- ing prevalence, we relied on the objectives developed by the NCI for the Year 2000 Project (40). The NCI has set a goal of decreasing the smoking prevalence from current levels to 15% of all adults by the year 1990. Our smoking prevalence projections ar6 based on assuming a linear decrease in the age-adjusted (ages -->15 adjusted to the 1980 population) smoking prevalence from the levels of 40.6% for males and 32.3% for females estimated from the 1978/1980 HIS. To project our exposure index of the tar X number sold per cap- ita, we assume that the average number of cigarettes sold per smoker would remain at the 1980 levels. Therefore, the proj~ected values of our tar × number sold index is the product of the projected tar level, the projected prevalence of smoking, and the number of cigarettes sold to smokers. These projections, along with the values observed in the past, are given in table 4. Because of our estimated 24-year lag, these average levels during 1934/1938 ..... 1954/1958 provide the calendar period component of the fitted mortality rates for the periods 1958/1962 ..... 1978/1982, while the actual and projected levels for the years 1959/1963 through 1999/2003 are components of our mortality rate projections through the period 2023/2027. As seen in figures 2 and 3, the pattern of fitted cohort parameters in our mortality model is similar to the pattern of smoking prevalence among young adults. Therefore, We use projections of the age-specific prevalence of smoking for males aged 20-24 and females aged 30-34 to estimate future cohort parameters for our mortality projections. We Table 4. Actual and projected 5-year averages of tar × number of cigarettes sold (in 1,000s) per capita Years Averages Males Females Actual 1934/1938 108.9 22.1 1939/1943 145.3 36.0 1944/1948 203.1 58.9 1949/1953 212.8 72.7 1954/1958 207.6 79.6 1959/1963 169.2 72.7 1964/1968 137.9 66.4 1969/1973 113.6 61.2 1974/1978 94.8 57.7 Projeaed* 1979/1983 59.4 48.1 1984/1988 35.3 31.4 1989/1993 18.0 17.9 1994/1998 14.1 14.1 1999/2003 14.1 14.1 *We assume: (1) tar content to decrease linearly to 5 mg in 1993 and (2) smoking prevalence to decrease linearly to 15% for all adults in 1990. found that the best fitting functional relationship between young adult smoking prevalence and the cohort parameters ro to be Yi = bXi2, where Yi represents the fitted cohort param- O eter and Xi represents the smoking prevalence for the ith co birth cohort ranging from 1888 to 1948. This relationship ~ assumes no intercept so that as the smoking prevalence 0~

decreases to zero the cohort parameters:.(would also go to zero and the age-specific lung cancer mortality risk would become the 1958-1968 nonsmokers' risk as noted in the previous section. The regression for males was estimated by least squares applied to the data in figure 2, which resulted in an estimated slope b = 4.4. The regression for females wag'estimated from the 1908-1948 data shown in figure 3 and resulted in an estimated slope of/~ = 8.2. Following the Year 2000 Project objectives, one scenario is that the prevalence of smoking among young adults drops to 15% by the year 2000. Our smoking prevalence projec- tions for males aged 20-24 assume a linear decrease from the 41% observed for those born during 1956-1960 to 15 or 0% for those born during 1976-1980 (being 20-24 in the year 2000). The projections for females aged 30-34 decrease linearly from the observed 39% for those born during 1946-1950. The smoking prevalence projections assuming 15% prevalence by the year 2000 are .given in table 5. Table S. Projection of smoking prevalence among young adults* _- Year of birth Smoking prevalence? Males Females 1946-1950 (0.54) (0.39) 1951-1955 (0.48) 0.33 1956-1960 (0.41) 0.27 1961-1965 0.345 0.21 1966-1970 0.28 0.15 1971-1975 0.215 0.15 1976-1980 0.15 0.15 *Cell entries are smoking prevalence for males of ages 20-24 and females of ages 30-34 by birth cohort; both sexes assumed to have 15% prevalence of smoking by the year 2000 in their respective age groups; as explained in the text, cohort parameters are given by: Males: parameter = 4.4 × (smoking prevalence)2 Females: parameter = 8.2 X (smoking prevalence)2 ?Observed prevalence in parentheses. Table 6. Actual and projected age-adjusted lung cancer mortality rates per 105 Years Actual rates Males Females 1958/1962 38.1 5.8 1963/1967 47.3 7.5 1968/1972 57.9 11.I 1973/1977 65.2 ..-15.2 1978/1982 71.0 20.6 Projecmd rams Males Females NCI obj.* No change NCI obj.* No change 1983/1987 72.6 72.6 25.6 25.6 1988/1992 72.9 73.0 30.8 30.8 1993/1997 -71.5 71.7 35.4 35.6 1998/2002 68.1 68.6 38.9 39.5 2003/2007 61.3 64.6 39.6 42.3 2008/2012 52.8 58.9 36.2 43.2 2013/2017 44.5 52.7 32.9 42.6 2018/2022 36.1 47.0 28.3 40.8 .. 2023/2027 '28.8 43.0 23.4 38.8 *NCI objectives: (1) tar content decreases linearly to 5% in 1993, (2) age-adjusted smoking prevalence drops to 15% in 1990, and (3) smoking prevalence in young adults dro, ps to 15% in 2000. there is a 40% difference in the projected rates, from 38.8 to 23.4 deaths per 100,000. Varying each of the three elements ofthe projection model, tar, age-adjusted smoking prevalence, and age- specific prevalence among new smoking cohorts, produces different projection tables for males and females. We examined more conservative projections, for example, tar levels not decreasing below 1982 levels and age-adjusted prevalence not declining below 25% for adults, as well as more liberal projections, including, for example, no new smokers among young adults by the year 2000. For the near term, the period 1998/2002, the differences in age-adjusted Projections of mortality. The age-adjusted projections are shown in table 6 and figure 4 for the NCI objectives com- pared to a "baseline" alternative in which tar content and smoking prevalence are assumed not to change after 1980. For both males and females, but especially for males, the two sets of projected rates for the period 1998/2002 differ very little because of the estimated 24-year lag for the effect of changes in tar content and age-adjusted smoking preva- lence. The only changes in our mortality projections in this short time span are due to different cohort effects, and these will affect only the young age groups with low mortality rates before the end of this century. For males, the projected year 2000 age-adjusted rate under the NCI objectives is 68.1 per 100,000, just 0.7% less than the rate of 68.6 per 100,000 for the no-change scenario. The differences in mor- tality rates projected for the period 2023/2027 between the two scenarios are noticeably larger. The projected rates for males are 43.0 and 28.8 per 100,000 for the no-change and NCI scenarios, respectively, representing a 33% decline due to accomplishing the NCI's smoking objectives. A larger dif- ference is apparent for females for the 2023/2027 period; PRO-TECTED 1960 1970 1950 1990 ~.000 ~010 ~0:~0 ~030 C~ t en(lar Year Figure 4. Agtual and projected age-adjusted lung cancer mortality rates for U.S. white males and females, 1960-2025. . = NCI objectives; --- = no change. Vol. 80, No. 1, March 2, 1988 2053628176

5~ rat~s are trivial. As the projection horizon lengthens, the dif- ferences~ become more noticeable. However, the rates based on more liberal objectives are similar to the NCI objective projections, illustrating that accomplishment of the NCI objectives (tar declines linearly, smoking prevalence at 15%) would include much of the potential reduction in lung cancer mortality for the next several decades. We considered the more dramatic scenario of a 0% prevalence of smoking by the year 2000 and obtained similar results. For example, under this scenario, the projected male age-adjusted mortal- ity rate for the period 2023/2027 is 27.9 per 100,000, only 3.1% less than the projection of 28.8 based on the NCI objectives. Females show a larger relative decline of 11% from 23.4 deaths per 100,000 to 20.9 in the 2023/2027 period. These detailed age-specific projection tables are available from the authors upon request. Projections at the age-specific level better illustrate why the rate of decline in the age-adjusted projections is so slight. When the different effects begin to be seen in each of the age groups and how fast these rates drop are crucial to interpreting the plausibility of the projections. As shown for males in figure 5, by the year 2000 only the age groups under .45 are affected by the decrease in new smokers represented by attaining the NCI objectives. For groups older than age 45, changes are not apparent until later. For males aged 55-59, the recent (1978/1982) lung cancer mor- tality rates are 169 per 100,000. Under the no-future-change model, these are projected to be 109 in the year 2000 and lower still to 80.5 by the year 2025. Therefore, the no- change scenario includes a substantial decline in lung cancer mortality for this younger age group to begin in the near future and to carry through this projection horizon due to changes in tar content and smoking prevalence that have already occurred. Achievement of the NCI objectives would further reduce these rates to 31.2 in 2025. A similar picture is evident for older males. Among the 75-79 year olds, com- pared to a recent rate of 487, the no-change scenario projects rates of 587 for the year 2000 and 311 by the end year of the projections (2025). The decline in mortality rates for this age group begins about the year 2005, 20 years after the turnaround for the 55-59 year age group. The general age- specific pattern for females is quite similar, but shows a later period of peak mortality rates (fig. 6). For the 55-59 year age group, the peak mortality rate of 79.6 is projected to be in the 1988/1992 period, 5 years later than males. The female rates for this age group for the years 2000 and 2025 are projected to be 69.4 and 56.9, respectively, under the no-change scenario. Attaining the NCI objectives reduces the later projected rates to 13.0 per 100,000. Discussion Trends in the United States during the past two decades have shown a dramatic turnaround in smoking prevalence. Because of these trends, changes in lung cancer and other tobacco-related diseases have been eagerly anticipated. De- clines in age-specific rates of lung cancer for young ages have been seen recently as a result of lower smoking preva- lence among new cohorts and as a result of lower tar content in cigarettes. However,. age-adjusted rates of lung cancer mortality have continued to climb for males, although recently there has been a noticeable flattening of these rates. For U.S. females, however, incidence and mortality rates have continued to dramatically increase. In the present analysis, we have constructed a model to take account of these trends of major public health interest and have provided a framework for projecting rates into the future. The model is based on age-period-cohort modeling, which has been used extensively in cancer epidemiology. In addition to fitting the model to available lung cancer mor- tality data, we have analyzed recent survey data on cigarette consumption and incorporated these findings into the projec- tion model. This allows the projection of lung cancer rates, providing one is willing to make a set of assumptions about the three key-factors that we believe are the major determi- 9O0 i100 10 ~CTUAL ; PROJECTED 1960 1970 1980 1990 2000 2010 2020 2030 Catendar Year Figure 5. Actual and projected age-specific lung cancer mortality rates for U.S. white males, 1960-2025. = NC1 objectives; .... no change. 900 ACTUAL PROJECTED 1960 1970 1980 1990 2000 2010 2020 2030 Figure 6. Actual and pr~ected age-specific lung cancer mortality rates for U.S. white ~males, 1960-2025. = NCI o~ective; --- = no change. Journal of the National Cancer Institute

nants of lung cancer mortality: starting the smoking habit, cbnfinuedr smoking prevalence, and tar content of cigarettes. Projections of lung cancer rates in the absence of any major changes in these factors show that the age-adjusted rates in males will be relatively flat through about 1990 and will then gradually decline. Mortality patterns for females lag:behind those for males, and in the absence of changes, rates are projected to peak about the year 2010. The age- specific peaking for females is only 5 years behind that for males; the age-adjusted ra(es peak later for females because of their more recent increase in smoking prevalence. Achievement of objectives for reducing smoking preva- lence advocated by the NCI will have little impact on these trends in the very near future. Little material changes can be expected by the year 2000 based on the empirical findings here; however, the longer term impact of the NCI objectives is much more positive. Compared to no changes in smoking prevalence and cigarette tar content, attaining the NCI objectives could reduce the age-adjusted male lung cancer mortality rate by almost 33% by the period 2023/2027 and the female mortality rate could be reduced by over 40%. Assuming attainment of the NCI objectives, we note that the projected age-adjusted rate for males for the period 2023/ 2027 of 28.8 per 100,000 represents almost a two-thirds reduction of the 1978/1982 mortality rates. Although the rate of 23.4 per 100,000 for females in 2023/2027 is about equal to recent rates, this is a considerable reduction on what would be projected under current trends. Advances in sec- ondary and tertiary prevention represented by screening and treatment might also affect future lung cancer mortality trends, but we have limited our analysis to the future effects of cigarette smoking behavior. We are somewhat surprised by the intractability of the rates in the near future. However, the empirical fit of the model to the data suggested lagged effects for parameters, which are consistent with previous literature (32). Thus reduction in lung cancer mortality rates in the next decade or two will occur only if recent decreases in smoking prevalence continue and efforts to reduce smoking further are adopted throughout the United States. References 1. SILVERBERG E, LUt~ERA J. Cancer statistics, 1986. CA 1986;36:9-25. 2. American Cancer Society. Cancer statistics. New York: Am Cancer Soc, 1985. 3. US Public Health Service. Smoking and health. Report of the Advisory Committee to the Surgeon General of the Public Health Service, US Department of Health, Education, and Welfare. Washington, DC: US Govt Print Off, 1964 (PHS publica- tion No. 1103). 4. DOLL R, PETO R. The causes of cancer, quantitative estimates of avoidable risks of cancer in the United States today. JNCI 1981;66:1191-1308. 5. DOLL R, PETO R. Cigarette smoking and bronchial carcinoma: dose and time relationships among regular smokers and lifelong non-smokers. J Epidemiol Community Health 1978;32:303-313. 6. HARRIS J. Cigarette smoking among successive birth cohorts of men and women in the United States during 1900-80. JNCI 1983;7 t:473-479. 7. WILSON RW, AAVEDAL MJ. Smoking data from the National Center for Health Statistics. Presented at the American Public Health Association, November 1984. 8. HORM JW, KESSLER LG. Falling rates of lung cancer in men in the United States. Lancet 1986;1:425-426. 9. DEVESA SS, HORM JW, CONNELLY RR. Trends in lung cancer incidence and mortality in the United States. In: Mizell M, Correa P, eds. Lung cancer, causes and prevention. Deerfield Beach, FL: Verlag Chemic lnt, 1984:33-45. 10. SCHNEIDERMAN MA, LEVIN DE Trends in lung cancer, mortality, incidence, diagnosis, treatment, smoking, and urbanization. Cancer 1972;30:1320-1325. I 1. BURBANK F. U.S. lung cancer death rates begin to rise proportionately more rapidly for females than for males: a dose-response effect? J Chronic Dis 1972;25:473-479. 12. TOWNSEND JL., Smoking and lung cancer, a cohort data study of men and women in England and Wales 1935-1970. J I~ Statist Sue A 1978;141:95-107. 13. MANTON KG, STALLARD E. A population-based model of respiratory cancer incidence, progression, diagnosis, treatment, and mortality. Comput Biomed Res 1982;15:342-360. 14. HORM JW, ASIRE AJ. Changes in lung cancer incidence and mortality rates among Americans: 1969-78. JNCI 1982;69:833-837. 15. CUMMINGS KM. Changes in the smoking habits of adults in the United States and recent trends in lung cancer mortality. Cancer Detect Prey 1984;7:125 - 134. 16. HAKULINEN T, PUKKALA E. Future incidence of lung cancer:, forecasts based on ~hypothetical changes in the smoking habits of males. Int J Epidemiol 1981;10: 233-240. 17. JANERICH DT. Forecasting cancer trends to optimize control strategies. JNCI 1984;72:1317-1321. 18. National Center~for Health Statistics. Health, United States, 1984. Washington, DC: US Govt Print Off. 1984 [DHHS publication No. (PHS)84-1232]. 19. Office on Smoking and Health, Public Health Service, US Department of He~'lth and Human Services. The health consequences of smoking. Cancer. A report of the Surgeon General. Washington, DC: US Govt Print Off, 1982 [DHHS publica- tion No. (PHS)82-50179]. TODD GF, LEE PN, WILSON MJ. Cohort analysis of cigarette smoking and of" mortality from four associated diseases. London: Tobacco Research Council, 1976. KERMACK WO, MCKENDRICK AG, MCKINLAY PL. Death rates in Great Britain and Sweden: expression of specific mortality rates as products of two factors and some consequences thereof. J.Hyg (Lond3 1934;34:433-457. BARRETT JC. The redundant factor method and bladder cancer mortality. J Epidemiol Community Health 1978;32:314-316. DAY NE, CHARNAY B. Time trends, cohort effects, and aging as influence on cancer incidence. In: Magnus K, ed. Trends in cancer incidence (causes and prac- tical implications). Washington, DC: Hemisphere, 1982:51-65. STEVENS RG, MOOLGAVKAR SH. A cohort analysis of lung cancer and smoking in British males. Am J Epidemiol 1984;I I9:624-641. OSMOND C, GARDNER MJ. Age, period and cohort models applied to cancer mortality rates. Stat Meal 1982;1:245-259. HOLFORD TR. The estimation of age, period and cohort effects for vital rates. Biometrics 1983;39:311-324. KUPPER LL, JANIS JM, KARMOUS A, et al, Statistical age-period-cohort analysis: a review and critique. J Chronic Dis 1985;38:811-830. MOOLGAVKAR SH. Risk assessment using vital data. In: Prentice RL, Whittemore AS, eds. Environmental epidemiology: risk assessment. Philadelphia: Society of Industrial and Applied Mathematics, 1982:175-192. RODGERS WL. Estimable functions of age, period, and cohort effects. Am Sue Rev 1982;47:774-787. US Federal Trade Commission. A report to Congress pursuant to the Federal Cigarette Labeling and Advertising Act for the year 1981. July, 1984. American Cancer Society. U.S. tar/nicotine levels dropping. World Smoking and Health 1981;6:47. PEACE LR. A time correlation between cigarette smoking and lung cancer. The Statistician 1985;34:371-381. PICKLE LW, CORREA P, FONTHAM E. Recent case-control studies of lung cancer in the United States. In: Mizell M, Correa P, eds. Lung cancer, causes and preven- tion. Deerfield Beach, FL: Verlag Chemic Int, 1984:101-115. GARFINKEL L. Time trends in lung cancer mortality among nonsmokers and a note on passive smoking. JNCI 1981;66:1061-1066. LEE PN. Lung cancer incidence and type of cigarette smoked. In: Mizell M, Correa P, eds. Lung cancer: causes and prevention. Deerfield Beach, FL: Veriag Chemic Int, 1984:273-283. LUBIN JH, BLOT WJ, BERRINO F, et al. Modifying risk of developing lung cancer by changing habits of cigarette smoking. Br Med J I984;288:1953-1956. LUBIN JH, BLOT WJ, BERRINO F, et al. Patterns of lung cancer risk according to type of cigarette smoked. Int J Cancer 1984;33:569-576. NELDER JA, WEDDERBURN RWM. Generalized linear models. J R Statist Soc A 1972; 135:370-384. Office of the Assistant Secretary for Health, Office on Smoking and Health, Public Health Service, US Department of Health and Human Services. The changing cigarette: a report of the Surgeon General. Washington, DC: US Govt Print Off, 1981 [DHHS publication No. (PHS)81-50156]. GREENWALD P, SONDIK EJ. Cancer control objectives for the nation: 1985-2000. NCI Monogr 1986;2:3-93. SHOPLAND DR, BROWN C. Area review: current trends in smoking control. Ann Behav Med 1985;7:5-8. 20. 21. 22. 23. 24. 25. 26. 27, 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 2063628178 Vol. 80, No. 1, March 2, 1988

2063628179

'Nicotine, ber 1986, ton, W1N 4BN, and tse, Guy's NICOTINE, SMOKING, AND THE LOW TAR PROGRAMME EDITED BY NICHOLAS WALD and SIR PETER FROGGATT St Bartholomew's Hospital Medical College OXFORD NEW YORK TOKYO .OXFORD UNIVERSITY PRESS 1989

,n d for. character- es as they ned about tes may be :he road to ted to give -~ that they ~ those who me seen as a formed and er. ~cil. iclds of UK try sources. ceightcd tar. :h Journal of .'hold Survey mber ). Relation of )kers. British ftar. nicotine ', 39, 361-4. .', Y. (1986). tar smokcrs. vision act as t preferences /our. HMSO. tar levels in a ion men and 17 The role of nicotine in the tar reduction programme SIR PETER FROGGATT and NICHOLAS J. WALD Abstract In the smoking habit nicotine has both advantages and d&advantages. Nicotine contributes to the apparent pleasure of smoking and it im- proves performance in smokers but an important disadvantage is the dependence it induces. Long-standing speculation that nicotine may be a cause of cardiovascular disease does not seem well-founded. Nicotine may be a co-carcinogen through its possible role in the formation of nitrosamines though the view that it is the nicotine in tobacco smoke that is largely responsible for the cancer induced by smoking is not supported by the epidemiological evidence. It is possible that nicotine may contribute to the anti-oestrogenic effect of smokh~g, though this is uncertain and requires further research. Nicotine, while not the only factor in controlling overall smoking behaviour, is recognized as being an important factor in regulating compensatory smoking. This can be used to advantage by reducing nicotine yields less than tar yields and thereby reducing the intake of tar as yields are gradually reduced. Reduction in tar yields will reduce the incidence of lung cancer and will probably also reduce the incidence of chronic obstructive lung dis- ease, but there is little evidence that it will affect the incidence of ischaemic heart disease. Trend data on smoking and its related diseases support these conclusions. On present evidence nicotine yields should be brought down, beet it would seem that the toxiciO' of cigarettes may be reduced more if nicotine yields are reduced to a lesser extent than tar yields. To the smoker nicotine confers undoubted benefits though it also has disadvantages. The benefits are well documented. There is no doubt that nicotine improves performance in smokers. It may also do so in non- smokers as evidenced, for example, by the observation cited by Jarvis'in Chapter 12 that nicotine administered nasally improves non-smokers' per- formance on a simple task of psychomotor speed, the enhancement being blocked by mecamylamine, a nicotine antagonist. Nicotine relieves stress,

230 Sir Peter Froggatt and Nicholas Wald enhances mood, and improves concentration at least in smokers. Further- more, the evidence shows that nicotine plays an important role in in- fluencing "compensatory smoking'. It follows that by maintaining the nico- tine yields of cigarettes while pari passu lowering the tar yield, tar intake can be reduced more than by lowering nicotine and tar in equal propor- tions. Understanding the determinants of compensatory smoking in this way is important in judging the health effects of changing cigarette yields. While compensation occurs it is reassuring to know that it is not complete, thus implying that while tar reduction may not produce its benefits to the extent which might initially have been supposed, it does, nonetheless, lead to a reduced tar intake. What are the disadvantages of nicotine? One is the dependence it produces in smokers and this can be considerable; tobacco smoking is still the most popular mode of nicotine delivery. The dependence on nicotine is a two-edged sword since the fact that people smoke, in part, for nicotine can be argued both as a reason to maintain nicotine in a tar reduction programme (to reduce compensation), and to lower it in order to wean the smoking population off the habit or discourage the development of dependence in new smokers--a particularly relevant point with the young. Certainly, this latter point argues that nicotine yields be not increased. The balance of the argument is. we believe, on present knowledge in favour of the view that nicotine yiel.ds be reduced less drastically than tar yields. From the toxicological point of view nicotine has been mainly considered in relation to two groups of disorders, namely cardiovascular disease and, more recently, cancer. That nicotine has a role in the cause of cardiovascu- lar disease has its adherents, but the evidence is not compelling. The fact that pipe smokers, who have high intakes of nicotine, do not have a materially increased risk of ischaemic heart disease, on the face of it dismisses chronic exposure to nicotine as a significant cardiovascular hazard. The possibility that acute nicotine exposure (by. say, "pulse-dosing' through deep inhaling) might be hazardous is likewise largely dismissed by the observation that smokers who have fatal heart attacks do not h'.ave them disproportionately after the first or second puff frofn the first cigar- ette of the day. Moreover, nicotine is not an agent which can induce chronic arterial damage in animals. A major focus of attention in these Proceedings is the possibility that nicotine may act as a co-carcinogen. Nicotine itself is not genotoxic. There is no laboratory evidence that it is a carcinogen or that it enhances the activity of known carcinogens. On the other hand, nicotine alkaloids present in tobacco can be nitrosated to produce nitrosamines for which there is conclusive evidence of carcinogenicity in animals. Nitrosamines are produced when secondary and tertiary amines are nitrosated. Nicotine is a tertiary amine and the tobacco alkaloids nornico- The role of, tine, anabasine, and ~ tobacco results from r~ derived from the nitrat the use of nitrogenous quantities in the roots a arise during tobacco cu one-third of nitrosamin~ to two-thirds are sYnth lysis).* It is likely, ther nitrate concentration of tertiary amines, or in increase in the total de good evidence that the nitrosamines in tobacco evidence that the nicotin Hoffmann in his paper p~ to a cigarette can increas the principal tobacco-sp~ nicotine yield of a cigar~ amines is still unclear, more influenced by nitra On general grounds, amine, such as nicotine, yield of total nitrosamine to be small in compari: nitrosamines, Thus, US samine yields than Britist and French cigarettes h~ filter cigarette will yield methyl-N-nitrosamino)- nitrosoanatabine); and F~ filter cigarette has yields, cigar yields nearly l0 tim for the 10-fold variation i cigarettes or the 10-fold c US cigarettes. Also, if th greater than that of UK and Britain similar amo amount, and who startec would be wise to engine~ *It is also possible that son absorbed nicotine, though this

ers. Further- role in in- ing the nico- d, tar intake quat propor- oking in this arette yields. ~ot complete, .merits to the :theless, lead 2pendence it noking is still on nicotine is • for nicotine tar reduction rder to wean celopment of th the young. ~creased. The e in fiivour of tar yields. dy considered r disease and. f cardiovascu- ling. The ~:Oa not have a e face of it rdiovascular "pulse-dosing" y dismissed bv ~ do not have the first cigar- :h can induce ~ossibility that lotoxic. There enhances the ,tine alkaloids ines for which The role of nicotine in the tar reduction programme 231 tine, anabasine, and anatabine are secondary amines. Nitrosation in tobacco results from reaction with nitrite and nitrogen dioxide which is derived from the nitrate in the tobacco which, in turn, is associated with the use of nitrogenous fertilizers. The nitrate is present in the greatest quantities in the roots and stems of the tobacco leaf and the nitrosamines arise during tobacco curing and processing. In the tobacco smoke about one-third of nitrosamines come directly from the tobacco while perhaps up to two-thirds are synthesized during the burning of the tobacco. (pyro- lysis).* It is likely, therefore, that factors that lead to an increase in the nitrate concentration of tobacco, or in the concentration of secondary and tertiary amines, or in the yield of nitrogen dioxide will all lead to an increase in the total delivery of tobacco specific nit.rosamines. There is good evidence that the nitrate content of tobacco affects the levels of nitrosamines in tobacco products and in smoke. There is, however, less evidence that the nicotine yield of a cigarette has a similar effect although Hoffmann in his paper presented data to show that the addition of nicotine to a cigarette can increase its yield of N'-nitrosonornicotine (NNN), one of the principal tobacco-specific nitrosamines. The relationship between the nicotine yield of a cigarette and its total yield of tobacco-specific nitros- amines is still unclear, though nitrosamine delivery appears to be much more influenced by nitrate yield than by nicotine yield. On general grounds, we must accept that the increase in a secondary amine, such as nicotine, in tobacco is likely to lead to an increase in the yield of total nitrosamines, though the magnitude of such an effect is likely to be small in comparison with other factors influencing the yield of nitrosamines. Thus, US and French cigarettes have much higher nitro- samine yields than Briti.sh cigarettes. US tobacco has a high nitrate content and French cigarettes have high yields of nitrogen oxides. A typical US filter cigarette will yield about 150ng of NNN, 200ng of NNK [4-(N- methyl-N-nitrosamino)-l-(3-pyridyl)-l-butanone], and 150ng NAT (N'- nitrosoanatabine); and French cigarettes have similar yields. A typical UK filter cigarette has yields only about one-tenth of these levels while a small cigar yields nearly 10 times more. The yield of nicotine could not account for the 10-fold variation in nitrosamine delivery between, say, US and UK cigarettes or the 10-fold difference between the yield from small cigars and US cigarettes. Also, if the relative toxicity of US cigarettes were 10 times greater than that of UK cigarettes why is the risk of lung cancer in the US and Britain similar among men of the same age who smoke the same amount, and who started to smoke at the same age? Unquestionably, it" would be wise to engineer cigarettes to reduce nitrosamine delivery, but ry amines arc *It is also possible that some nitrosamine can come from metabolism in the body of lloids nornico- absorbed nicotine, though this has not been firmly demonstrated.

232 Sir Peter Froggatt and Nicholas Wald altering the nicotine yield is not the best way to accomplish this. In any case, we must not forget that whatever toxic effects tobacco:specific nitros- amines may have, the carcinogenicity of tobacco smoke cannot be satisfac- torily explained in terms of its nitrosamine delivery. It is possible, though not yet proven, that nicotine may contribute to the anti-oestrogenic effect of tobacco smoking and the consequential effect on the risk of oestrogen-related disorders among smokers. Clarifying the role of nicotine is important since it may represent the most serious toxic effect of nicotine in humans. At present, there is too little information upon which to make a judgement and the position must be re-examined when further data are available. We may therefore say that, at present, there ~i-e no strong reasons for believing that nicotine is an important toxic component of tobacco smoke though we cannot completely exclude the possibility that it may play a role in co-carcinogenesis and in the development of hormone-related disease, Nicotine, though an important factor in regulating compensatory smok- ing and therefore a determinant of tar intake, is not the only factor. It is clear from the presentations at the Symposium that compensation is not due to a single mechanism and, indeed, nicotine does not determine all aspects of smoking behaviour as evidenced by the large number of smokers who manage to give up smoking altogether and by the observation that in certain countries, such as West Germany, cigarettes with relatively low nicotine yields are widely used. It is possible that people starting for the first time to smoke may be satisfied by a relatively low nicotine delivery cigarette, whereas established smokers who have got used to a higher nicotine delivery cigarette could not easily or immediately accept less nicotine. Table 17.1. Trends in lung cancer mortality ratios bv sex and age 1950-84 (England and Wales) Sex Year Mortality ratios for age group 30-34 40-44 50-54 60-64 7(I-74 195(I-54 1.00 1.00 l.ll(I l.(lO 1960-64 (I.97 11.91 1.03 1.52 2.(14 197(I-74 0.65 (I.75 0.89 1.49 2.88 198(I-84 (I.35 (1.48 (I.66 1.26 2.72 195(I-54 1 .(1() 1.00 1.00 1.00 1.00 1960-64 (I.73 1.33 1.49 1.49 1.40 1970-74 0.53 1.35 2.33 2.34 2.26 198(I-84 (I.47 1.13 2.35 3.45 3.56 The datt especially persistence girls and p~ consumptk percentage smoking p~ designed to smoking-re of epidemic the reducti, younger ag would have Tables 1- years (an a habits than

~this. In any ecific nitros- :annot be satisfac- J contribute to the • quential effect on 21arifying the role erious toxic effect information upon e-examined when ;trong reasons for of tobacco smoke it may play a role e-related disease. npensatory smok- : only factor. It is npcnsation is not not determine all umber of smokers bservation that in ith relatively low le starting for the nicotine delivery used to a higher lately accept less : and age 195II-84 group 60-64 70-74 1.00 1.011 1.52 2.O4 1.49 2.88 1.26 2.72 I .l)O I 1.49 1.40 2.34 2.21~ 3.45 3.56 The role of nicotine in the tar reduction programme 233 Table 17.2. Trends in weekly consumption of manu- factured cigarettes per person 1950-84 (Great Britain, sales adjusted) Sex Year Age 30-34 35-49 60 + Men Women 1950-54 83 84** 39 1960-64 79 87 51 1970-74 84 87 50 1980-84 56* 56 32*** 1950-54 36 26** 9 1960-64 42 46 13 1970-74 60 59 18 1980-84 46* 51 17"** *25-34; *'35-59; ***65 +. The data presented in Chapter 5 on smoking trends are encouraging, especially those in men and young women. The major concern is the persistence of relatively high rates of cigarette smoking among teenage girls and persons in the lower socio-economic groups. The decline in the consumption of cigarettes, both expressed on a per person basis and as a percentage o'f smokers in the general population, shows that a low tar smoking policy can be successfully implemented pari passu with a policy designed to reduce smoking in general. This is important because trends in smoking-related diseases support the view. principally based on the results of epidemiological studies, that the reduction in tar yields is associated with the reduction in lung cancer. This effect has been most marked in the younger age groups in which exposure to high tar cigarettes in the past would have been less than in older smokers. Tables 17.1-17.3 show how lung cancer mortality in men aged 30-34 years (an age group young enough to be less affected by past smoking habits than older men) in 1980-84 were only one-third of the mortality 30 Table 17.3. Trends in annual sales- weighted tar yield (rag/cigarette) (United Kingdom) Year Annual sales weighted tar yield (mg/cig) 1948-54 30 1962-68 26 1970-74 21 1980-84 16

234 Sir Peter Froggatt and Nicholas Wald years before, while cigarette consumption declined to two-thirds and tar yields to nearly one-half.* The decline in lung cancer can not be satisfactor- ily explained by the fall in cigarette consumption alone, but it can by the fall in tar-weighted cigarette consumption. CompensatorY smoking might have been expected to attenuate the fall in tar yields, but there are grounds for believing that increases in the extent of inhaling probably leads to relatively less deposition of smoke particles on the proximal bronchial airways and, hence, less lung cancer than might otherwise be expected (though not necessarily less than would be expected with diseases that may be associated with the contact or absorption of smoke particles in the peripheral parts of the lung). The association between tar-weighted cigarette consumption and lung cancer in young British men provides important support for the view that the reduction in tar yields hav~ been beneficial. The reduction in lung cancer in young women supports this thesis, but the reduction appears greater than would be expected from the decline in tar-weighted cigarette consumption--lung cancer fell to about half over the same 30-year period, while tar-weighted cigarette consumption declined to about two-thirds (28 per cent increase in consumption and a 47 per cent decrease in tar yields). While the reason for this striking reduction in lung cancer is not clear it may have arisen because the decline in tar yields in young women may have been greater than the average reduction for the population as a whole. It may also have been due to other environmental changes, notably improvements in the control of air pollution in Britain during the 1950s. The percentage reduction in lung cancer that may have arisen as a result of cleaner air is likely to be more apparent in young women with their low background rates than in men whose rates were much higher. There is also evidence, though less compelling, that reducing tar yields has reduced the risk of chronic obstructive lung disease. On the other hand, the data on ischaemic heart disease and aortic aneurysm are equi- vocal. However, on the basis of epidemiological studies (not cited in these Proceedings) and on the basis of the trends (that are described), there is no evidence that the prevalences of these diseases have been detrimentally affected by decreases in tar yields. Thus, the conclusion viz, that the product modification programme has resulted in net benefits, is sound and supported by the evidence. It is, of course, possible that factors other than tar yields and the preval- ence of cigarette smoking have contributed to the reduction in lung cancer, for example a reduction in the specific carcinogenicity of tobacco smoke. The accompanying table, however, which shows declines in lung cancer in younger age groups over some 30 years reflecting declines in tar yield, *This assumes that the secular change in tar yields affected all age groups to the same extent: it probably had a greater impact on the young than the old.

.................................... ' .................................. ~ ...................................... [111[11 I111111111111111 ~ ..... /11111111 [11 I Illfllll ~u~ , wo-thirds and tar not be satisfactor- but it can by the ry smoking might there are grounds )robably leads to • oximal bronchial wise be expected diseases that may e particles in the ~mption and lung for the view that reduction in lung -eduction appears veighted cigarette ~e 30-year period, out two-thirds (28 :ase in tar yields). ~cer is not clear it sung women may • . population as a 1 changes, notably .ring the 1950s. as a result of ~en with their low dgher. • educing tar yields tse. On the other leurysm are equi- (not cited in these • ribed), there is no ,een detrimentally ~ion viz, that the efits, is sound and tds and the preval- ion in lung cancer, ~f tobacco smoke. s in lung cancer in :lines in tar yield, ge groups to the same The role of nicotine in the tar reduction programme 235 suggests that it is not necessary to invoke such an explanation though, clearly, it cannot be excluded solely from the trend data. Several speakers during the Symposium expressed the view that many smokers who switch to low tar cigarettes do so preparatory to stopping smoking altogether, and that this was not simply an intermediate step in a self-selected regimen towards giving up the habit, but could in itself actively help the smoker who wants to stop smoking achieve this goal. If these views are substantiated then a more active policy of inducing smokers onto low tar brands would be justified; for the last flve years the proportion of smokers smoking low tar brands in the UK has plateaued at 12-15 per cent. The low tar policy would then be a useful step, not only in reducing the hazard to continuing smokers, but also as a way of converting such smokers into ex-smokers. o r...o 0~

---

N.ENGL J MED V.320 NO.24 PP.1619-1621 YEAR:I989 COPIES ORDERED: 1 BENOWITZ N.L. HEALTH AND PUBLIC... 33332 WORLDWIDE SCIENTIFIC AFFAIRS 02038709 ADVANCED INFORMATION CONSULTANTS 1200 BROAD STREET DURHAM, NC 27705 Authorized by : Lynn Larson 804-274-3642 -L NIA 8.25.97 Do Not Fax Max Fee: $75 Thank you for choosing Advanced Information Consultants for your document delivery needs U-Team NC/MI Mail (Use FedEx for courier .ship.ments, do not use Airborne) VenOor Cost Trans Type • $ Ve.fica~on Oel C~arge Ship IDaze

"Vol. 320 No. 24 EDITORIALS ". 1619 The New England Journal of Medicine Owned and Published by the Massachusetts Medical Society William 13. Lavelle, M.D. President William M.~McDermott, Jr., M.D. Charles S. Amorosino, Jr. Executive Vice President Executive Secretary THE COMMITTEE ON PUBLICATIONS OF TrlE MASSACHUSETTS MEDmAL SOCIETY james F. McDonough, M.D., Chairman Henry H. Banks, M.D. Edward E. Jacobs, Jr., M.D. Frank E. Bixby, Jr., M.D. Brian J. McKinnon Howard M. Ecker, M.D. Daniel Miller, M.D. Howard Epstein, M.D.Percy W. Wadman, M.D. Arnold S. Relman, M.D., EDITOR-IN-CHIEF Marcia Angell, M.D., EXECUTIVE EDITOR Edwin W. Salzman, M.D., DEeUTY EDrrOR Gregory D. Curfman, M.D., DE~,UrY EDITOR Edward W. Campion, M.D., DEPUTY EDITOR Robert D. Utiger, M.D., DEPUTY EDITOR ASSOCIATE EDITORS Jane F. Desforges, M.D. - Norman K. Hollenberg, M.D., Ph.D. Ronald A. Malt, M.D. Morton N. Swartz, M.D. Franklin H. Epstein, M.D. Francis D. Moore, M.D., Boox P, Xv~E'~ EDITOR John (3. Bailar, III, M.D., Waiter Willett, M.D., STATISTICAL (~ONSULTANT~ John K. Iglchart, NATIONAL CORRESPONDENT Marlene A. Thayer, EDITORIAL OFFrCE MANAGER Stephen E. Cinto, MANAGER OF EDITORIAL PRODUCTION Lorraine W. Loviglio, MANAGER OF MANUSCRIPT EDITING EDITORIAL BOARD Eugene Braunwald, M.D. Robert J. Mayer, M.D. Aram V. Chobanian, M.D. Kenneth McIntosh, M.D. Theodore Colton, Sc.D. David G. Nathan, M.D. Richard H. Egdahl, M.D. Lawrence G. Raisz, M.D. John T. Harrington, M.D. Kenneth J. Rothman, Dr.P.H. Homayoun Kazemi, M.D. Thomas J. Ryan, M.D. EDITORIAL OFFICE Nancy A. Brady, Editorial Production Assistant; Helen Connors, Research Assistant; Karen M. Daly, Editorial Assistant; Briana boherty, Editorial Assistant; Kathleen Eagan, Manuscript Assist- ant; Dale R. Golden, Editorial Assistant; Kate L. Haas, Editorial Production Assistant; Christie L. Hager, Editorial Assistant; Re- becca H. Hale, Editorial Assistant; Susan L. Kaplan, Editorial Pro- duction Layout Artist; David F. March, Manuscript Editor; Sandra S. McLean, Manuscript Editor; Brian Middleton, Editorial Assist- ant; Henry S. Miller, Jr., Manuscript Editor; Stephen Morrissey, Manuscript Editor; Sylvia L. Parsons, Editorial Assistant; Marilyn Seaqulst, Receptionist; Deborah A. Stone, Senior Editorial Production Coordinator. Frederick Bowes, IIl, DIRECTOR OF PUBLISHING OPERATIONS Ann Relnke Strong, DEptrry DiP~c'roa HEALTH AND PUBLIC POLICY IMPLICATIONS OF THE "LOW YIELD" CIGARETTE BETWEEN 1955 and 1987 the average tar and nico- tine yield of American cigarettes declined substantial- ly. The average tar yield (weighted for the volume of sales of the cigarette) fell from 34 to 13 rag, and the average nicotine yield declined from 2 to 0.9 mg. Ad- vertisements for "low yield" cigarettes often imply that the health hazards associated with smoking these c{garettes are less than_the hazards associated with higher-yield cigarettes. The case-control study by Palmer and coworkers reported in this issue of the Journal shows that modern low-yield cigarettes do not reduce the risk of nonfatal myocardial infarction among women smokers under 65 years old) Similar data have been reported for memo It seems clear that the hazards of coronary heart disease are not reduced by smoking low-yield rather than high-yield cigarettes. However, there, is evidence that smoking the low-yield cigarette may af- fect the overall risks of adverse health effects in a pop- ulation Of smokers. The implications of these findings for physicians and public policy makers are the subject of this editorial. Yields are determined by analyzing the smoke produced when a machine consumes a cigarette, using specific "puffing" characteristics. In the United States, a 35-ml-puff is taken over a period of two seconds, and one puff is taken every minute until the cigarette has burned to a specific length. Cigarette testing was performed by the Federal Trade Commis- sion between 1967 and 1987. The commission began testing to deal with the competing advertising claims of tobacco companies concerning tar yields. Govern- mental testing was discontinued for economic and oth- er reasons. Cigarette manufacturers, overseen by the Federal Trade Commission, now undertake testing on a voluntary basis. Historically, the first and most important step in reducing tar and nicotine yields was the addition to cigarettes of a filter tip that selectively removes Pl~osPEcrrv~ authors should consult "Information for Authors," which ap- pears in the first issue of each month and may be obtained from the Journal Editorial Office (address below). -ARTXCLES with original material are accepted for consideration with the understanding that, except for abstracts, no part of the data has been pub- lished, or will be submitted for publication elsewhere, before appearing here. Normr.s should be sent at least 30 days before publication date. THE Journal does not hold itself responsible for statements made by any contributor. Statements or opinions expressed in the Journal reflect the views of the author(s) and not the official policy of the Massachusetts Medical Society unless so stated. AL'rHOUCrl all advertising material is expected to conform to ethical stand- ards, acceptance does not imply endorsement by the Journal. MATernaL printed in the Jou,,u~t is covered by copyright. No part of this publication may be reproduced or transmitted in any form without written permission. FoR information on subscriptions, permissions, reprints, and other services see the "Business Information for Readers" page preceding the Classified Advertising section. EmTOgtAL OFFtOmS: 10 Shattuck St., Boston, MA 09115-6094. Telephone: (617) 734-9800. FAX: (617) 734-4457. Bo~Nr.Ss, Su~cau~r~or~ Oranc~s: 1440 Main St., Waltham, MA 02154-1649.

1620 THE NEW ENGLAND JOURNAL OF MEDICINE June 15, 1989 these elements (but not carbon monoxide or other gas- eous components) from tobacco smoke. More recent engineering refinements to d~crease tar and nicotine yields include the use of reconstituted sheet tobacco containing larger amounts of stems, which has less nicotine; expanded or puffed tobacco, which results in less tobacco per cigarette; faster burning times, more porous paper, and longer filter overwraps, which cause the smoking machine to take fewer puffs per cigarette; and ventilated filters that allow dilution of the tobacco smoke with air. It is important to recog- nize that modern low-yield cigarettes contain the same type of tobacco and the same amount of nicotine by weight as higher-yield cigarettes.~ Thus, the low- yield cigarette is not low in yield because it contains less of anything, but because it is engineered to make less smoke available to the smoker (or at least to the smoking machine). However, people do not smoke the way machines do. Most smokers are addicted to nicotine; they tend to compensate for their lower-yield cigarettes by smoking them in such a way as to opdmize the intake of nicotine.* By taking more frequent puffs, inhaling more deeply, occluding the ventilation holes with lips or fingers, and smoking more cigarettes, people take in considerably more tar, nicotine, and carbon monoxide than would be predicted by smoking machines. Stud- ies of people who smoked their own selected brands of higher- or lower-yield cigarettes indicate similar or only slightly lower levels of cotinine (a metabolite of nicotine commonly used as a marker of nicotine in- take) or carbon monoxide in smokers of all but possi- bly the lowest-yield cigarette (1 mg of tar).3,5-8 De- spite a great deal of promotion and advertising, very-low-yield cigarettes (1 to 3 mg of tar) are not very popular and account for only a small percentage of sales,9 presumably because most smokers do not ob- tain enough nicotine to find them satisfying. In gener- al, low-yield cigarettes do have a lower rado of tar to nicotine yield as tested by smoking machines, which has suggested that even if smokers compensate for nicotine, their exposure to tar will be reduced.7 Unfor- tunately, intensively smoking low-yield cigarettes in- creases the tar-to-nicotine ratio and reduces or even negates any possible benefit of selective differences in yield.5 Epidemiologic data indicate that low-yield ciga- rettes are less hazardous than high-yield cigarettes with respect to lung, laryngeal, esophageal, and other cancers and possibly chronic obstructive lung dis- ease.m-t2 However, it is important to recognize that the definition of "low yield" has changed over the years. A low-yield cigarette in the 1960s (18 mg of tar or less) would be a high-yield cigarette today. Many of the older cigarettes were unfiltered. Not only was their yield of tar much greater, but the tar was also qualita- tively more toxic than that in modern cigarettes. As is appropriate when one is studying diseases that may take 20 years or longer to develop, most epidemiologic studies of cancer and lung disease have used the older cigarettes in their comparisons. Studies indicate that the risk of lung cancer is reduced substantially (by 20 to 40 percent) in smokers of the old-style low-yield as compared with the old-style high-yield cigarettes; however, that risk is still markedly higher than the risk in nonsmokers.t°-12 Similar results have recently been reported for laryngeal, esophageal, and other cancers.1~ The data on chronic lung disease are less clear. Som~ studies suggest a reduction in cough and phlegm, fewer deaths due to emphysema, or slightly less seriously impaired pulmonary function in smokers of filtered as compared with unfiltered cigarettes,m'~ Other stfidies find no difference in lung function as related to cigarette yield. ~3 It is noteworthy that the greatest reduction in lung cancer among people who switch from unfiltered to filtered' cigarettes is among those who do not increase the number of cigarettes they smoke per day. 14 The risk in those who compen- sate by smoking more than 10 additional filtered ciga- rettes a day is as great as or greater than their risk when they smoked unfiltered cigarettes. For myocardial infarction, studies comparifig smok- ers of high-yield and low-yield cigarettes, either old style or modern, show no evidence of a difference in disease risk.L2'mJ| We can draw valid.conclusions concerning modern cigarettes and the risk of acute myocardial infarction or sudden death, because these events are closely related to current smoking habits. The risk of these events diminishes within a year or less of stopping smoking, so the brand most recently smoked is likely to influence disease risk. AlthOugh conclusions about the relative risks of modern high- yield and low-yield cigarettes cannot yet be reached for cancer or chronic lung disease, studies using bio- chemical markers of nicotine or smoke intake in peo- ple who smoke various brands indicate only small dif- ferences in exposure to the toxins of tobacco smoke. The expected reduction in disease risk for a person who smokes a low-yield cigarette is small, although the consequences of a small reduction in a population of smokers could be considerable. On balance, the movement toward low-yield ciga- rettes has been worthwhile, although in reducing the risk of disease it may have reached the limit. There are, however, potential risks in encouraging the smok- ing of low-yield cigarettes. The availability of low- yield cigarettes may make it easier for adolescents to begin smoking. Additives, which elahance the flavor of low-yield cigarettes, may be harmful, although no data concerning this issue are available. Most impor- tant, information about low-yield cigarettes may be used to convince people that smoking is not as hazard- ous as it once was. As a result, some smokers may switch to low-yield cigarettes rather than quit. The implications of the low-yield cigarette differ for physicians and public health planners. The benefits for any person of smoking low-yield rather than high- yield cigarettes are small, and the benefits of quitting are great. Physicians should give their patients the unequivocal message that low-yield cigarettes are not safe cigarettes. The only reliable way to reduce the adverse health consequences of smoking is to stop.

EDITORIaLs .... 1621 4. 5. 6. the perspective of public health~ however, the movement toward low-yield cigarettes makes sense. There has been considerable progress in reducing the prevalence of smoking in the United States, Canada, and many European countries, but smoking rates are much higher in other parts of the world, and the ciga- rettes smoked in many other countries have a much higher yield than their American counterparts. A worldwide attempt should be made to reduce the yields of toxic substances and to make the yield of all cigarettes as low as possible. Public health policy should encourage smokers who have not yet quit to smoke cigarettes with the lowest possible yield. The yidds of American cigarettes should not be allowed to drift higher as research finds that low-yield cigarettes are not less hazardous. Mandated ceilings for tar, car- bon monoxide, and other toxic components of tobacco smoke that could be lowered gradually over the years, or a progressive tax on higher-yield cigarettes, are logical ways to implement.such goals. " $m Francisco General Hospital San Fra~isco, CA 94410 NEAL L. BENOWITZ, M.D. Palmer JR, Rosenberg L, Shapiro S. "Low yield" cigarettes and the risk of nonfatal myocardial infarction in women. N Engl J Med 1989; 320:1569-73. Kaufman DW, Helmrich SP, Rosenberg L, Miettinan O$, Shapiro S. Nico- tine and carbon monoxide content of cigarette smoke and the risk of myocar- dial infarction in young men. N Engl J Med 1983; 308:409-13. Benowitz NL, Hall SM, Heming R.I, Jacob P I11, Jones RT, Osman A-L. Smokers of low-yield cigarettes do not consume less nicotine. N Engl J Med I983; 309:139-42. Benowitz NL. Pharmacologic aspects of cigarette smoking and nicotine addiction. N Engl J Med 1988; 319:1318-30. Benowitz NL, Jacob P/]/, Yu L, Talcott R, Hall S, Jones RT. Reduced tar, nicotine, and carbon monoxide exposm~ while smoking ulwalow but not low-yield cigarettes. JAMA 1986; 256:241-6. Gori GB, Lynch C.J. Analytical cigarette yields as predictors of smoke bioavailability. Regul Toxieol Pharmacol 1985; 5:31426. 7. Russell MA, Jarvis MJ, Feyerabend C, Saloojee Y. Redaction of tar, nice- line, and carbon monoxide intake in low tar smokers. J Epidemiol Commu- Mty Health 1986; 40:80-5. 8. M~on DJ, Fortmarm SP. Nieot~e yield and measures of cigarerm smoke exposure in a large population: are lower-yield cigarettes safer? Am 2I Public Health 1987; 77:546-9. 9. Kozlowski LT. Evidence for limits on the acceptability of lowest-tar ciga- rette. Am J Public Health 1989; 79:198-9. I0. Department of Health and Human Services. The health consequences of smoking: the changing cigarette: a report of the Surgeon General. Washing- ton, D.C.: Government Printing Office, 1981. (Publi~tion no. DHHS (PHS) 81-50156.) II. Participants of the Fourth Scarboreagh Conference on Preventive Medicine. Is there a future for lower-tar-yield cigarettes? Lancet 1985; 2:1111-4. 12. Kanfman DW, P~Imer JR, Rosenl~a'g L, Stolley P, Wa~haner E, Shapiro $. Tar eonmnt of cigarettes in relation to lung cancer. Am .I Epidemio11989; 129:703-11. 13. Sparrow D, St~fos T, Boss~ R, Weiss ST. The relationship of tar eontefit to decline in pulmonary function in cigarette smokers. Am Rev Respir Dis 1983; 127:56-8. 14. Augusline A, Harris RE, Wynder EL. Compensation ~ a risk factor for lung ear, cer in smoke~s who switch from nonfiltor to lllter cigarettes. Am J Public Health 1989; 79:188-9I. PREDNISONE THERAPY FOR DUCHENNE'S MUSCULAR DYSTROPHY IN the past five years, progress in understanding the molecular basis of Duchenne's muscular dystrophy has been substantial. The affected gene in this disease has been cloned, and its protein product, dystrophin, characterized.1 The importance of dystrophin in the pathogenesis of this disorder has been defined: dystro- phin is absent from muscle in Duchenne's muscular dystrophy and is usually present but of abnormal size in Becker's muscular dystrophy, a milder variant.2 Unfortunately, these dramatic advances have not yet had an effect on the clinical management of muscular dystrophy. Duchenne's muscular dystrophy remains invariably fatal. The disease is common, occurring in approximately I in 3000 male infants. A third of the cases result from new mutations in the dystrophin gene. It is therefore essential to develop effective treat- ment for this disorder. In this issue of the Journal, Mendell and colleagues report that the administration of prednisone in sin- gle doses each day over a six-month period improved the strength of patients with Duchenne's muscular dystrophy) This confirms the results of three pre- vious unrandomized, unblinded studies.4-6 In an ear- lier study, these investigators found prednisone effec- tive in such patients as compared with historickl controls who were only observed.5 They now report similar results of a randomized, blinded trial. Sev- eral factors were evaluated to gauge muscle status, including strength in several muscles, joint contrac- tures, timed functional tests (e.g., the time needed to climb four stairs), overall functional grading of the limbs, and pulmonary-function tests. In all catego- ries except joint contractures, progressive improve- ment was detected at one, two, and three months; the improvement in muscle function was maintained for three to six months. Several points commend this report. It "demon- strates benefit from a therapy for Duchenne's muscu- lar dystrophy in a double-blind, controlled trial. It exemplifies the value of well-executed multicenter col- laboration for rapid, statistically accurate drug trials. It is also a tribute to the Muscular Dystrophy Associ- ation, which has been a dominant force promoting research into the pathogenesis and treatment of the disease. These points notwithstanding, the paper raises sev- eral questions. Perhaps most important, can any trial of steroids at these doses remain truly blinded and free of a placebo effect? The answer is not clear. Certainly, it ~s unlikely that observer bias explains the significant 61inical improvement at one month in this study, since the cushingoid appearance induced by steroids had not developed in most of the prednisone-treated pa- tients by that time. Although the results of many of the functional tests might have been influenced by a placebo effect, it is doubtful that this alone would have accounted for the overall pattern of improve- ment in the prednisone groups. Twenty-four-hour uri- nary creatinine levels, which reflect total muscle mass, increased during prednisone treatment. This observation is objective and presumably independent of any placebo effect. Furthermore, some trials of drugs have not benefited patients with Duchenne's muscular dystrophy,7 indicating that placebo effects

2063628193

STEPHEN, A. "NICOTINE, SMOKING, AND THE LOW TAR PROGRAMME (111 SMOKING YIELDS AND COMPENSATION (8 ESTIMATING THE EXTENT OF COMPENSATORY SMOKING))" BOOK 53; TAB 21 HERE REMOVE WHEN ARTICLE HAS ARRIVED O~ 03 O~

S6~gzgBgOZ

i Vol. 320 No. 24 EDITORIALS 1619 The New England Journal of Medicine Owned and Published by the Massachusetts Medical Society WilIiam G. LaveIle, M.D. President William M-.'McDermott, Jr., M.D. Charles S. Amorosino, Jr. Executive Vice President Executive Secretary THE COMMrFFEE ON PUBLICATIONS OF THE MASSACHUSETTS MEDICAL SOCIETY James F. McDonough, M.D., Chairman Henry H. Banks, M.D. Edward E. Jacobs, Jr., M.D. Frank E. Bixby, Jr., M.D. Brian J. McKinnon Howard M. Ecker, M.D.Daniel Miller, M.D. Howard Epstein, M.D. Percy W. Wadman, M.D. Arnold S. Relman, M.D., EDITOR-IN'CHIEF Marcia Angell, M.D., EXECUTIVE EDITOR Edwin W. Salzman, M.D., DEPUTY EDITOR Gregory D. Curfman, M.D., DEPUTY EDITOR Edward W. Campion, M.D., DEPUTY EDITOR Robert D. Ufiger, M.D., DEPUTY EDITOR ASSOCIATE EDITORS Jane F. Desforges, M.D. - Norman K. Hollenberg, M.D., Ph.D. Ronald A. Malt, M.D. Morton N. Swartz, M.D. Franklin H. Epstein, M.D. Francis D. Moore, M.D., BOOK R~vE~' EDrrOR John C. Bailar, III, M.D., Walter Willett, M.D., STATISTICAL CONSULTANTS John K. Iglehart, NATIONAL CORRESPONDENT Marlene A. Thayer, EDITORIAL OFFICE MANAOER Stephen E. Cinto, MANAGER OF EDITORIAL PRODUCTION Lorraine W. Loviglio, MANAGER OF MANUSCRIPT EDrriNO EDITORIAL BOARD Eugene Braunwald, M.D. Robert J. Mayer, M.D. Aram V. Chobani~n, M.D. Kenneth McIntosh, M.D. Theodore Colton, Sc.D. David G. Nathan, M.D. Richard H. Egdahl, M.D." Lawrence G. Raisz, M.D. John T. Harrington, M.D. Kenneth J. Rothman, Dr.P.H. Homayoun Kazemi, M.D. Thomas J. Ryan, M.D. EDITORIAL OFFICE Nancy A. Brady, Editorial Production Assistant; Helen Connors, Research Assistant; Karen M. Daly, Editorial Assistant; Briana Doherty, Editorial Assistant; Kathleen Eagan, Manuscript Assist- ant; Dale R. Golden, Editorial Assistant; Kate L. Haas, Editorial Production Assistant; Christie L. Hager, Editorial Assistant; Re- becca H. Hale, Editorial Assistant; Susan L. Kaplan, Editorial Pro- duction Layout Artist; David F. March, Manuscript Editor; Sandra 8. McLean. Manuscript Editor; Brian Middleton, Editorial Assist- ant; Henry S. Miller, Jr., Manuscript Editor; Stephen Morrissey, Manuscript Editor; Sylvia L. Parsons, Editorial Assistant; Marilyn Seaquist, Receptionist; Deborah A. Stone, Senior Editorial Production Coordinator. Frederick Bowes, llI, DIRECTOR OF PUBLISHING OPERATIONS Ann Reinke Strong, DEPUTY DIRECTOR HEALTH AND PUBLIC POLICY IMPLICATIONS OF THE "LOW YIELD" CIGARETTE BETWEEN 1955 and 1987 the average tar and nico- tine yield of American cigarettes declined substantial- ly. The average tar yield (weighted for the volume of sales of the cigarette) fell from 34 to 13 rag, and the average nicotine yield declined from 2 to 0.9 rag. Ad- vertisements for "low yield" cigarettes often imply that the health hazards associated with smoking these cigarettes are less than the hazards associated with higher-yield cigarettes. The case-control study by Palmer and coworkers reported in this issue of the Journal shows that modern low-yield cigarettes do not reduce the risk of nonfatal myocardial infarction among women smokers under 65 years old.1 Similar data have been reported for men? It seems clear that the hazards of coronary heart disease are not reduced by smoking low-yield rather than high-yield cigarettes. However, there, is evidence that smoking the low-yield cigarette may af- fect the overall risks of adverse health effects in a pop- ulation of smokers. The implications of these findings for physicians and public policy makers are the subject of this editorial. Yields are determined by analyzing the smoke produced when a machine consumes a cigarette, using specific "puffing" characteristics. In the United States, a 35-ml puff is taken over a period of two seconds, and one puff is taken every minute until the cigarette has burned to a specific length. Cigarette testing was performed by the Federal Trade Commis- sion between 1967 and 1987. The commission began testing to deal with the competing advertising claims of tobacco companies concerning tar yields. Govern- mental testing was discontinued for economic and oth- er reasons. Cigarette manufacturers, overseen by the Federal Trade Commission, now undertake testing on a voluntary basis. Historically, the first and most important step in reducing tar and nicotine yields was the addition to cigarettes of a filter tip that selectively removes PROSPECTIVE authors should consul~ "Information for Authors," which ap- pears in the first issue of each month and may be obtained from the Journal Editorial Office (address below). ARTICLES with original material are accepted for consideration with the understanding that, except for abstracts, no part of the data has been pub- lished, or will be submitted for publication elsewhere, before appearing here. NOTICES should be sent at least 30 days before publication date. THE Journal does not hold itself responsible for statements made by any contributor. Statements or opinions expressed in the journal reflect the views of the author(s) and not the official policy of the Massachusetts Medical Society unless so stated. ALTHOUGH all advertising material is expected to conform to ethical stand- ards, acceptance does not imply endorsement by the Journal. MATERIAl. printed in the joun~al is covered by copyright. No part of this publication may be reproduced or transmitted in any form without written permission. FOR information on subscriptions, permissions, reprints, and other services see the "Business Information for Readers" page preceding the Classified Advertising section. EDITORIAL OFFICES: 10 Shattuck St., Boston, MA 02115-6094. Telephone: (617) 734-9800. FAX: (617) 734-4457. BUSLNESS, SUBSCmPTtON OFFmES: 1440 Main St., Waltham, MA 02154-1649.

1620 THE NEW ENGL, MND JOUt~NAL OF MEDICINE June 15, 1989 these elements (but not carbon monoxide or other gas- eous components) from tobacco smoke. More recent engineering refinements to d6crease tar and nicotine yields include the use of reconstituted sheet tobacco containing larger amounts of stems, which has less nicotine; expanded or puffed tobacco, which results in less tobacco per cigarette; faster burning times, more porous paper, and longer filter overwraps, which cause the smoking machine to take fewer puffs per cigarette; and ventilated filters that allow dilution of the tobacco smoke with air. It is important to recog- nize that modern low-yield cigarettes contain the same type of tobacco and the same amount of nicotine by weight as higher-yield cigarettes.3 Thus, the low- yield cigarette is not low in yield because it contains less of anything, but because it is engineered to make less smoke available to the smoker (or at least to the smoking machine). However, people do not smoke the way machines do. Most smokers are addicted to nicotine; they tend to compensate for their lower-yield c!garettes by smoking them in such a way as to optimize the intake of nicotine.* By taking more frequent puffs, inhaling more deeply, occluding the ventilation holes with lips or fingers, and smoking more cigarettes, people take in considerably more tar, nicotine, and carbon monoxide than would be predicted by smoking machines. Stud- ies of people who smoked their own selected brands of higher- or lower-yield cigarettes indicate similar or only slightly lower levels of cotinine (a metabolite of nicotine commonly used as a marker of nicotine in- take) or carbon monoxide in smokers of all but possi- bly the lowest-yield cigarette (1 mg of tar).a,5"s De- spite a great deal of promotion and advertising, very-low-yield cigarettes (1 to 3 mg of tar) are not very popular and account for only a small percentage of sales,9 presumably because most smokers do not ob- tain enough nicotine to find them satisfying. In gener- al, low-yield ciggrettes do have a lower ratio of tar to nicotine yield as tested by smoking machines, which has suggested that even if smokers compensate for nicotine, their exposure to tar will be reduced.7 Unfor- tunately, intensively smoking low-yield cigarettes in- creases th~ tar-to-nicotine ratio and reduces or even negates any possible benefit of selective differences in yield.5 Epidemiologic data indicate that lnw-v~,q~ ,-;--- re[tes are less hazardous than high-yield" cigarettes with respect to lung, laryngeal, esophageal, and other cancers and possibly chronic obstructive lung dis- ease.I°'z2 However, it is important to recognize that the definition of "low yield" has changed over the years. A low-yield cigarette in the 1960s (18 mg of tar or less) would be a high-yield cigarette today. Many of the older cigarettes were unfiltered. Not only was their yield of tar much greater, but the tar was also qualita- tively more toxic than that in modern cigarettes. As is appropriate when one is studying diseases that may take 20 years or longer to develop, most epidemiologic studies of cancer and lung disease have used the older cigarettes in their comparisons. Studies indicate that the risk of lung cancer is reduced substantiall.v (by 20 to 40 percent) in smokers of the old-style low-yield as compared with the old-style high-yield cigarettes; however, that risk is still markedly higher than the risk in nonsmokers,m'~2 Similar results have recently been reported for laryngeal, esophageal, and other cancers.~ The data on chronic lung disease are less clear. Som~ studies suggest a reduction in cough and phlegm, fewer deaths due to emphysema, or slightly less seriously impaired pulmonary function in smokers of filtered as compared with unfiltered cigarettes.1°'1~ Other studies find no difference in lung function as related to cigarette yield.~a It is noteworthy that the greatest reduction in lung cancer among people who switch from unfiltered to filtered' cigarettes is among those who do not increase the number of cigarettes they smoke per day.~4 The risk in those who compen- sate by smoking more than 10 additional filtered ciga- rettes a day is as great as or greater than their risk when they smoked unfiltered cigarettes. For myocardial infarction, studies comparirig smok- ers of high-yield and low-yield cigarettes, either old style or modern, show no evidence of a difference in disease riskJ''2'~°'~ We can draw valid, conclusions concerning modern cigarettes and the risk of acute myocardial infarction or sudden death, because these events are closely related to current smoking habits. The risk of these events diminishes within a year or less o~ stopping smoking, so the brand most recently smoked is likely to influence disease risk. AlthOugh conclusions about the relative risks of modern high- yield and low-yield cigarettes cannot yet be reached for cancer or chronic lung disease, studies using bio- chemical markers of nicotine or smoke intake in peo- ple who smoke various brands indicate only small dif- ferences in exposure to the toxins of tobacco smoke. The expected reduction in disease risk for a person who smokes a low-yield cigarette is small, although the consequences of a small reduction in a population of smokers could be considerable. On balance, the movement toward low-yield ciga- rettes has been worthwhile, although in reducing the risk of disease it may have reached the limit. There are, however, potential risks in encouraging the smok- ing of low-yield cigarettes. The availability of low- yield cigarettes may make it easier for adolescents to ...~ ............. ~. Add,u~ c~, wi~icn elnnance the flavor ot low-yield cigarettes, may be harmful, although no data concerning this issue are available. Most impor- tant, information about low-yield cigarettes may be used to convince people that smoking is not as hazard- ous as it once was. As a result, some smokers may switch to low-yield cigarettes rather than quit. The implications of the low-yield cigarette differ for physicians and public health planners. The benefits for any person of smoking low-yield rather than high- yield cigarettes are small, and the benefits of quitting are great. Physicians should give their patients the unequivocal message that low-yield cigarettes are not safe cigarettes. The only reliable way to reduce the adverse health consequences of smoking is to stop.

Vol. 320 No. 24 EDITORIALS "-. 1621 the perspective of public health, however, the movement toward low-yield cigarettes makes sense. There has been considerable progress in reducing the prevalence of smoking in the United States, Canada, and many European countries, but smoking rates are much higher in other parts of the world, and the ciga- rettes smoked in many other countries have a much h~her yield than their American counterparts. A worldwide attempt should be made to reduce the yields of toxic substances and to make the yield of all cigarettes as low as possible. Public health policy should encourage smokers who have not yet quit to smoke cigarettes with the lowest possible yield. The yields of American cigarettes should not be allowed to drift higher as research finds that low-yield cigarettes are not less hazardous. Mandated ceilings for tar, car- bon monoxide, and other toxic components of tobacco smoke that could be lowered gradually over the years, or a progressive tax on higher-yield cigarettes, are logical ways to implement such goals. ' San Francisco General Hospital San Francisco, CA 94410 NEAL L. BENOWITZ, M.D. I. Palmer JR, Ros~nberg L, Shapiro S. "Low yield" cigarettes and the risk of nonfatal myocardial infarction in women. N Engi J Med 1989; 320:I 569-73. 2. Kanfrnan DW, Helmrich SP, Rosenberg L, Miettinen OS, Shapiro S. Nico- tine and carbon monoxide content of cigarette smoke and the risk of myocar- dial infarction in young men. N Engi J MeAt 1983; 308:4139-13. Benowitz NL, Hall SM, Heming RI, Jacob PIII, Jones RT, Osman A-L. Smokers of low-yield cigarettes do not consume less nicotine. N Engl J Med 1983; 309:139.42. 4. Banowitz NL. Pharmacologic aspects of eigaret~ smoking and nicotine addiction. N Engl J Med 1988; 319:I318-30. 5. Benowitz NL, Jacob P Ill, Yu L, Taleott R, Hall S, Jones RT. Reduced tar, nicotine, and carbon monoxide exposure while smoking ultralow but not low-yield cigarettes. JAMA 1986; 256:241-6. 6. God GB, Lynch CJ. Analytical cigarette yields as predictors of smoke bioavallability. Regul Toxicol Pharmaeol 1985; 5:314-26. 7. Russell MA, larvis MJ, Feyembend C, Saloojee Y. Reduction of tar, nico- tine, and carbon monoxide intake in low tar smokers. J Epidemiol Commu- nity Health 1986; 40:80-5. 8. Maron DJ, Fortmarm SP, Nicotine yield and measures of cigarette smoke exposure in a large population: are lower-yield cigarettes safer? Am J Public Health 1987; 77:546-9. 9, Kozlowski LT. Evidence for limits on the acceptability of lowest-tar ciga- rette. Am J Public Health 1989; 79:198-9. 10. Department of Health and Human Services. The health consequences of smoking: the changing cJgar~Jte: a report of the Surgeon General. Washing- ton, D.C.: Government Printing Office, 1981. (Publication no. DHHS (PHS) 81-50156.) 11. Participants of the Fourth Scarborough Conference on Preventive Medicine. Is there a future for lower-tar-yield cigarettes? Lancet 1985; 2:1111-4. 12. Kanfman DW, Palmer JR, Rosenberg L, Stolley P, Warshauer E, Shapiro S. Tar content of cigarettes in relation to lung cancer. Am J Epidemiol 1989; 129:703-11. 13. Sparrow D, Stefus T, Boss6 R, Weiss ST. The relationship of tar content to decline in pulmonary function in cigarette smokers. Am Rev Respir Dis 1983; 127:56-8. 14. Augustine A, Harris RE, Wyadar EL. Compensation as a risk factor for lung cancer in smokers who switch from nonfilter to filter cigarettes. Am J Public Health 1989; 79:188-91, PREDNISONE THERAPY FOR DUCHENNE'S MUSCULAR DYSTROPHY IN the past five years, progress in understanding the molecular basis of Duchenne's muscular dystrophy has been substantial. The affected gene in this disease has been cloned, and its protein product, dystrophin, characterized.1 The importance of dystrophin in the pathogenesis of this disorder has been defined: dystro- phin is absent from muscle in Duchenne's muscular dystrophy and is usually present but of abnormal size in Becker's muscular dystrophy, a milder variant,z Unfortunately, these dramatic advances have not yet had an effect on the clinical management of muscular dystrophy. Duchenne's muscular dystrophy remains invariably fatal. The disease is common, occurring in approximately 1 in 3000 male infants. A third of the cases result from new mutations in the dystrophin gene. It is therefore essential to develop effective treat- ment for this disorder. In this issue of the Journal, Mendell and colleagues report that the administration of prednisone in sin- gle doses each day over a six-month period improved the strength of patients with Duchenne's muscular dystrophy,s This confirms the results of three pre- vious unrandomized, unblinded studies.~-6 In an ear- lier study, these investigators found prednisone effec- tire in such patients as compared with historical controls who were only observed.5 They now report similar results of a randomized, blinded trial. Sev- eral factors were evaluated to gauge muscle status, including strength in several muscles, joint contrac- tures, timed functional tests (e.g., the time needed to climb four stairs), overall functional grading of the limbs, and pulmonary-function tests. In all catego- ries except joint contractures, progressive improve- ment was detected at one, two, and three months; the improvement in muscle function was maintained for three to six months. Several points commend this report. Itdemon- strates benefit from a therapy for Duchenne's muscu- lar dystrophy in a double-blind, controlled trial. It exemplifies the value of well-executed multicenter col- laboration for rapid, statistically accurate drug trials. It is also a tribute to the Muscular Dystrophy Associ- ation, which has been a dominant force promoting research into the pathogenesis and treatment of the disease.. These points notwithstanding, the paper raises sev- eral questions. Perhaps most important, can any trial of steroids at these doses remain truly blinded and free of a placebo effect? The answer is not clear. Certainly, it is unlikely that observer bias explains the significant clinical improvement at one month in this study, since the cushingoid appearance induced by steroids had not developed in most of the prednisone-treated pa- tients by that time. Although the results of many of the functional tests might have been influenced by a placebo effect, it is doubtful that this alone would .have accounted for the overall pattern of improve- ment in the prednisone groups. Twenty-four-hour uri- nary creatinine levels, which reflect total muscle mass, increased during prednisone treatment. This observation is objective and presumably independent of any placebo effect. Furthermore, some trials of drugs have not benefited patients with Duchenne's muscular dystrophy,7 indicating that placebo effects

206~628t~9

International Journal of Epidemiology (~) rnternational Epidemiological Association 1989 LEADING ARTICLES • Determinants of Policy on Smoking • andHealth "": " "= " Vol. 18, No. 1 Printed in Great Britain PETER FROGGATT The opinions expressed in this article are personal and do not necessarily reflect the views of the Independent ,~cientific Committee o.n Smoking and Health o~ any. of its'othermeml~ers. ~ .....: . .. ".. Smoking cigarettes is arguably the greatest public health hazard in developed countries and may become so in much of the developing world. The International Agency for Research on Cancer (IARC) identifies some 30 diseases or ~roups within ICD rubrics as being positively associated with smoking, most causally.~ 'Excess deaths' attributabl.e, to smoking run into tens of thousands every year in the UK alone. Faced with this scourge governments have appeared to act with an almost uniform timidity: 'paltry and hesitating' even in the measured language of the Royal College of Physicians.2 Many critics however do not understand the factors which government weigh in formulating policies. In this article I try to describe simply these often competing determinants, and government's response to them, since the results of the early case-control studies on lung cancer and cigarette smoking3-9 became parliamentary currency in 1951.1° I deal exclusively with the UK where I have been concerned in the government's principal scientific advisory machinery as a member, and from 1981 chairman, of the Independent Scientific Committee on Smoking and Health. Comparison with practices in other countries would be instructive but is beyond the scope of this review. THE VIEW FROM GOVERNMENT The mounting evidence during the early. 1950s i~t~rim.i~aating .cigarette svhokin~, in il.ung canc.e.r, set diffic.ult problems for government, .the tobacco industry, the medical profession, and (not leagt) the smoker. The critical factors were seen to be: (i) smoking was widespread: in 1950, 77 % of men and 38 % of women s.moked, mostly (in the case of women exclusively) cigarettes,11 (ii) smoking was not a passing fad but a widely accepted and growing habit, 3 Strangf6rd Avenue, Belfast BID 6PG, N Ireland,, UK. (iii) tobacco products were advertised without restraint and lawfully marketed to all over 15, (iv) the public favoured strong, non-filtered cigarette.s, (v) the irritant properties and alien aroma of tobacco smoke were offensive, but not harmful to non-smokers, (vi) death rates from lung cancer were increasing sharply, especially in men; and (vii) cigarette smoking and lung cancer risk were seemingly dose-related with no threshold effect. Medical and some lay opinion considered that these demanded action; it was less certain what this action should be. Government on the other hand was instinctively cautious since each of the above had a wider, political, perspective. Thus (i) above, to the doctor an index of high and increasing public health risk, was to the politician a measure of the widespread popularity of smoking. There was no popular mandate to move against it, only political risk. For (ii), nicotinewas known to be an habituating drug but in the (small) doses involved in smoking it was considered to be harmless. Moreover, like alcohol, it had a long history of popular acceptance. Addictive or habituating properties, however, were not per sea ground for action; that nicotine acts as a proxy for harmful tobacco products and could be considered on this basis was too subtle a concept for an unreceptive legislature. For (iii), the facts did not warrant any infringement, of the industry's existing commercial freedom. For (iv), there was no refi.able evidence relating strength of.toba~co products to their, to.x[.eity." l~oir tobacco ~rnoke, while offensive to ~ miriority, harmed only the smoker, and self-poisoning was no longer a crime. Furthermore, the putative tobacco-related diseases were non-contagious. As then seen by government the smoker polluted nothing and infected nobody: to constrain his smoking may or may not have protected his inalienable right to life but would certainly have infringed his equally inalienable rights to liberty and the .pursuit of happiness! Even for (vi) and (vii) the message • was qu~i.fied: for (.vii) the evidence w.as tenuous u.ntil

2 the results of later prospective studies; while the role of Smoking in (vi) had many distinguished sceptics -- including R A Fisher.m4 Circumspection, but not inaction, was adopted as government's watchword• In 1951 they responded tO the initial 'incriminating research findings by: referr.i.ng:.the m~tter to their 'Standing Advisory Comniittee on Cancer and R/tdio~he.rapy, then to a panel under the Go~/drniiaent Actu~ry~ meanwhile stonewalling in pai'liament. Then, as the facts became incontrovertible and on the advice of the Chief Medical Officer (CMO), the Minister of Health (lain Macleod) on 12' Februapj 1954 stated "I accept the [Standing] Corfimittee's view that the'statistical evidence points to .'smoki.n'g.as a:f~etor in.lung cancer. ?5 He now moved positively if cautiously and c(iunselled further research, annonnced a research grant of "£25Q 000 from the tobacco companies to the Medical Research Council, ~5 and waited on events. It is true that tobacco duty (then some 30% of all government taxes on expenditure and 12% of total government revenue~6) gave government a vested interest and that introducing even a degree of substitute taxation would have been difficult and unpopular in the fiscal circumstances of the time, but even in retrospect the in!tial government response seems judicious rather than supine. Supine or judicious, certain principles were clear and were acted on. The young had to be protected; the Children and Young Perso.ns Act, 1933, forbad tobacco sales to children under 16 years and though widely flouted by inanimate vending machines and animate retailers alike (the average annual prosecutions under the Act, 1945-52, was only 20!)]7 it was considered adequate in the circumstances, m9 The public had to be told the medical facts, and here government were to their supporters scrupulous, to their detractors neglectful. They were certainly wary. Despite heavy pressure in and out of parliament they refused to sponsor or encourage a national education campaign and conceded only that they 'would take such steps as are necessary to ensure the public are kept informed:2° Until 1957, three years after the Minister's acceptance of the role of smoking in lung cancer]5 this was mainly through parliamentary answer and muted references in Ministry of Education handbooks for teachers,2~ but following an MRC report of June 195.7.2z goverhtnen.t involved the local;heal'tl5 authorities and Central (~oun~i'l "for "Health Education23 and by mid-1959 were claiming substantial success?4 Again they claimed to be acting judiciously, a belief bolstered by a philosophy of accepting the rights of adults freely to purchase and smoke tobacco and of industry to make, advertise, and sell their products in the marke~-place. Their third responsibility -- to encourage research -- they considered to be adequately INTERNATIONAL JOURNAL OF EPIDEMIOLOGY discharged through existing general research resources and the industry's £250 000 grant to MRC. Ih 1962 the Royal College of Physicians of London (RCP) published its inflffential monograph Smoking and Health2~ with its clear message for action. Pressure on • governmdrit now mounted2 At. first this pushed them " faster andfurther alo.ng'tile health educatiofftrack':'Iocal health auihority education spending was stepped up, • posters were distributed by the Ministry of Health including, from 1964, the message 'cigarettes cause lung cancer' to a total of £92 000 in 1966-67, anti-smoking colour films were made, millions of inscribed bookmarks ' were issued, and a postn.aa~k cancellation. .'~cigarettes. harm yotir health' g?as bri6fly us6d.2~'29"It al.so'ptished "' them to support anti-smoking .clinics and issue smoking cessation leaflets,a° More daringly, they constrained advertising for the first time in 19653~ (industry had tried to pre-empt this with a modest self-denying voluntary ordinance on over-persuasive advertising to the young in 1962), and, harried in parliament,a2-aa the Minister of Health (Kenneth Robinson) in 1967 went further and tried to persuade the tobacco manufacturers to voluntarily restrict their aggressive press and poster advertising and sales promotion campaigns, though unsuccessfully?4 The mood of the country however had changed since the 1950s and government now answered this rejection of their advances by the industry by deciding 'in due course to take powers to ban coupon gift schemes and other promotional schemes, to forbid or limit certain forms of cigarette smoking, and to limit expenditure on advertising of cigarettes',a4 This threat ultimately succeeded and did much to convince government to rely mainly on persuasion (and a middle- sized stick!). They still however saw themselves as marginal players and the 1960s ended with their repeated refusals to imitate the USA initiative of printing incriminating labels on cigarette packets or to consider fiscal measures. With the harsh judgement of hindsight government, in their policies on smoking, were poor custodians of the public health in the 1950s and 1960s. Smoking- related diseases rose inexorably. Much of this increase Was due to earlier• exposure for which tiost-1950 governments could not be blamed, but the nuniber of .' the you~.. ~.e.nteri.ng.the ~mokiflg ranks'could be'affected • " by g6v.ernlnent policies and here the si .t-oation'had grossly • deteriorated: the percentages of men who smoked in 1951 ' and 1971 were respectively 62 % and 55 % and those for women were 33% and 49%.35* This latter year (1971) * This was for 1956 and 1971; reliable prior data are not available. For manufactured cigarettes, the almost exclusive tobacco product smoked by women, the 1951 and 1971 figures were 28% ~/n~l 48%.36 tht ad agt (G M~ ' rel act SUl sta~ cla • se~, rec on] for, soc in TH Th~ dist ind cor star resl tha~ tw~ put~ hea the~ tho.' ofs the~ mal eact rob~ inte dea the tOS, tern ..ve~ • t.ox'i .¢ • • go~ y.eai cant thro (the Tob. rese 196~ hou,,

DETERMINANTS OF POLICY ON SMOKING AND HEALTH the tobacco manufacturers spent some £63m on advertising and sales promotion, government and its agents less than. £0.5m.3~' Small wonder the CMO (George Godber)"called this .situation" 'incredible:38 Moreover, during:the~ two decades t0.b'a~co pro'duct~ r~lative to wages ~ha'd becbme much cheaper, ~6 the rate of duty had hardly changed, and tobacco taxation as a • proportion of total expenditure on tobacco products actually declined.39. Not surprisingly tobacco con- sumption increase.d during the 1950s and when it did. start to fall in the mid-1960s it was mainly in higher social class' mefi and in response to" m~dical opiniom Ot~ ~he seven courses of 'possible action by gover.nment' recommended in S~noking and Health in 1962, at most only three had been acted on by 1970.40 A far more forceful policy from government was needed and was soon to be demanded with medical professional bodies in the vhn. THE VIEW FROM INDUSTRY The demonstration of the association of smoking with disease had immense implications for the whole tobacco industry. Its common policies, and those of its principal constituents, are only known from actions and statements; much crucial material lies in classified and restricted archives. It is, however, broadly true to say that the tobacco manufacturing industry has followed a two-prong policy: (i) to make cigarettes 'safer' (or putatively 'safer') to smoke -- in this way the increasingly health-sophisticated public might continue to smoke them; and (ii) to market alternative tobacco products for tho~e who wished to remove themselves from the hazards of smoking cigarettes without at the same time removing themselves from the pleasure of smoking. Naturally the manufacturing companies competed vigorously with each other for market share, and naturally they also robustly defended their perception of their shareholders' interrsts, and their fiduciary responsibilities, in their dealings with government, the medical profession, and the public. They had, however, and still have, no desire to sell a harmful product either in the shorter or longer term when they can sell a :safer' one, and th'ey invested. yery heavilS, in're,search, arid d'ev.eldp.tnent into tobace0 toxicity as well as produdt accel~tability. Iri 1954, under government pressure, they gave £250 000 (over seven years) to the MRC 'for research int.o the cau.ses of lung cancer',15,4~ in 1956 they coordinated their research through a Tobacco Manufacturers' Standing Committee (the Tobacco Research Council since 1963, now the Tobacco Advisory Council), in 1962 they opened research laboratories at.•Harrogate with 250 staff, by 1969 they had committed some £6 million, in out-of- house research over .and above their in-house R and 3 D,42 and today they are by a very long way the largest sponsors of research into smoking and health in the UK. Though much of their effort is directed to their credo that most smokers will contin.ue .to sriapk.e.so.me, fo .rm. of tobae'.co, unlegs deterred.by its eor/tinued, toxicity, .a credo jtistified by events certainly un(il the 1970s, they have iubstantially increased our knowledge of nearly every aspect of tobacco toxicity. The industry, therefore, found themselves in part-chorus with government, the smoking public, and the medical profession: all wanted .. safer s.moking .but for differen.t reasons! ,.They we.re. • " l~vever in•discord o+er"a'nti-smoking objectives v~hich were seen at the time in very simple, and in retrospect simplistic, terms. How well in the 1950s and 1960s did the policies of the tobacco manufacturers succeed? Undoubtedly very well. Aggressive marketing converted the smoking population to filter tips: 2.3% of all tobacco products by weight in 1956 they were 64.5 % in 1970 by which time four of every five cigarettes smoked was faltered.43 Each cigarette was 'safer' than its unfiltered counterpart less through absorption of noxious substances than because the filter tip displaced a significant weight of tobacco. To compensate for this tobacco loss more cigarettes were bought; between 1956 and 1970 the total weight of cigarette tobacco consumed declined by 9.6 % but the number of cigarettes sold increased by 28.5% .44 Furthermore, the industry's second prong -- developing alternative products -- was also successful. In 1961 the number of 'cigars and cigarillos' sold was 315 million; in 1967, only six years later, it was 1135 million,4~ meeting the demands of many ex-cigarette smokers not least doctors.46 While all the time the industry was garnering information through its substantial R and D programmes and supplying, and to an extent creating, the smoking market's needs. Its success was as marked as was the failure of the government's anti-smoking policies. THE NEW INITIATIVES In the early 1970s renewed pressure in parliament47 and from the. pro.f.ession48 forced.goye .r~n. ent's hand. Their •-first:action,, in Mai:ch.:19"~0, .whs riece'ssa.ry though moddst: they appointed a Senior Medical Officer to coordinate inter-departmental work on smoking and health. More significant action was not long delayed: in June 1971 the Minister (Keith Joseph) announced details of an agreement with the industry on the labelling of cigarette packets and adverts with appropriate health warnings.~9 This was the first of the 'voluntary agreements' -- portrayed as gentlemen's agreeme.nts they were (and are) toughl3~ negotiated compromises -- presage.d four years previ.ously by,.which to the present

4 day government has sought to effect its smoking and health policies in preference to enforced action. Government now acted with more vigour. They had, since 1954, warned of the risk to cigarette smokers of lung cancer; now in 1973 they told the public which cigarette bran.ds-con.tained the most tar and. nicotine by • igublishing a brand l~ague table --'which "is noW. a biannual ser~e.s with (from.1984)' carbon monrxide' yields added. They "also, belatedly, recognized the need for systematic and above all impartial scientific advice, and in 1973 .established the Independent Scientific Committee o.n Smoking and Health (ISCSH): 'independent' in that its members are.not fro.in government, the civii service or .industry, ake unpaid, an'd it reports direb~l~ to the health ministers; and 'scientific' in that the members are prominent scientists from cogfiate disciplines and the Committee's advice is based wholly, necessarily, and exclusively on the scientific evidence without regard to any other consideration. Its general terms of reference are to advise the health ministers 'and where appropriate, the tobacco companies' on the scientific aspects of matters concerning smoking and health. It has no role in making recommendations on other prongs of government policy, ie in advertising products or in fiscal policy, though it has a role in commenting where relevant on health appraisal and health education. Apart from ad hoc advice it has published four scientific reports which have had considerable influence on government .policy,5°53 with up to £7 million supplied by TAC under the 1980 and 1984 voluntary agreements it sponsors, through the Tobacco Products Research Trust, 25 research projects, and has sponsored an international symposium on the role of nicotine in the product modification programme54 and is organizing another on the role of smoking in hormone-related diseases. This Committee at once tackled the basic irony of smoking, namely that most scientists agree (with the exception of one important group55) that it is tar that contains the lung cancer-producing or inducing agents but the smoker smokes mainly to obtain nicotine, which confers undoubted benefit to smokers and may improve performance of non-smokers.54 Absorbing (harmful) tar constituents is in fact an unwanted by-product of absorbing useful (harmless) nicotind. This is an over- simp.lific~tion but. close.to .the .truth. If tob'acco could' .th6refore be stripped, of its tar, or the absorbable ta~ greatly diluted, or if all the harmful tar prrducts could be removed in some way or rendered innocuous or inoperative, smoking cigarettes could keep its appeal and lose its main danger, a seductive goal for government, smokers, and the tobacco industry alike. Looking beyond this, since smokers seek mainly nicotine why bother with tobacco at all; why not have an inert combustible INTERNATIONAL JOURNAL OF EPIDEMIOLOGY substance which contains adequate nicotine which can then be inhaled when burnt, or nicotine tablets, or wadding impregnated with nicotine which will distil over at room temperature or, with .a source of. heat, at lower temperatures than pyrolysis, or some other, product of ingenious ~o.bacco'techn.ology; or w. hy not suck or chew tobacco ~q e~hang~rig a low ri.sk 6f bhcc.al, can66r. a higli bne:6f lung cancer? AH of these have been tried at one time or other, some are cut:renf today and other~ will be with us tomorrow. Their basic objective is the same, viz to supply nicotine to the brain in a rapid, efficient, and effective manner, where it is pharma- colog!cally.act!ve, withou.t, absorbing too much harmful tar;" px%ferably Without absorbing- any. V~'hbn "confined to smoked tobacco this policy is called 'product modification', and the programmes designed by the ISCSH and industry and negotiated by government with the Tobacco. Advisory Council (TAC), in a series of 'voluntary agreements' (in 1973, 1977, 1980, 1984) is called the 'product modification programme'. Product Modification The first method tried was an ambitious one, viz tobacco substitutes, ie replacing much of the tobacco in a cigarette with essentially a tar-free substance, taste and other attractions of pure tobacco being preserved by additives. The first work of the ISCSH was in fact in establishing with industry acceptable means of testing tobacco products and substitutes, and the possible toxicity of additives. In July 1977 two large tobacco manufacturers marketed cigarettes containing res- pectively the 'substitutes' Cytrel and New Smoking Material (NSM) -- these cigarettes were a commercial failure and were eventually withdrawn losing for the companies concerned many millions of pounds, for a decade something of their zeal for such radical change, and possibly something of their admiration for the fledgling ISCSH! This caused some strain but government, wisely in my view, resisted pressure for more rigorous alternative action, and in many ways endorsed its own faith in voluntary agreements. Government in fact had come a long way since 1970: it now had a clear and agreed scientific policy (product modification) 6a~ed on increasingly sound a priori • ' ~ro~n~s; i.t ha.d the.haehns'6f implementing it(volufltary agreements); ahd it had reasonably cohrrent generkl strategies. Most importantly it has supportec~ the ISCSH: it has widened its terms of reference, has only once failed to accept the Committee's advice and that unim- portantly,56 and in March last year accepted the recommendations in the Committee's FourthReport. If government ,was laggardly up to 1970 it has since done more, and often much more, than most advanced

DETERMINANTS OF POLICY ON SMOKING AND HEALTH countries and in the p.ast decade has seen its policies produce a 25 % reduction in total smoking, an average tar reduction per cigarette of 30 %, a substantial decrease in lung cancer in younger age groups,5a and the transition of smoking from" social norm to. a miriority" ind.u|gence. O.niy in the lower socioecot~omic..groups and in. young'wo.men has progress beeia disappoit~ting. It ig mere specula.tion as to whether'the more robust measures that some recommend would have produced more beia~ficial results or been count~rproducti~,e. Having failed with substitutes the Committee, govern- ment and the industry espoused systematic gradualism. • Un.der rep.ea.ted volun.tary ag.reem.ents sal.es-weigh.t.ed.tar, nicotine and toxic gases v~ere reduced gradually, tai" most of all (Table 1). Frill details of this product modiflcatiqn programme are in the Committee's Fourth Report.53 Briefly, there are two pre-conditions for its success. First, that the Committee's scientific assumptions are correct: these are, th.at the noxious materials in tobacco are in the tobacco tar and/or are products of its combustion; and that there is a dose response with the diseases they cause (most particularly lung cancer) without a significant threshold of safety. Second, that the programme develops in the context of an increasingly sympathetic environment for less smoking and for lower- tar products and is not so ambitious as to prompt resistence from the consumer. This 'consumer operates through so-called 'compensatory smoking; ie mechanisms by which the smoker maintains the nicotine dosage which his body is conditioned to need for optimum efficiency, effectiveness, and contentment. It is a crucial phenomenon and is briefly as follows. There are three main mechanisms: first, the smoker can simply smoke more cigarettes; second, he can switch to a stronger brand; third, he can 'oversmoke' a ~igarette -- longer and more frequent pulls, shorter butts, etc. "T~,BLE 1 Annual sales-weighted tar, nicotine, and carbon monoxide yields (mg/cigarette) of manufactured cigarettes in the UK Year Tar Nicotine CO 1934-40 32.9 2.00 18.6 1955 -61. 30.4 2.03 20.6 1965. 31.5 2.08 • 18.8 .1970. 22.5 . :1.56. 17A 1975 • 17.9 1.34" 1~5.2" 1980 16.3 1.30 "16.6 1982 15.4 1.32 15.2 1984 14.6 1.28 14.1 1985 14.4 1.31 14.7 Now* 13.55 1.21 14.42 Source: Wald Net al (reference no 11). Sect 8; ISCSH. Fourth Report (reference no 53)• Sect I. * Based on date from the 27th Survey of the Laboratory of the Government Chemist, January-June 1988. 5 Each of these increases his tar exposure; combating them is a cornerstone of the Committee's policies. The ISCSH can have little to say on obvious measures to limit the number of cigarettes confirmed smokers smoke other than in health awareness fields but it can adopt policies with respect to the other, two mectianisms of;.. .. " :.compensatory ~moking'. One. such means is to'ensure • that stronger brands are not'on the market. For some . years this has held for new brands, but last year ISCSH recommended an upper tar limit of 16mg per cigarette for all brands 'as soon as possible' and reducing to 14mg after four years.57 Another is to gradually reduce tar levels ia pop.ular brands .t.hough without risking .sig- • niflcant switching to higher tar brands. This is being done by ingenious tobacco technol0gy.5s Another is to • develop the market for low tar cigarettes by selective advertising and brand promotion supplementing the health education message. The industry, despite its critics, has been active: by February 1986, 33 of 138 brands included in the Laboratory of the Government Chemist biannual survey were low tar (0-9.99mg per cigarette) as against 19 of 114 ten years previously, though since about 1980 their market share has plateaued at about 13%.59 (Some are unhappy with 'selective' advertising on the grounds that it is still advertising of tobacco: this issue is too complex to discuss here)• Yet another is to exploit the desire for nicotine of irrevocable smokers as a means of reducing their tar exposure: since nicotine plays an important role in influencing 'compensatory smoking' it follows that by maintaining the nicotine yields of cigarettes while paripassu lowering the tar yield (by sophisticated tobacco and cigarette technology), tar intake can be reduced more than by lowering nicotine and tar in equal proportions. This in fact has been pursued over the last few years (Table 1). There are technological limits to this approach and it sets moral, philosophical, and political problems concerning using an habituating drug in this way especially in the young and new smoker, as recently discussed;6° nevertheless the Committee recommends someqimited trials.6| This product modification policy based on gradualism has proved robust and effective. The low tar programme • has contributed sig.nificantly to the reduction in lung cancer mortalR3/in the'younger age grohps-.(who have - not been" exposed to high tar products in their shorter smoking history) and possibly also to the redtiction in chronic obstructive airway disease. Its role in ischaemic heart disease, however, is equivocal.62 Further infor- mation will emerge from on-going studies worldwide including many of the 25 projects sponsored by the Tobacco Products Research Trust on behalf of ISCSH.63

6 CURRENT PERCEPTIONS Anti-'smoking groups argue that the above approach merely plays at the margins. They seek more direct government action including draconian increases in tobacco duty arguing that until about 1980. tobacco product.s were, bec.om!fig ehe.aper in terms.of di.Sposable income,:..that .tobacco taxation, as a. 'share of total expenditure.on tobacco t~r6dficts was actually falling,~6 and that swingcing increases in duty had helped control. the gin-swilling epidemic in the mid-eighteenth century. They also e.mphasize that the perceptions on which govern.merit originally devised its strategies have drastica!ly changed especially since the early 197.0.s. These'two contentions are Row examified tojgeth~rwith recent proposed EC Directives. Change in Strategic Perceptions Of the seven 'critical points' listed at the start of this article and which influenced government policy in the 1950s, all but two had changed by the 1980s; some through government action. The two which remain as irrefrangibIe fact are: the non-contagious nature of tobacco-related diseases; and the acceptance of a low or zero threshold effect of cigarette smoking and lung cancer. Such a dramatic change in strategic perceptions would alone make a reassessment of policy timely; recently government has merely pressed on along established paths and tightened existing screws. There is angther, more pressing, reason, viz environmental tobacco smoke (ETS, also called 'passive' or 'involun- tary' smoking) has since 198664.66 been generally accepted as a health hazard: a (small) increased risk of lung cancer in consistently exposed non-smokers, and respiratory symptoms, episodes of respiratory illness, and decrements in lung function in similarly exposed young childrem67 The argument that smokers poison only themselves (or their unborn children?) can no longer be convincingly sustained. The conceptual framework within which government, industry, and the profession have worked, is fundamentally changed. Naturally the industry opposes the belief that ETS is harmful to health and seeks vigorously to dismiss the supportive scientific findings as methodological artifacts,68 and some workers agree.69 It is easy to see whythe industry's opposition is total. It is filso easy'to's6e.why many call for more dynarriic action froin government. R is, however, less easy to foresee how and to. what extent government will respond. In March 1987 they accepted the ISCSH Interim Statement incriminating ETS,7° and in March 1988 also the Committee's Fourth Report (including a detailed re-statement of its earlier position) now with specific recommendations,~ which will probably form a basis for formulation of government INTERNATIONAL JOURNAL OF EPIDEMIOLOGY policies. Only one of these, viz possible segregation of smokers from non-smokers in work and indoor leisure environments may lead to more direct action. The question of ETS is undoubtedly the most difficult one for "the tobacco industry and its" formal and informal ..regulatg.rs since the causal associa.tion of "cigarette.i: smoking gndlung" cancer vOas demonstrated. Tobacco Duty The effect of changes in overall cigarette price (largely dictated by tax) on the amount and distribution of smoking is, like that of advertising, complex.16,71-75 Soine facts are not in dispute. Tobacco.tax. though far. frofn ~ "negligibl~ sburce of revenue --~ eq£iivalent in 1986-7 to about 4p on the basic rate of income tax or raising VAT from 15% to 20%16 -- now provides only some 4% of government revenue as against over 16% 40 years ago: VAT and petroleum revenue tax are greater. Cigarettes are far cheaper now in terms of wage rates and at least 25 % below their 1948 price level in real terms.76 Low income groups spend a larger proportion of their income (on average) in taxes on tobacco but they reduce their smoking more in response to tax increases; some even hold that the downward drift in real cigarette prices has been a major factor in widening the gap between upper and lower class smoking patterns and largely negating the benefits of health education programmes.77 Furthermore, if, against the evidence, tax increases led to diminishing returns, losses could now easily be recouped from other sources. Econometricians have recently tried to quantify the relationships with varying results16,76,77 though most agree that the higher the price of tobacco the less will be smoked and that there would be no significant diminishing tax returns in the shorter run. One group suggested that a 10 ~o real increase in taxation would cut tobacco consumption by 5-6% and increase tobacco revenue by up to 7%.16 This would be a highly desirable public health outcome: better health and more tax revenue[ Why then has government been reluctant to sanction significant increases in tobacco excise duty (there was a 16% increase between 1980 and 1981 but since then the average increase in real duty per cigarette has only marginall!¢ .~xceeded inflation) or institute a full-blooded i~oli~" of dlffer~fitial duty dependent on tar';t~:~ngth" -- that on cigarettes'yielding more than.. 20rag tar, imposed in September 1978, was withdrawn in 1981 -- especially since the net effect on the overall economy by further reduction in tobacco manufacturing and distributing activity would be minor.16 The main reasons seem to be: (i) a political caution at 'over-taxing' what is still a widely practiced indulgence (alcohol is a similar case), (ii) a belief that if price response is low then tobacco 0 O~ Co 0 tax is effect and chalk remo~ gover a pac' The reduc EC 13 On 4 Coian to aplc of to~ cigare of pa( 'Toba langu langu viz's disea~ on ea, seriot howe' of tar 1992, of sal 1988 by IS, If ace retro~ so wk in the range profil incre~ in Br In : one c on pr of ag, curre use o tines levels for ff curre CON Gove healtI neafli

DETERMINANTS OF POLICY ON SMOKING AND HEALTH tax is regressive and socially inequitable,78 and (iii) its effect on RPI. The first lies in social attitudes which can and should be changed. Recently (ii) has been challenged,!6,76,77 and it .may be that~ modem work will remove the "regressive' .ai'gument. While on (.iii).a recen, t government estimate is that an increase of 30p duty on a packet of 20 cigarettes would increase RPI by 0.7%.79 The'economic argument against fiscal measures to reduce smoking is not impre.ssive. EC Directives • - Oh 4 FebrUary 1988 the-.Cornmission, of the.European Communities submitted to the Council draft Directives to approximate the laws dfmember states on (i) labelling of tobacco products and (ii) the maximum tar yield of cigarettes. The former, if accepted, will require all units of packaging of tobacco products to carry the message 'Tobacco s~r.iously damages health' in the official languages of the country and, in the country's own language, as a minimum one of two specific warnings, viz 'smoking causes cancer' and 'smoking causes heart disease'. Also, tar and nicotine yields are to be indicated on each packet of cigarettes which are unlikely to cause serious problems in the UK. The latter draft Directive, Ohowever, if accepted, will. It specifies an upper limit of tar yield for all cigarette brands (15mg by 31. December 1992 and 12mg by 31 December 1995) without mention of sales-weighted average tar levels which in the UK in 1988 was about 13rag per cigarette and is recommended by ISCSH to be reduced to about 12rag by end-1991.8° If accepted in its present form this Directive couM be retrogressive in the UK by allowing manufacturers if they so wish to increase the tar levels of many of their brands in the Low to Middle Tar (10-14.99mg per cigarette) range and, without countervailing change in market profiles, the sales-weighted average tar levels would increase. Representation on these lines has been made in Brussels. In addition two further Directives are being drafted, one on curtailing smoking in public places, the other on preventing sales of tobacco to those under 16 years of age. "In principle neither would cause problems to current government thinking-although in the former the use of mandatory directives rather than discretionary lines of sectoral action might. Only on the matter of far levels may harmonization 'post-1992' lead to problems for the UK anti-smoking policies as the Directive is currently drafted. COMMENT Government has a responsibility to improve the public health. Smoking tobacco, especially cigarettes, has for nearly 40 years been a demonstrable health hazard to 7 the consumer, and recently ETS has been increasingly accepted as a health hazard to exposed non-smokers through 'involuntary". smoking. This third party risk introduces a new dimension to the traditional problem of smoki.ng and health, which the tobaecO i.n..dustry has been quick to i~erceive in its publicity .campaigns and in the thrust of much of its sponsored resehrch towards questioning the methodology of the supportive studies. Hopefully government will be as alert in discharging its own responsibilities. There are many means by which tobacco consumption can be reduced and t.he.se depend upon action by smokers, non-smokers, educators, health, care pro- fessionals, and most importantly government -- who should consider all means open to it to effect a reduction in smoking. Part of government's past failure and seeming dilatoriness has been the diffusion over several bodies of responsibility for effective action: thus taxation, art and sport sponsorship, health education among the young, sales promotion, and environmental restrictions, the armoury of government's anti-smoking campaign, are the responsibility of various departments other than the Department of Health. These structural problems have undoubtedly compounded any lack of political will and deprived government policies of much necessary focus, cohesion, and impetus. One area which is however very much a Department of Health responsibility is the toxicity of tobacco smoke in marketed products, and the product modification programme. The importance of this issue is often paid only lip-service by anti-smoking lobbies -- and (for different reasons) also some pro-smoking lobbies -- since they hold that espousing it weakens all-out resolve to terminate smoking. They argue that commitment to cessation and not to encouragement of 'less hazardous' smoking with its tolerance to (low tar) advertising and its countenancing of continued smoking, is needed. I have explained in this article why this criticism is based on a misinterpretation of the ISCSH policies. The Committee perhaps keeps too low a profile since often its work seems not just misunderstood but hardly even to be known -- the latest (1983) report o.f the RCP in its discussion" on product modification i~ an example8~ -- yet raising it might coml~rorr~ise the excellent working relations it has developed with all pai-ties which has ensured the quality and acceptance o.f its advice. More valid is the criticism that whateyer is the strength of the a priori case for a low-tar programme, product modification, the key ISCSH policy, has not been shown to have contributed to a reduction in smoking-related disease. After the Committee's Fourth Report this no longer holds for cancer of the lung and chronic obstructive pulmonary disease: the situation concerning

8 ischaemic heart disease is more equivocal and the undoubted role ot~ smoking in its aetiology is still an enigma.60 While the search for the cancer producing agents in tobacco continues the reduction of tar and other 'noxa" must continue also. Such empiricism is not new in public health. But should, it be effected through voluntary agreement~ or le~islation?.As thing~ staiadI can see rio advatitage, and some disadvantages, in :the latter. Government, as already noted, has been able to incorporate in its four voluntary agreements with the industry all but one 0fthe ISCSH recommendations and has accepte~t the recent Fourth Report completely. If the situation changed, however; I might think differently! In any event the.'post~1992-' situation whfch lib6 current and pr~isosed EC Directives foreshadow, will lead to a new situation. Less conjectural is the Committee's need to continue "to supply government and the industry with the best scientific advice as its brief demands. In the important and intensely complex field of smoking and health the Committee needs to be enabled to sponsor research and not have to rely on an electic choice from the often unsystematic and unhelpful (in the context) world literature. The UK is an .acknowledged leader in °less hazardous' smoking policies; crucial population research is not likely to be centred elsewhere. The monies supplied by TAC to ISCSH since 1981 for research to monitor the effects on human health of product modification and which supports 25 projects are now fully 'committed. It is vital to the whole future of national public .health policies in smoking and health that further funds from some source be made available to the Committee under terms which it can accept without compromising its status or standards. We must await the terms of any new voluntary agreement which may replace the one which expired at the end of 1987. REFERENCES I International Agency for Research on Cancer• Tobacco smoking. Lyon: I.ARC, 1986 (Carcinogenic risk of chemicals to humans; vol 38). Royal College of Physicians of London. Smoking or health. London: Pitman Medical 1977 pp 15. Schrek R, Baker A, Ballard G P, Dolgoff S. Tobacco smoking as an etiologic factor in disease. I: Cancer• Cancer Re.s 1950; 10: 49-58. - • . Miffs C A, Porter M M. Tobacco smoking habits and cancer of the mouth and respiratory system. CancerRes 1950; 10: 539-42. 5Levin M L, Goldstein H, Gerhardt PI R. Cancer and tobacco smoking. A preliminary report. JAMA t950; 143t 336-8. Wynder E L, Graham E A. Tobacco smoking as a possible etiologic factor in bronchiogenic carcinoma. A study of six hundred and eighty-four proved cases. JAMA 1950; 143: 329-36. Doll R, Hill A B. Smoking and carcinoma of the lung: preliminary report. Br Med J 1950; 2: 739-48• INTERNATIONAL JOURNAL OF EPIDEMIOLOGY s McConnell R B, Gordon K C T, Jones T. Occupational and personal factors in the aetiology of carcinoma of the lung. Lancet 1952; 2: 651-6. 9 Doll R, Hill A B. A study of the aetiology of carcinoma of the lung. Br Med J 1952; 2: 1271-86. Io Parliamentary Debates (Hansard). House of Commons, session " 1950-1, fifth.series, vol 489, co1"1549-51. (Hereinafter cited as Hansard, HC). 1! Wald. N; Kirylult S, Darby S, Doll. R, Pike M, Peto R, 'eds. UK " smoking #tatisn'c~. Oxford: Oxford. University Press, 1988." Table 3.3. i~ Ibid. Tables 8.1, ~a Fisher R A. Lung cancer and cigarettes. Nature 1958; 182: 108. ~4 Fisher R A. Cancer and smoking. Nature 1958; 182: 596• t5 Hansard HC, 1953-4, vol 523, col 173-4W. ~6 Godfrey C, Maynard A. Economic aspects of tobacco use and taxation policy.. Br Med J .198.8; 297: 33.9.-4.3. ' 17 Hansiz/d HC, 1952-3, "vO1 514, col 1394. ~8 Hansard HC, 1956-7, vol 569, col 84W. 19Hansard HC, 1956-7, vol 572, col 1284-5. ~o Hansard HC, 1955-6, vol 552, col 803. 2~ Hansard HC, 1956-7, vol 560, col 1356-7. 22 Medical Research Council. Tobacco smoking and cancer of the lung. Br Med J 1957; 1: 1523-4. ~s Hansard HC, 1957-8, vol 577, col 81W. ~aHansard HC, 1958-9, vol 605, col 712-22. 25 Royal College of Physicians of London• Smoking and health. London: Pitman Medical, 1962. 26 Hansard HC 1962-3, vol 673, col 1171-2. ~7 Hansard HC 1962-3, vol 675, col 13-14. zs Hansard H( 1963-4, vol 688, col 16W. ~9 Hansard HC 1967-8, vol 753, col 604-8. ~o Hansard HC 1968-9, vol 777, col 410W. ~tHansard HC 1964-5, vol 706, col 11-14. ~2 Hansard HC 1964-5, vol 710, col 160W. 33 Hansard HC 1965-6, vol 725, col l17W. ~ Hansard HC 1967-8, vol 751, col 1327-9. as Wald Net aL UK smoking statistics. Oxford: Oxford University Press, 1988: Tables 4.8.1 and 4.8.2. a6 Ibid. Table 4.1.2. 37 Royal College of Physicians of London. Smoking or health. London: Pitman Medical, 1977: Tables 1.1(a), (b). as Department of Health and Social Security. On the state of the public health. London: HMSO, 1969: 9. 39 Wald Net al. UK smoking statistics. Oxford: Oxford University Press, 1988: Chapt 11. 4o Royal College of Physicians of London. Smoking and health• London: Pitman Medical, 1962: $8. 41 Thomson A L. Halfa century of medical research. Volume one: Origins mad policy of the Medical Research Council (UK). London: HMSO, 1973: 212-3. 42 Tobacco Research Council. Review of past and current activities. London: Tobacco Research Council, 1963. 43 Wald et al. UK smoking statistics. Oxford: Oxford University Press; 1988: Tables 1.1 and 1.2. . 4~ Ibid. Tables 1..2 and 1.3. as Ibid. Table i.2.. 46 Royal College of Physicians of London. Smoking and health now. London: Pitman Medical and Scientific Publishing, 1971: Fig 1.4. 47 Hansard HC, 1969-70, vol 794, col 519-22. '~Royal College of Physicians of London. Smoking and health now. London: Pitman Medical and Scientific Publishing, 1971. 49 Hansard HC, 1970-1, vol 819, col 340-4W. 1

DETERMINANTS OF POLICY ON SMOKING AND HEALTH 50 Independent Scieptific Committee on Smoking and Health (ISCSH). First report: Tobacco substitutes and additives in tobacco products. London: HMSO, 1975. 5~ ISCSH. Second report: Developments in tobacco products and the possibility of "lower-risk' cigarettes. London:. HMSQ, 1979. 5z ISCSH. Third report. London: HMSO, 1983. 53 ISCSH. Fourth report: Smoking and health. Lon(~6n: HMSO, 1988. 54 Wald N, Froggatt P, eds. Nicotine, smo[dng and the low tar programme. Oxford: Oxford University Press, 1989. 55 Hoffman D. Nicotine, a tobacco-specific precursor for carcinogens. In: Wald N, Fmggatt P, eds. Nicotin¢, smoking and the low tarprogramme. London: Oxford University Press, 1989: 24-40. 56 ISCSH. Fqurth report: Smoking and health. London: HMSO, 1988, para 18. 57 Ibid, pare 31. s~ Wald Net ~l. UK smoking statistics. Oxford: Oxford University Press, 1988: Table 8.8. 59 Ibid. Table 8.14. ~o Froggatt P, Wald ~. The role of nicotine in the tar reduction programme. In: Wald N, Froggatt P (eds). (See reference no 54). Chapt 1Z 6~ ISCSH. Fourth report: Smoking and health. London: HMSO, 1988. .Para 34. 62 Ibid. Sect 2. 6s Ibid. App 4. ~ US Department of Health and Human Services. The health consequences of involuntary smoking: A report of the Surgeon General~ Rockville MD: DHHS Office on Smoking and Health, 1986. 65National Research Council. Environmental tobacco smoke. Measuring exposures and assessing health effects. Washington DC: National Academy Press, 1986. 66 ISCSH. Fourth report: Smoking and health. London: HMSO, 1988. Sect 3. 67 Ibid. Pare 70-1: 68 Tobacco Advisory Council. The clouded issue. London: Tobacco Advisory Council, 1986. 69 Lee P N. Passive smoking and lung cancer association: a result of bias? Hum Toxicol 1987; 16: 517-24. 70 ISCSH. Interim statement on passive smoking. Hansard HC, 1986-7, vol 112 (ser!es 6), col 328-9W (13 March 1987). 7~ Sumner M T. Demand for tobacco in the UK. The Manchester' School 1971; 39: 23-36. 72 Atkinson A B, Skegg J L. Anti-smoking publicity and the demand for tobacco in the UK. The Manchester School 1973; 41: 265-82. 73 Russell M A. Changes in cigarette price and consumption by men in Britain 1946-1971: a preliminary analysis. Br J Prey Soc Med 1973; 27: 1-7. 74 Peto J. Price and consumption of cigarettes: a case for intervention. Br J Prey Soc Med 1974; 28: 241-5. 75 McGuinness T, Cowling K. Advertising and the aggregate demand for cigarettes. European Economic Review 1975; 6: 311-28. 76 Townsend J L. Cigarette tax, economic welfare and social class patterns of smoking. Applied Economics 1987; 1~: 355-65. 77 Townsend J L. Economic and health consequences of reduced smoking. In: Williams A, ed. Health and economics. London: Macmillan Press, 1987: 139-61. 78 Atkinson A B, Townsend J L. Economic aspects of reduced smoking. Lancet 1973; 2: 492-6. 79 Hansard HC, 1987-8, vol 123 (series 6), col 396-7W. 8o ISCSH. Fourth report: Smoking and health. London: HMSO, 1988, para 12. 8~ Royal College of Physicians. Health or smoking? London: Pitman Publishing, 1983: 127-8.

60~g~9S90~

The relevance of tobacco-specific nitrosamines to • human cancer .... STEPHEN S HECHT and DIETRICH HOFFMANN. • Naylor Dana Institute for Disease Prevention, American Health Foundation, Valhalla, New York 10595 I Introduction II Human exposure'to TSNA HI Bioassays'ofthe TSNA IV Evidence that TSNA cause cancer in humans O~al cancer in smokeless tobacco users and betel quid chewers 2 Lung cancer in smokers 3 Oesophageal cancer in smokers 4 Pancreatic cancer in smokers and smokeless tobacco users 5 Nasal cancer in smokers and snuff-dippers V Approaches to quantifying the relationship between TSNA and human cancer VI Approaches to prevention of TSNA-induced cancers Conclusions Keywords: Tobacco, tobacco-specific nitrosamines (TSNA), NNK, NNN, lung cancer, oral cancer, oesophageal cancer, pancreatic cancer, nasal cavity cancer. I Introduction Major prospective studies in North America and in Europe in the 1960s and the 1970s have demonstrated that the risk of lung cancer moJ:tality for "smokers is 7.8 to 15.9 times higher than that for non-smokers. These findings were confirmed by more than 100 case--control studies and have also clearly established a dose-response relationship between number of cigarettes smoked and the risk of lung c.ancer. The studies are summarized in the reports of the Royal College of Physicians of London (1983) and. of the Surgeon General of the United States Public Heath. Service (US Department of Health and Human Services, !989). These reports aI.s~ "describe the caushl relation- ship of smoking to canc6r of the lhrynx, pharynx, oral cavity, oesophagus, pancreas, renal pelvis and urinary bladder. In addition, cigarette smoking is (~ lmperial Cancer Research Fund 1989 Cancer Surveys [Zol.8 No.2 1989

274 Stephen S Hecht and Dietrich Hoffmann Table 1. TSN. assoriated:with cancer of the nasal cavity" and of the cervix and also wi.~h .leukarmia (.IARC, 1986; Kinlen and Rogot, 1988; US .D.epartment of Health" andHunian Services, .1989). Sni0king.rigars and pipes is. caus.ally fel.ated.to:., I ... p~du.ct cancer of the respiratory tra.ct, oral cavity and 0esrphagu~, ~lthoJag'h th6 ' : " "-" " causal relationship with lung cancer is in-this case not as strong as'that of Snuff, moist" cigarette smoking (IARC, 1986; US" Department of Health and Human Services, 1989). More recently, exposure to environmental tobacco smoke (inv.oluntary, or passive smoking) has been incriminated as a risk factor for cancer of .the lu..ng in no.n-smoker.s.. (IARC, 1986; U.S National R.esearch i Snuff, dry" Council, 1986; US Department of Health dnd Hurhan Servi.ce~, 1989).:• ~" I " " " " Chewing o.f tobacco and especially the oral use of snuff are associated with Chewing cancer of the oral cavity (IARC, 1985; Department of Health and Human tobacco Services, 1986) and possibly with cancer of the nasal cavity, pancreas, kidney and bladder (Brinton et al, 1984; Goodman et al, 1986; Kabat et al, 1986; US Department of Health and Human Services, 1986; World Health Organiza- tion, 1988). In India and in other Asian countries, chewers of betel quid with tobacco and chewers of khani, a mixture of tobacco and lime, are at high risk Mashed" for cancer of the oral cavity, pharynx, larynx and oesophagus (Jussawalla and Zardaa Deshpande, 1971; IARC, 1985). Most of the epidemiological observations on tobacco usage and cancer have _ Nassa been supported by evidence of carcinogenicity from bioassays with whole Cigarettes smoke and with the particulate matter ('tar') of the smoke. Such studies in laboratory animals have been summarized in several reviews (Wynder and Hoffmann, 1967; Mohr and Reznick 1978; US Department of Health and Little cigars Human Services, 1982; IARC, 1986). Bioassays with smokeless tobacco have Cigars indicated, though not proved, that these products are carcinogenic in the oral cavity of rats and hamsters (IARC, 1985; Hecht et al, 1986a; US Department Pipe tobacco of Health and Human Services, 1986). Analytical investigations have led to t.he isolation and identification of approximately 3000 individual components _ in tobacco and 4000 in tobacco smoke (Roberts, 1988) including various aSpecifictot carcinogens in processed tobacco, and a large number of turnout initiators, bn.d.---not d tumour promoters, cocarcinogens and organ-specific carcinogens in tobacco "I. Brunner smoke (Hoffmann and Hecht, 1988, 1989). 4. Brunne It is the purpose of this overview to delineate the contribution of the 7. Hoffm~ tobacco-specific nitrosamines (TSNA) to the carcinogenicity of chewing 10. Ohshi tobacco, snuff and tobacco smoke. Although TSNA as a group represent the most abundant, highly active carcinogens in these products, consideration reactions d~ must be. given to the fact that tobacco extracts and tobacco smoke are highly alkaloids in complex mi~tures.. Conseque.nt.ly, tobarco and its smoke .contain :not only. TSNA .i~ tumorigenic agents but also tum~ur inh~bitors. The abso/'ption'bf biologically Pyr~yhth4~ active agents from mixtures ~s governed by various factors including the 46%bf NN physical and chemical state of the compound and the' pH of the mixture. The from. tobac carcinogenic activities of the TSNA can be influenced by factors such as i during smo alcohol and diet. Despite these limitations there is evidence that the TSNA quantitative contribute appreciably to the increased risk of tobacco users for cancers of the The level lung, oral cavity, oesophagus, pancreas and nasal cavity. -

Tobacco-specific nitrosamines 275 II Hum~an exposure to TSNA Nicotine :~.nd the minor Nicotiana alkaloids (Fig. !) ~epresent a. major=~.r.oup of pharmacologically actiye., compounds in tbb.a6co pr~duets-.(US Department Some Tobacco-Alkaloids N|coWr ne Anabaslne 1,1"-Methylanabaslne Anatablne N'-Methylanatabirle Fig. 1. Structures of nicotine alkaloids of-Health and Human Services, 1988a). Depending on the tobacco type, variety and plant components utilized, processed tobacco contains from 0.5- 5% of alkaloids with nicotine as the predominant compound (90-95% of the total alkaloids). Commercial cigarettes smoked under standardized laborat- ory conditions (Brunnemann et al, 1976) deliver 0.1-3.0 mg of nicotine and up to 0.3 mg of minor alkaloids in the mainstream smoke and 1.3-20-fold higher amounts of the alkaloids in the sidestream smoke of cigarettes, the smoke geri~rated between puffs (Adams et al, 1987; US Department of Health and Human Services, 1989). Cigars of various shapes, sizes and tobacco types, as well as pipes, can generate up to several milligrams of nicotine in the mainstream smoke (US Department of Health and Human Services, 1982). Indoor environments polluted with tobacco smoke were found to contain 1- 13.8 tzg/m3 of nicotine (US Department of Health and Human Services, 1989). The alkaloids together with the aminoacids and proteins a~e'.thg/hOSt • abundant amino compounds in tobacco.products.-Nicotine is h tertia~ .amin~,.." while nornicot!n.e, anataNnd a~d .anabasine at6 .~ed0ndary-arrffhe.g (Eig." 1),. Nitrdsatiori bf alkaloids )iel~ts TSN. A; seven have been identified in "tobacco :products (Fig. 2). Althot~gh.tra~es of some of these nitrosan~ines have hlso ... been found in green tbbadco leaves bef6re harvesting (Andersen et al, 1989; Djordjevic et al, 1989a), the largest ~mounts are formed during the processing of tobacco. The yields of nitrosamines are dependent on the conce/atrations of

II IIIII IIIII IIIIIII1 II III I II , ...... 276 Stephen S Hecht and Dietrich Hoffmann : Formation of Tobac~=o Specific N-Nitrosamines : -. ~o Fig~ 2. Formation of tobacco-specific N-nitrosamines alkaloids and nitrate in lamina and ribs and on the tobacco processing methods including curing, fermentation and ageing (Tso, 1972; Brunnemann et al, 1983). As seen in Table i the highest concentrations of TSNA have been determined in snuff. This is due to the favourable conditions for TSNA formation during fermentation (Hoffmann and Hecht, 1988; Djordjevic et al, 1989b; US Department of Health and Human Services, 1989). The sometimes exceptionally high TSNA levels in commercial snuff are a likely consequence of the ageing of products during a long shelf life. But even concentrations of the TSNA in smokeless tobacco products of recent manufacture exceed those of nitrosamines reported in other consumer products by at least two orders of magnitude (US National Research Council, 1981; Preussmann and Eisen- brand, 1984). Analyses of saliva of tobacco chewers and snuff-dippers have demonstrated that the TSNA are extracted from these tobacco products (Hoffmann and Adams, 1981; Nair et al, 1985; Palladino et al, 1986; Bhide et al, 1986; Brunnemann et al, 1987; Oesterdahl and Slorach, 1988). While the individual TSNA in saliva may reach levels up to 400 ppb, unusually high concentrations (up to 2600 ppb N'-nitrosonornicotine (NNN) and a mean of 980 ppb) have been measured in the saliva of some Canadian Eskimos who dip snuff (Brunnemann et al, 1987). Factors other than high TSNA levels in tobacco which contribute to high TSNA concentrations in the saliva of ~nuff-dippers .. include number qf years, of practis.ing the habit and the frequency of chewing .It has been. estimated on the b~sis.of'usilag popular )krr~ei~an products in r. 1987--188 that. a snuff/dip/gei v41i6 'co.n~.iimes 10:~/day over' ~i perio .d of 40 years is thus exposed to .about 48.00 mg Of NNN ~nd 260 mg of 4-(methylnitrosamino)- 1-(3-pyridyl)-l-butanone (NNK). This exposure to NNN and NNK, the two most carcinogenic TSNA, is not far below the levels which induce tumours in rats. These calculations of human exposure to NNN and NNK may be lower than thereal levels of exposure if one considers the likelihood that nitrosation

~e ly ly .ly Table I. TSNA in tob.acco Tobacco-specific nitrosamines 277 1"ob~cco" Country .product , , Snuff, moist USA . C.n'rrada " UK Sweden Denmark Snuff. dry USA Che.wing USA tobacco UK Sweden FRG Belgium India .Thailand MasherP~ndia Zarda" Nass~ India USSR "Tobacco-s.p. ecific nitrosamines~ (ppm) .NAB&NAT Reference* 1,2,3;10 " 2,4,7 5. 2,3,6,10 2 2 3 2 2,3 • 10 2,3 3 3 3,9 1 Cigarettes Cigars Pipe tobacco USA 0.6-7.9 0.1-1.3 0.5-5.8 7,10 UK 0.3 0.1 0.2 7 France 0.58-18.6 0.13-1.5 0.23-10.0 7,8,10 USA 11.2 4.5 13.0 7 USA 3.0-10.7 1.1-3.5 2.5-33 7 Netherlands 6.8-53.0 2.9-4.3 4.6-20.4 10 UK 3.0 0.6 2.5 10 France 6.9 1.1 4.9 10 Netherlands 3.8 n.d.b 2.0 10 . pecffic tobacco products used m certain regions (.see WHO, 1988) Un.d. = not detected "!. Brunnemann etal (1985); 2. Hoffmann and Hecht (1988); 3. Tricker etal (1988); 4. Brunnemann et al (1987); 5. Hoffmann et al (1988); 6. Oesterdahl and Slorach (1988); 7. Hoffmann et al (1984a); 8. Djordjevic et al (1989b); 9. Tricker and Preussmann (1988); I0. Ohshima etal (1985) ae :from tobacc~-'by, dii-ect transfer, while the remainder is igyros3;nthesized as . diaring smoking (Hoffmann et al, 1980; Adams et al, 1~983). Table2 presents. A '~ quantitative analytical data for TSNA in the smoke of cigarettes and cigars. ae~ The levels of TBNA in tobacco smoke are up to 100-fold higher than those ' reactions during chewing will yield additional amounts of TSNA from the alkaloids in the snuff (Nair et al, 1987). TSNA in .cigarette smoke" originate from tobacco as v~ell as being . . .pyrosynthesized durin~ smoking. Under stand~'rd smoldng ~6ndi~bns .40-;... 46% o.f NNN'and 2623-7%. ~f Nf~K i~a cigarette l~ainstt~am sr~oke originate

278 Stephen S Hecht and Dietric.h Hoffmann Table 2. TSNA in mainstream tobacco smoke Tobac(o. .Country product. ... Cigarettes ".." F USA" F-VLT FRG. F-LT F-MT ", F-HT F NF" NF-O NF'V NF-T NF-B NF-BL NF-BL Fr/mce NF_Bua .. NF_Va .. Little cigars F USA Cigar NF USA Fraffce . USA FRG Tobacco-~pecific nitrosarnines (ng/cig) • "" " 310 ~50 24-106 6-69 38-99 26-55 "19=-179 . . 21-1"45 11-122 27-73 1000-3200 "~ 1.90--430 1202950. ": .... "80-770' 3-19 n.d.-4 16-32 36-91 77 59 85-255 70-156 "512-625 108-432 575-590 127-220 3700 320 620 420 Reference* NAB di N)AT • " 370 15-128 33-98 • 35-285 27-108 190-649 :.'" ~40-990" 6-20 40-90 102 80-225 • 2.66-353 200-350 4200 410 2 2 2 2 2 5500 4200 1700 4 3200 1900 1900 4 aExpedmental cigarettes F, filter; NF, non-filter; VLT, very low tar; LT, low tar; MT, medium tar; FIT, high tar; O, oriental; V, Virginia; T, Turkish; B, blended tobacco; BL, black tobacco; Bu, Burley tobacco; n.d., not detected *1. Hoffmann et al (1984a); 2. Fischer et al (1989a); 3. Djordjevic et al (1989b); 4. Hoffmann et al (1980) of: o.the.r nitrosamines, in the human environment; except in some specific occupational settingg (US Ni~tional Resghrch Council, 1981; Preussmann and Eisenbrand, 1984). On the bas~s of TSNA yidlds such as are present inthe smoke of a machine-smoKed non-filter cigarette, human exposure estimates at a rate of 40 cigarettes per day over a 40-year span would reach about 590 mg of NNN and 250 mg of NNK. Again, we believe that this is a low estimate since, among other factdrs, it is arrivedat without regard for the possible endogenous formation of TSNA due to alkaloids and nitrosating agents inhaled as constituents of tobacco smoke. Consideration must also be given to TSN puff-ta .TSNA • smoke. • " ,17.3 n~ " sidestr¢ pollute, Levels ~ ng/~ f, The int their c; Rivens~ assayed (methyl Their o exampl, irrespec elicits t~ subcuta observe (LaVok maligna given b~ N'-nitro nitrosoa NNAL ~ • recently IV Evi~ 1 Oral Epidemi tobacco, as high z the fact that smoke yields obtained, under standi~rdized machine-smoki.'ng • IARC, " • ."..... :. '. .. conditmns ..are fre.q.uently'low.ertlaan th.o~e generated by clgar.ette smokers, ~ : mokele: • " "" ' ":..v)h6 ~le~iat~'fr~m~'~ta~nda~l'coi~ditiong;.b'y'dr~.v~i_'ng "15"fiffs.f~r ffabr~"ft~qUer~tly..l."~bm~i~sii • " of low than doe • a~nd i/~halin.g, th&n ~.o:.fe !nteflseiy.. Th/'s, ~[pplies"e.spedially io srr[ok~s ' .." ~ l' ," ." yield cigarettes (US Department of Health ~nd Human Services," ~989). One [ "type~" of would, therefdr~, hi,re t6 ffssunie additional body burdens of TSNA for hyd.es;-p exposure estimates among'low-yieid Cigarette smokers (Fischer et at, 19.89bj. [ mann et

Tobacco-specific nitrosamines 279 TSNA are also generated during sidestrea/n smoke formation !n. b.etw~en m ~ .... 19uff-tiaking. Under standardized:conditions in the lhbbr.atory; the rele..ase. ce~ . ,..I.TS.N.A" into sjd~stream" srrioi~d 'e~.ce:6~ts.tlie leqdls ~en~rated, iri inainstr~a~ smgke.This is especially pronounced in the calse of cigarettes with perforated filter ~i~,. A low-yield American tiltercigarette delivered 66.3 ng NNN and 17.3 ng NNK in mainstream smoke, and 338. ng NNN and 386 ng NNK in sidestream smoke (Adams.et al, 1987). Attempts to measure TSNA in smoke- polluted indoor environments were until now reported only in one instance. Levels ranged .from below detection limit .to 3.7. ng/m3,-f.or NNN ~ind .up to'. fi~/m~ foi NNK ~Kl~is~t'til, 19g~/)/ " III Bioassays of the TSNA The interest in the TSNA in ~obacco andt0bacco smoke is primarily due to their car.cinogenicity in laboratory animals (Hecht and Hoffmann, 1988; Rivenson et al, 1988). Five of the seven TSNA identified thus far have been assayed for carcinogenic activity (Table 3). Three of them, NNN, NNK and 4- (methylnitrosamino)-l-(3-pyridyl)-l-butanol (NNAL), are strong carcinogens. Their organospecificities depend in. part on the route of administration. For example, in rats the powerful carcinogen NNK causes turnouts of the lung, irrespective of the mode of application. When given in drinking water, it elicits tumours of the exocrine pancreas in addition to lung turnouts. Upon subcutaneous injection tumours of the nasal cavity, liver and lung are observed. NNK, .but not NNN, is also a tumour initiator in mouse skin (LaVoie et al, 1987). NNN given in drinking water causes benign and malignant tumours of the oesophagus as well as nasal tumours in rats; when given by subcutaneous injection it induces mainly tumours of the nasal cavity. N'-nitrosoanabasine (NAB) is weakly carcinogenic, in rats while N'- nitrosoanatabine (NAT) is non-carcinogenic in rats at doses up to 9 mmol/kg. NNAL elicits tumours of the lung and pancreas in rats. Iso-NNAL and the recently identified iso-NN-acid are currently being assayed for carcinogenicity. IV Evidence that TSNA cause cancer in humans 1 Oral cancer in smokeless tobacco users and betel quid chewers Epidemiological studies have demonstrated that chronic use of smokeless tobacco, in the form o~ snuff-dipping, causes oral cancer. The relative risk is as high as 50-fold.for c~nc.er of the~gum and:b.u, cc,al.mucqsa.(Wj.!m.~.t a./,~ 198.1; .I.A-RC, .t~85'; US" D~p~'tfnent:df"Healtfi.and H.ii.mhn Se/fices, 1986). Si{a~e smiskeless tobacco is nota, dom6ustion product like tobAc6o srhok6; .iti .composition" is simpler aiad'it contairis a less complbx mixtur~ of carcinogeris than does tobacco smoke. The only ¢~rcinog.ens kndwn to be p.reseiat in the types of smokeless tobacco used for snuff-dipping are nitrosamines, alde- hydes, polonium-210 (2t°po) and polynuclear aromatic hydrocarbons (Hoff- mann • et al, 1987). NNN and NNK are typically present in amounts ranging..

280 Stephen S Hecht and Dietrich Hoffmann Table 3. Carcinogenicity of TSNA TSNA . Animal Route of " Principql Dose ~ Reference* Hum than th (strain) " .. .i applica.tion. . . targdt organ . . (mmol/a~imal) • ~us~?(S.en~ar) "Topicai.(TI').a. None'. .-" )'-~.~'.: ..... 0.0i8 ,....~.. : M~ia~e (A~.J)" . " I~tr.aperito'nelil. "Lth~'g" 'q:.." i' ' ? ". ") " 0.-1 :: . i :.'." NNK Rat(F344) " Subcutgneo.us Nasal cavity, Oral Rat (Spra~ue- Oral Dawley) SG Hamster Subcutaneous Mouse (Sencar) Topical (TI)a Mouse (A/J) Intraperitoneal Rat (F344) Subcutaneous Oral • • SG." Hamster Subcutaneous NNAL Mouse (A/J) Intraperitoneal Rat (F344) Subcutaneous • NAB Rat (F344) Oral SG Hamster Subcutaneous NAT Rat (F344) Subcutaneous oesophagus ". .0.2-3.4 Oesophagus, nasal cavity i.0~3.6 Nasal cavity 8.8 Tracl~ea, . • "" ha~al cavity'.: .... "" ~0.9-~.1 Skin 0.028 Lung 0.02-0.12 Nasal cavity, lung, liver 0.2-2.8 Lung, pancreas, liver 0.075--0.31 Trachea, lung, nasal cavity 0.005-0.9 Lung 0.12 Lung, pancreas 0.32 Oesophagus 3-12 None 2 None 2.8 1 2,3 2 4 2 5 4 2 2 2 carcino, tQbgcco laborat( • 1986.a; . compon et al~ 19 surgieali and this extract, in comb - .ttimours is ubiqui a cocarci Oral ~ chewing only wh~ NNN an, saliva ol (Sipahim • aTI, tumour initiation assay with TPA as promoter * 1. LaVoie et al (1987); 2. Hoffmann and I-Iecht (1985); 3. Hecht et al (1988a); 4. Rivenson et al (1988); 5. Castonguay etal (1983a) NNN, N'-nitrosonornicotine; NNK, 4-(methylnitrosamino)-1-(3-pyridyl)-l-butanone; NNAL, 4-(methylnitrosoamino)-l-(3-pyridyl)-l-butanol; NAB, N'-nitrosoanabasine; NAT, N'-nitrosoanatabine from 1-l.00 /zg/g~i(Table 1),. which .is about 1000-fold .ab.ove'the levels of p"blynuclear akomatic hydrocarboris. F~rmaldeh'~de, acetaldetiyde arid 'cro- tonaldehyde are presenVat levels from 1-10/zg/g, and 2~°po from 0.2-1 pCi/g. NNN and NNK are the only carcinogens in smokeless tobacco that have been shown to induce oral cavity tumours in laboratory animals (Hecht et al, 1986a). A mixture of NNN and NNK, swabbed daily on the oral tissues of rats in a total dose of 1.6 mmol/kg, induced papillomas of the oral cavity in 8 of 30 rats. The calculated exposure of a snuff-dipper to NNN and NNK, over a 40-year period, is approximately 0.4 mmol/kg, which is similar to the dose which produced tumours in rats: The presence, of NNN and NNK in the saliva ....: 9(snu. ff~dipper.s h,hs. b.e.e, n co.rlftrm.,eid. ( .Ho .ffma~.: ar~, ¢: Adam._...s;.1..9.8i)-. Met..abofi.'s..m... • ~tudi'eff ivlth "hurri~ih fis~fle~"fiavd "i:l~.rh6h~trafed: thhl]:..buedal ni~cosa can act{val?e:NNN and NN.K to.intermediates tha~ carl binfft9 DNA, as observed with oral tissue of rats (Castonguay. et al, 1983b, 1984). Taken togethei', these. data provide strong support for the tdle of NNN and NNK as causative factors in oral cancer induction by smokeless tobacco. al, 1987) strongest and the Nitrosam present i pionitrile .or.al.ca.rci • Thus tt causative 2 Lung, Since the lung canc Horn, 19: pipes is Hu.man "...toba.ceo, : :I)igher'ri~l .: 1983),,.Le' smoke df • ,In .addi several

Tobacco-specific nitrosamines 281 qAT, ~ls of t cro- ~Ci/g. been et al, ff rats of 30 ~ver a 'sali~ad°Se [ ~olis.m~ a can .erved these actors Human exposure to NNN and NNK through snuff-dipping is more interise than .that to any o[her .nitrosamine, and possil~ly that. to any other., strong : :. c~irc.in~gen; it r.epreser~ts ffri u~acceptable: rigt~ Nvhich.'..sho'uldbe'. 6.orrected bY legislative action. -. .. Synergisms"may be "impdrtant in oral cancer induction b'y smokeless tobacco. Extracts of snuff, have not been proved to induce tumours in laboratory animals, despite the presence of NNN and NNK (Hecht et al, 1986a; US Department of Health and Human Services, 1986). In fact, surgically, cre~tted canal in the lower lip of the rat (Hirsch and Johansson, 1983; Hecht et al, 1986a). Chronic irritationmay enhance oral carcinogenesis, and this may play a part in accounting for the fact that whole snuff, but not its extract, can induc~ oral tumours in the rat (Konsta.ntinidus et al, 1982). Snuff in combination with herpes .simplex virus type I has been shown to induce tumours in the Syrian golden hamster oral cavity (Park et al, 1986). This virus is ubiquitous in man (Nahmias and Roizman, 1972) and may be important as a cocarcinogen with NNN and NNK. Oral cavity cancer is the leading cancer in males in India. It is cz'used by chewing betel quid with tobacco. The evidence for cancer causation is strong only when tobacco is included in the quid (IARC, 1985). The presence of NNN and NNK in the tobacco used for preparing the quids, as well as in the saliva of betel quid chewers, .tias been demonstrated in several studies (Sipahimalani et al, 1984; Wenke et al, 1984; Nair et al, 1985; Prokopczyk et al, 1987). As in the case of smokeless tobacco, NNN and NNK are the strongest carcinogens known to be present in the tobacco used for the quids, and the only ones known to induce oral tumours in laboratory animals. Nitrosamines derived from arecoline, the major alkaloid of areca nut, are also preserit in the quid and in saliva. One of them, 3-(methyln'itrosa.mino)pro- pionitrile, is a powerful Carcinogen in the rat; its activity as a locally applied oral carcinogen has not yetbeen evaluated (Prokopczyk et al, 1987). Thus the available evidence strongly supports the role of NNN and NNK as causative factors in oral cancer induced by tobacco products. 2 Lung cancer in smokers Since the first large scale epidemiological studies on cigarette smoking and lung cancer (Wynder and Graham, 1950; Doll and Hill, 1952; Hammond and Horn, 1958) it has been well establi.shed that smoking cigarettes, cigars and tob.acco, such as are.common in.France, North Nfrid~ and..Caba, have a "higher risk fdr lung canc6r than 'do smokers Of" bleiaded cigarettes (Joly e~. 1983). Levels of TSNA in the smoke of black cigarettes are higher than in the smoke of other cigarettes (Table 2). In addition to epidemiological observations there is strong evidence t~rom several laboratory studies which supports the concept that TSNA cont.n.'bute

282 Stephen S Hecht and Dietrich Hoffmann i. appreciably to the risk of lung cancer in smokers. NNK induces benign and ' malignant lung tumours in mic+, ~'atg and hamste~?s (Table 3). In rats 30 mg: ".- ! . )N.N .K/kg, gi.ven, ifi, ttie. dfi~l~in~ W~it.~.t., i.ncluees, ffi.aligi~.a.nt, lung tumofirs "" ;[ • (Rivensofi etal, 1988) ;.~h!16 a.)ingle.sub~ut.~ife~s aps~ ;f i mg NNK. i.n~luces : ' ~"" " a s;gnifi~antinciden~e-ofrespil~atory tract tumours in l~amster~I (He.cht et al, 1983a). This latter dose of 1 mg NNK per hamster corresponds to about 6 mg .NNK/.kg. A smoker of 40 cigarettes per day. (425 ng NNK/cigarette) is ex.pose.d to abgut 250 mg bf NNK" or 3.6 in),/kg over a .40-year.period.. Hamsters treated "v0ith NNK by'intratracheal instillation devel.o.ped preneo- .pl.agtic'cellul~r ch~ia'grs similar, to.thrse i3bserar~d in-the pulmomir.y eigithdlium' of smokers (Boutet et al, 1987). Huinan respi)atory epithelium treated with NNK in vitro,, and after transformation transplanted into nude mice, produced undifferentiated carcinoma (Parsa et al, 1986). .- As we have discussed it~ earlier review .articles, NNK and NNN are metabolically activated by a-hydroxylation to intermediates which bind to DNA in the lung (Hoffmann and Hecht,1985; Hecht and Hoffmann, 1988). Tissue explants obtained from human bronchi and peripheral lung have the capacity to metabolize NNK and NNN by a-hydroxylation to reactive and DNA-damaging electrophiles (Castonguay et al, 1983b). In mice and rats, NNK-derived elctrophiles react with nucleophilic centres in DNA to yield a variety of products including O6-methylguanine (Hecht and Hoffmann, 1988). The latter adduct has been shown to cause miscoding (Loechler et al, 1984). In strain A mice NNK induces lung adenomas which contain an activated K- ras oncogene (Belinsky et al, 1988). In lung adenocarcinomas of smokers the K-ras oncogene appears to be activated by point mutations in codon 12, which result, in part, from O6-methylguanine-induced miscoding. It is likely that K- ras oncogene activation is an important event in the pathogenesis of adenocarcinomas of the human lung (Rodenhuis et al, 1988; Fig. 3). ~CIH~ Processing ~-~O I;I.N=O Metabolic or ~.y~ C,:z Act,vat, ....d other DNA with actlva,ed Smoking Nicotine NNK Fig. 3. 'Scheme linking nicotine to formation of DNA adducts including O6-methyguanine. The latter leads to DNA miscoding, activation of K-ras oncogene and lung tumQurs Th.e 'epidemioldgical..and biochemical data discussed'above do not.prove ....] - .t.h.at. the ".T.SN.~ ga.use~.l.tmg .egaaeet in. ~aokerg'~4au.t,!d~. l£ad..t6 the'e6~adtis.~o/a~ . that thrT.SNA represent an imiabi-ta'n[ risk faetdr for lur~g cancer in smokrrs, 30e~ophage:~l cancer in smokers" • ." • " • Tobacccr smol~ing is an important cause of oesophageal cancer (IARC, 1986). Nitrosamines are the only known constituents of tobacco smoke tliat are organospecific for cancer of the oesophagus in laboratory animals. Four nitrosam " hitrosom ...' NNN .is b ""iffdu~es. t tile drink 1983b; ( approxim tion to r~ could be • NNN'li~ oesophag different 1983b). A the role c evidence smoke co~ 4 Pancr~ Cancer of both men Society, 1' cigarette ~, women. A the relatN (IARC, 1~. study in N for snuff-d So far,, induce pm enzymatic water give pancreas i~ drinking ~ (Rivenson events suct not produc available d tobacco use qurstion. - • :..-:-! .~.':" 5' Nasal ee .Alfiio~gh n o~cupat.ion. the nasal p dippers (Br is suggestiv

gn and 30 mg tmours aduces t et al, I.. ltrmg '.1 :" .tie) .is ~driod. reneo- helium d with mice, N are dnd to ~ve the ve and d rats, gield a 'mann, 1984). ted K- :sis of he. The prove :fusion .kers. ".2 1986). at are X:~our Tobacco-specific nitrosamines 283 nitrosamines in tobacco smoke--NNN, NAB, N-nitrosodiethylamine and N- nitrosomethylethylamine--cause oesophageal tumours in rats. Of these, NNN is by far the most prevalent in mainstream cigarette smoke (Table 2). It induces .high incidences of oesophageal turnouts in F344 rats when .given in the dlrinking watefor i.n a liquid dirt. (Hoffmann et al,. 1975; Hecht et. al, 198.3b;-Castongu@ .~.t /il;'.1984). .Only:.~el~tively high. ~tota! .d~se~ of approximately 3-9 mmol/kg of NN.N have thus f~r been tested b3~ administra- tion to rats in the drinking water. Lifetime exposure of a smoker to NNN could be estimated as approximately. 50 /zmol/kg. Metabolic activation of NNN has been observed in ct/ltured rat oesophagus as well as cultured human ".oesophagus, but the extents of activation by the various pathways are "..d.iff~ere..n.t. in the rat.and ttum.an...tissues.(Hechtet al,,1982;. Castonguay et."al; 1983b). Although further studieswill be necessary to more definitely establish the role of NNN as a cause of oesopha.geal cancer in smokers, the available evidence is suggestive and more compelling than for any other tobacco smrke constituent. 4 Pancreatic cancer in smokers and smokeless tobacco users Cancer of the pancreas, one of the major cancers in the USA, has increased in both men and women during the last three decades (American Cancer Society, 1988). Both cohort studies and case-control studies have shown that cigarette smoking is caus~ally associated with pancreas cancer in men and women. A dose-response between number of cigarettes smoked per day and the relative risk of developing cancer of the pancreas has been established (IARC, 1986; Department of Health and Human Services, 1989). In a large study in Norway, an increased risk for cancer of the pancreas was indicated for snuff-dippers and tobacco chewers (Heuch et al, 1983). So far, only two agents have been isolated from tobacco products which induce pancreatic tumours in laboratory animals. These are NNK and its enzymatic reduction product NNAL. A dose of 1.0 ppm NNK in the drinking water given to Fischer rats in a lifetime study led to turnouts of the exocrine pancreas in 9 of 80 animals, while 8 of 30 rats treated with 5 ppm NNAL in drinking water developed adenoma and adenocarcinoma of the pancreas (Rivenson et al, 1988). NNK has been shown to induce early morphological events such as metaplasia and hyperplasia in human pancreas explants but did not produce carcinoma in this in vitro system (Parsa et al, 1986). While the available data on the possible contribution of TSNA to pancreatic cancer in tobacco users are limited, they do indicate a need for an in-depth study on the question. 5 Nasal cancer in smokers ahd snuff-dippers Althou~,h r/asal.~ancer in "humans is.r~i.r.e; as{de.: .fr~.m i.N .oecu.r?ene~ ir~ certain." occfipatibnals~t~ihgg,.it .~pp.eai:s that ti~e"risk forsqfiamous c'elf darcinoma of the nasal passagei is signifie.antly elevated in long-term smokers and snuff-. dippers (Brinton e.t.al, 1984). The observation of nasal cancer in snuff-dippers is suggestive of an organospecific effect. In rats and hamsters both NNN and o~ o I

284 Stephen S Hecht and Dietrich Hoffmann NNK induce tfimours of ,the nasal cavify when administered by.subcutaneous injectio.n .(Hoffmann and Hecht, •1985). NNN. anff o~z.e of. itg maj.o~ " "m6iabqiites~ ~NN-~N-0x.id.e.,::als.o. give.. na..~hl .cav.ity ~u.mb.fi.rs w.he.n ~inis, . tered orally tb rats (He.cht .et al, 1.983.b;' ~Castonguay. et al, 1.984), The malignant tumours induced by subcutaneous •injection bfNNN or NNK are found mainly in the olfactory portion of the nose. However, squamous cell ~carcin.oma, like those seen in humans, are the predominant malignant tumours induced in F344 rats by oral administration of NNN and NNN-N- • . oxide .(H.e.ch.t. et .a/,~.19.8.0, 19.8.3.b; Caston~.uay et al, 1.984; Hoffmann e.tal, 1984b). In ratg, the nasal mucosa has ah 6XCel~tionally high bapacityfor, the • • metabolic activation of.NNN and NNK; consequently, extensive methylation of DNA in the nasal mucoga is observed in NNK-treated rats (Brittebo et al, • 1983; Hecht et al,. 1986b). The tentative link between TSNA and cancer of the nasal cavity in humans is worthy of furthdr investigation.. V Approaches to quantifying the relationship between TSNA and human cancer Fig. 4. Metabolic activ DNA and protein The hundreds of millions of tobacco users in the world represent a unique population that is exposed on a regular basis to TSNA and other carcinogens. The risk of smokers and tobacco chewers for developing lung cancer and/or other tobacco associated cancers is great enough to be measured in epidemiological studies. How can we quantify the role of TSNA in the induction of these cancers? At present, the most promising approach appears to be the measurement of macromolecular adducts of these nitrosamines in tobacco users. In prospective epidemiological studies, it may be possible to relate these adduct levels to risk for cancer development. Although the levels of TSNA in tobacco products have been extensively investigated, and valid.a.ted (Tables i and 2), there remain uncertainties about the extent .of individual uptake ofthese, carcinogens. Thig will-depend, for example, on inhalation practices, modes of chewiiag or dipping and exposure to environmental smoke (IARC, 1985, 1986). These individual variations can be assessed by measuring such parameters as carboxyhaemo- globin or plasma cotinine, but it is not known whether these components or others could be surrogates for uptake of TSNA. The extent to which TSNA m.ay be formed endogenously is also not clear, although the endogenous related carbonium i~ of unknown structur mild base, releases t 1988a). At present, analysis of 9 release, Our continuing res superior to either 32~ 9 or its precursor ad~ also a methylating methylguanine in hu .9 6r related materia] .or endogenous sou~ measurement of co~ related to TSNA. ]3 tobacco products, th to tobacco product~, tobacco associated c~ formation of N-nitrosgproline in smokers has been dem6nstrated (Hoffmann and Brunnemann, 1983; Bartsch .and Montesano, 1984; Ladd et al, 1984). . ..VI Approa.ch.es to. • .~..M~io~ int.e~n.diyid.u.a.l ~diffe(.rence.s i.n the e.xten..t of ~aetabolic .adtivat!on of . .:.. It is .clear ,~hat the : " " TSNA have been~ observe~t in "cialtur~d hiamafl.t.issxf6~, (Casfon~6a!z ei" a/, ": ~: . "~oba~'p~(~d/t~ts .an, 1983b)..These observations demonstratb the need.for a methbd t6.'asses.s an. ~".: .and to 'a~oid exp¢ individual's uptake and metabolic activation Of TSNA. Among" th~ val:iou~ methods that might be considered, haemoglobin adducts aiad DN)k addlacts appear to be the most promising at.present (Hecht et al, 1988b; Wogan, 1988) As illustrated in Fig. 4, metabolic activation of NNK and NNN lead to a common reactive intermediate, tho.ught to be the diazohydroxide 7, or a measures.and smoki US. Among males, over the period 196: and 28% (US Depa~ relationship betwee

d e ;. Y d tl )- ,r \ S ,f rl s s ) a Tobacco-specific nitrosamines 285 .... : ": :.7: X" o Fig. 4. Metabolic activation of NNK and NNN to hypothetical intermediates which bind to DNA and protein related carbonium ion. This reacts with both DNA and globin giving adducts of unknown structures. Hydrolysis of DNA with strong acid, or of globin with mild base, releases the keto alcohol 9 (Carmella and Hecht, 1987; Hecht et al, 1988a). At present, we are developing mass spectrometric methods for the analysis of 9 released by hydrolysis of DNA or globin obtained from humans. Our continuing research indicates that mass spectrometry is likely to be superior to either 32p-postlabelling or immunoassays for the quantification of 9 or its precursor adducts in human blood or tissue samples. Although NNK is also a methylating agent, and approaches towards measurement of 06- methylguanine in human DNA are promising, we favour the quantification of 9 or related materials (Foiles et al, 1988). There are numerous environmental or endogenous sources of DNA ~and globin methylation. In contrast, the measurement of compounds such as 9 is more likely to be unambiguously related to TSNA. Because these nicotine derived compounds occur only in tobacco products, these measurements can be specifically related to exposure to tobacco products and may provide a realistic index of susceptibility to tobacco associated cancers. Vl Approaches to prevention of TSNATinduced" ca.ncers .- It is .c.lear-that::the most'effe~'/.iv6 ~vay .o~.:ivre~,~nti~"~a'ncer ~flduc~idn)by .'~" tobacdo pr6dfi~t~ hnd their cbnstituents is to avoid using toba'.6co in any form, and to avoid exposure to environmental" tobacco smoke. Educational measures and smoking cessation programmes have had a major impact in the US. Among males, the percentage of smokers has dropped from 52 to 33 over the period 1965-87. Among females the corresponding figures are 34% and 28% (US Department of Health and Human Services, 1989). An iziverse relationship between the number of years of education and smoking

286 Stephen S Hecht and Dietrich Hoffr~ann l~revaIence hasbeen established .(US Department of Health and Human mice .PEITC vi Services, 1988.a):-Despite th.rse g~iins and:.the, a.tt.endhnt decrrase in lung.~ well as O6-meth ~ancer in~ider~ceiri(3~6un~ ~ol-i6..~.~.'ttia.t. h~s:b'ee~ obser#~d s{he.e 1986, the~..e ' .!-. i.-:i.,. "trr~te~l~.ffi.~h .Nb are still more than 50 mill.!onsmokers in thd .US and- more than 10 iriilliori, :. ~." bti~- h/~d no effrc snuff-dippers (IARC, 1986; US Drpar~tn~rit of Health andHuman Services, 1989). There are htindreds of millions of tobacco users worldwide. In developing countries especially, tobacco use is on the rise (Tominaga, 1986). Presently, ,there are 250 million smokers in mainland China wh.ere a major epidemic, of.tung.ca.ncer., has .been predicted (,peto, 198.6; Crofton, 19.87). Thes~ dauflting statistics attest to l/he pontiac3;' ~f'n'icotihe"as a "habituating agent. There is a.very, low probability that tobacco use will be eliminated in the near future. In view of this, one must consider devising methods to reduce exposure to carcinogenic constituents of tobacco srrioke as a means of lowering cancer risk in those individuals who continue to use tobacco products. Epidemiological studies have demonstrated that smokers of filter cigarettes have a lower risk for developing cancer of the lung and larynx than those who use non-filter products (Wynder and Stellman, 1979; IARC, 1986). There is an inverse relationship between cancer risk and exposure to cigarette smoke total particulate matter (IARC, 1986). Based on the arguments presented in this paper, we assume that part of this decreased risk is due to reduced exposure to TSNA, the levels of which are decreased with filtration to approximately the sameextent as other constituents of total particulate matter (Adams et al, 1987). The further reduction of TSNA levels in tobacco and tobacco smoke should be a major priority. Standards for permissible levels of TSNA in tobacco products should be developed by national and international public health agencies. Continuing research on the formation of TSNA has confirmed our initial finding that they are produced primarily during the curing an.d processing of tobacco (Hecht a.nd Hoffmann, 1988; Andersen et al, 1989; Djordjevic et al, 1989a). Tobacco bleflds made' with Burley stems, which have a high nitrate content, have higher levels of TSNA than those products with lower nitrate (Brunnemann et al, 1983). International compar- isons of TSNA levels in products such as snuff have clearly shown that processing techniques can lead to significant reductions in the levels of these carcinogens in tobacco (Brunnemann et al, 1985). From 26-46% of the NNN and NNK in tobacco is transferred into mainstream cigarette smoke while the remainder th..at is found in smoke is foi'rned by the reaction of alkaloid pr.ecu.rsorg with nitrog.ext oxides (Hoffmann et al, 1980; Adams et al, 1983), N~trar.e coiatent-,of tobacco, is' correlatdd...wi~ TSNA le.velr~ in mainstream'" smoke (Ada.ms ~t al,"i98.4). Thus, ~fie"&~ ~f-I~w~i~ iii~r.a~e'l~eiads,!w.oti~l" lead'" to lower levels of TSNA in smoke (Hecht and Hoffmann, 1988).:HoweVer, it should .be noted that this strategy may le.ad t6 increased levrls of p61ynucleai~ aromatic hydrocarbons in mainstream smoke (Hoffmann and Rathkamp, 1968). Chemopreventive agents which inhibit TSNA-induced cancers are being developed. One promising candidate is phenethyl isothiocyanate (PEITC). In .: hl;i989b). ~EIq inhibition of NI lead compound These studies s~ induce cancers • carcinogenic 1St( The design c Groups of subje are taking non- should be possil to assess the po~ be desirable to ~3-carotene and Services, 1988b decreased. VII Conclusi¢ Tobacco use, e developed coun of all male can region) and 28~ can be attribute and Gori, 1977 USA to t.obacce are lung, uppe~ urinary bladdm biochemical dal In this overv alkaloid-derive, the oral cavity tobacco smoke and nasal cavi tobacco. The ~ ~. tobacco and cm ," ..ap'd ,NN)k~ ar, their I~vels in bioassay doses metabolically a in DNA. Such causes activati~ have been obse

tn )n In or in tO ;al sk ;er se tal tre in )lic ~.he al, ns~ 3se .ar- hat . ese the oid am ~ad -, it ear np, :ing • In Toba.cco-specific tzitrosarnines 287 mice PEITC virtually completely inhibited NNK-induced lung. tumours as well as O6-methylguanlne f.orma.tibn.in lung DNA (Morse et al, 1989a). In rats" treated with NNK, PEITC c~us~ct a.50%.decr.e.ase in.lung tu.mo, ur incidence but l~d nO effect. on tumourinduction in the liver and nasal cavity (Morse et al, 1989b). PEITC appears to have minimal toxic dffects in rats at doses where inhibition of NNK tumorigenesis is observed. Thus, PEITC is a promising lead compoundupgn which further structure-activity studies are being based. These studies should lead to new insights on the mechanisms by which TSNA induce, cancers in laboratory arii.ma.ls ahd on methods for inhibiting the carcinog'e£aic ~rocdss. '- ...... " " " ' " " : "" " " " The design of intervention studies in tobacco users can be envisaged. Groups of subjects who are using products with lower levels ~f TSNA, or who are taking non-toxic chemopreventive agents, can be studied. Initially, it should be possible to measure levels of TSNA haemoglobin adducts in order to assess the potential effectiveness of tl~ese strategies. Subsequently, it would be d~sirable to carry out prospective trials, as are presently being done with 13-carotene and related compounds (US Department of Health and Human Services, 1988b), in order to determine whether cancer incidence would be decreased. VII Conclusions Tobacco use, especially smoking cigarettes, is a major cause of cancer in developed countries. It has been estimated that between 1960 and 1975 30% of all male dancers and 7% of all female cancers in England (Birmingham region) and 28% of all male cancers and 8% of all female cancers in the USA can be attributed to the use of tobacco (Higginson and Muir, 1979; Wynder and Gori, 1977). Doll and Peto (1981) attribute 25-40% of all cancer in the USA to tobacco use. The sites that are most at risk for cancer in tobacco users are lung, upper respiratory and upper digestive tract, pancreas, kidney and urinary bladder. These epidemiological data are supported by chemical and biochemical data and by in vitro and in vivo bioassays. In this overview we have discussed the concept that the tobacco-specific alkaloid-derived nitrosamines (TSNA) contribute appreciably to cancer of the oral cavity in snuff-dippers and to cancer of the lung and oesophagus in tobacco smokers (Table 4). They also play a part in cancer of the pancreas and nasal cavity in cigarette smokers and possibly in users of smokeless tobacco. The concept that TSNA have a major role in the association of tobacco and chncer is based primarily on the fol,lowa.'ng evidence.: N.NN, NN.K. ' and-NNAL" hr~ p~werh/l organospeeific car.cinoggns in :labgratoby anlfn~its;" their levels in tobacco smoke an"0 ~mokeless tobacco are comparable to bioassay doses which induce tumours. In laboratory animals, TSNA are metabolically activated to electrophiles which react with nucleophilic centres in DNA. Such DNA adducts can cause miscoding of the DNA which, in turn, causes activation of specific oncogenes. Comparable biochemical processes have been observed in tissue explants from tobacco smokers.

288 Stephen S Hecht and Dietrich Hoffmann • Tabl.e 4. The role of TSNA in tobacco-induced/~ancer in humans • ~Ca.ncersite .. Cause orassociation. 6raicavity• •Snuff dipping'and • • betel qtiid chewing .Evidenke for TSNA as causative factors" "" "~;tr~n'ffl NNK and ~lN~~ire.th~/~niy'ibb~cc~ ' constituents known to cause ora~ tumours'in lab3rat0ry hhimals. Human exposure levels at6 comlSarable levels of NNK and NNN that cause tumours in animals Cigarette smoking Lung Highly suggestive• NNK is a powerful lung carcinogen in all species tested. Human exposure levels are ..., ...:. .cor~parg~ble.to levels that cause.tumours in laboratdry '~ " "animals,'Pafallel activatio'n.mechariisi-as.of NNKin .. humans and laboratory animals Oesophagus Cigarette smoking Suggestive. Among the constituents of tobacco smoke that cause oesophageal tumours in rats, NNN occurs at the highest levels Pancreas Cigarette smoking Limited but suggestive. NNK and NNAL are the only tobacco constituents known to induce tumours of the exocrine pancreas in laboratory animals Nasal cavity Cigarette smoking Tentative. Oral administration of NNN to rats causes and snuff dipping squamous cell carcinoma of the nasal cavity similar to that seen in humans In future studies, emphasis should be placed on delineating the endogenous formation of TSNA in chewers and smokers and developing biological markers for the dosimetry of TSNA exposure of humans. Although the only safe way to prevent the occurrence of tobacco-related cancers is cessation of tobacco use, chemopreventive measures are feasible to lower the risk for those who do not cease the tobacco habit. Acknowledgements We thank Ilse Hoffmann and Bertha Stadler for their editorial assistance. Our studies in tobacco carcinogenesis are supported by Grants No CA- 29580, CA-32391, CA-44161 and CA-44377 from the US National Cancer Institute. References. Adams JD, Le~ S J, Vinchkoski N, Castonguay A and Hoffmann D (1983) On the formafiog 6I the ~ob.a_c.co-.specifi.c e~rci.nogen 4-(methylnitrp.sa~ino)-l-(3-pyridyl)-l- • b'utanone..d.uring ~mo!~ing. Cafi'.cer.L~tt~rs l-~ 339-346- " Y :' ." i "' ,"" ? " " ." . ': " Adams JD, Lee SJ and Hoffmann D (~984) Carcinogenic agents in cigarette srfioke and the influence of nitrate on their formation. Carcihog'enesis 5 221.-223" Adams JD, O'Mara-Adams KJ and Hoffmann D (1987) Toxic and carcinogenic agents in undiluted mainstream smoke and sidestream smoke of different types of cigarettes. Carcinogenesis 8 729-731 American Cancer Society (1988) Cancer statistics, 1988. Ca-A Cancer Journal for Clinicians 38 5-27" Andersen N'-Acyl doting g "~.- ~a~ch H • C. arcino; Belinsky S K-ras p nitrosam Proceed Bhide SV, .... j.rt t.h.e.s.~ 24 299-2 Boutet M. tobacco- the resF 438-442 Brinton L Fraume~ paranas: Brittebo li specific ~ Brunnema tobacco cigarettt 3rd Wo, Health, 76-1221 Bmnnema formatk smoke., Brunnema tobacco 1178--11 Brunnem~ the salb Letters 2 Carmella : of F34~ pyridyl) Castongu~ Klaunig culturec (3-p..yrid • .¢. a~to.~ specific Academ Castongu~ B and t and cart 2290 Crofton J

)ratory ' tO nimals nogen ato.ry i in ,, ~moke :curs at e only )f the aU$~S i lar to e only ion of sk for ) CA- ~ancer ~n the idyl)-l- smoke agents ,pes of 'hal for Tobacco-specific nitrosamines 289 Andersen RA, Flemming PD, Burton HR, Hamilton-Kemp TR and Sutton TB (.1989) N'-Acyl arid N'-nitroso pyridine alkaloids in alkaloid lines of Burley tobacco .. during growth, and a!.r?c.ufing:.Jburnal ofAgric.ulture an.d t7ood Chemistry 37. ~ .44--50 Bartscfi H. a/ad Montesafio R" (J:984~ .Relevance of nitros~ffnines" to hurhan .d~incer. Carcinogenesis'6"1381-1393 . Belinsky SA, Devereux TR, Stoner GD and A~derson MW (i988) Activation of the K-ras proto oncogene in lung tumors from mice treated with 4-(N-methyl-N- nitrosamino),l-(3-pyridyl)-l-butanone (NNK) and nitrosodimethylamine (NDMA). Proceedings of the American Association for Cancer Research 29 139 Bhide SV, Nair N J, Noir J, Spiegelhalder B and Preussmann R (1986) N-Nitrosamines • in thesali;ca of tobacco .ctiek0ers iirid iZnash~ri :tlsers. "Food. dn~l ~hemikizl Tb'xicolog~ "" 24 299-297 Boutet M, Mbntminy L and Castonguay'A (1987) Cellular changes induced by the toba.cco-specific carcinogen 4-(N-nitrosomethylamino)-l-(3-pyridyl)-l-butanone in the respiratory tract of Syrian golden hamsters. IARC Scientific Publications 84 438-442 Brinton LA, Blot WJ, Becket JA, Winn DM, Browder JP, Farmer JC Jr and Fraumeni JF Jr (1984) A case-control study of cancers of the nasal cavity and paranasal sinuses. American Journal of Epidemiology 119 896-906 Brittebo EB, Castonguay A, Furuya K and Hecht SS (1983) Metabolism of tobacco- specific nitrosamines by cultured rat nasal mucosa. Cancer Research 43 4343-4348 Brunnemann KD, Hoffmann D; Wynder EL and Gori GB (1976) Chemical studies on tobacco smoke XXXVII. Determination of tar, nicotine and carbon monoxide in cigarette smoke. A comparison of international smoking conditions. Proceedings 3rd World Conference on Smoking and Health, pp 441-449. US Department of Health, Education and Welfare; Washington DC (DHEW Publication No. (NIH) 76-1221). Brunnemann KD, Masaryk J and Hoffmann D (1983) Role of tobacco stems in the formation of N-nitrosamines in tobacco and cigarette mainstream and sidestream smoke. Journal of Agriculture and Food Chemistry 31 1221-1224 Brunnemann KD, Genoble L and Hoffmann D (1985) N-Nitrosamines in chewing tobacco: an international comparison. Journal of Agriculture and Food Chemistry 33 1178-1181 Brunnemann KD, Hornby AP and Stich HF (1987) Tobacco-specific nitrosamines in the saliva of Inuit snuff dippers in the Northwest Territories of Canada. Cancer Letters 37 %16 Carmella SG and Hecht SS (1987) Formation of hemoglobin adducts upon treatment of F344 rats with the tobacco-specific nitrosamines 4-(methylnitrosamino)-l-(3- pyridyl)-l-butanone and N'-nitrosonornicotine. Cancer Research 47 2626-2630 Castonguay A, Lin D, Stoner GD, Radok P, Furuya K, Hecht SS, Schut HAJ and Klaunig JE (1983a) Comparative carcinogenicity in A/J mice and metabolism by cultured mouse periphbral 1.ung of N'-nitrosonornicotine, 4-(methylnitrosamino)-l- (3-pyridyl)-I-butan.one,. and their an.alo.gue.s...Cancer ~Rese.qrch 43 1223-1229 " • - "Castonguay:A, Stoner GD, S~hut'HA~l and Hecht SS (1983b). Metabolism of tbbacco- " specific N-nitrosamin~s by cultured human tissues. Proceedings of the National Academy of Sciences of the USA 80 6694-6697 Castonguay A, Rivenson A, Trushin N, Reinhardt J, Stathopoulos S, Weiss CJ, Reiss B and Hecht SS (1984) Effects of chronic ethanol consumption on the metabolism and carcinogenicity of N'-nitrosonornicotine in F344 rats. Cancer Research 44 2285- 2290 Crofton J (1987) Smoking and health in China. Lancet ii 53

290 Stephen S Hecht and Dietrich Hoffmann Djordjevic MV, Gary SL, Bush LP and Chaplin JF (1989a) Tobacco-specific nitrosamine accumulation and distribution in flue-cured tobacco alkaloid isolines, Journal of Agriculture and Food Chemistry (in press) Djo.~djevic .MV, Bru.nnemann. KD'.a.nd Hoffmann D (1989b) Identifi.ca.tibn and analysis" o£a nicotine-defi.wd N-'n.itr.o.~amino.a..6id. and..other nitrosa.m, ilao, acids in. Doll R and Hill AB (1952))k stt/dy of the aetiolog3~ of carcinoma of the lung~ British" Medical Journal 2 1271-1286 Doll R and Peto R (1981) The causes of cancer: quantitative estimates of avoidable risk of cancer in the United States today. Journal of the National Cancer Institute 66 1191-1308 Fischer S,. Spiegelhalder B al).d Pr.e.us~mann R (1989a) Tobacco-specifienitrosamines m mainstream smoke o4 West German cigarettes "'t~ 'alon.e is not g'sufficlent index for the carcinogenic potential of cigarette smoke'. Carcinogenesis 10 169-173 Fischer S, Spiegelhalder B and Preussmann R (1989b) Influence of smoking parameters on the delivery of tobacco-specific nitrosamines in cigarette smoke. A contribution to relative risk evaluation. Carcinogenesis (in press) Foilles PG, Miglietta LM, Akerkar SA, Everson RB and Hecht SS (1988) Detection of O6-methyldeoxyguanosine in human placental DNA. Cancer Research 48 4148- 4188 Goodman MT, Morgenstern H and Wynder EL (1986) A case-control study of factors affecting the development of renal cancer. American Journal of Epidemiology 124 926-941 Hammond EC and Horn D (1958) Smoking and death rates--report on forty-four months of follow-up of 187,783 men. II. Death rates by cause. Journal of the American Medical Association l~i~i 1294-1308 Hecht SS and Hoffmann D (1988) Tobacco-specific nitrosamines( an important group of carcinogens in tobacco and tobacco smoke. Carcinogenesis 9 875-884 Hecht SS, Chert CB, Ohmori T and Hoffmann D (1980) Comparative carcinogenicity in F344 rats of the tobacco-specific nitrosamines, N'-nitrosonornicotine and 4- (methyl-N-nitrosarnino)-l-(3-pyridyl)-l-butanone. Cancer Research 40 298--302 Hecht SS, Reiss B, Lin D and Williams GM (1982) Metabolism of N'-nitrosonornicotine by cultured rat esophagus. Carcinogenesis 3 453-456 Hecht SS, Adams JD, Numoto S and Hoffmann D (1983a) Induction of respiratory tumors in Syrian golden hamsters by a single dose of 4-(methylnitrosamino)-l-(3- pyridyl)-l-butanone (NNK) and the effect of smoke inhalation. Carcinogenesis 4 1287-1290 Hecht SS, Young R and Maeura Y (1983b) Comparative carcinogenicity in F344 rats and Syrian golden hamsters of N'-nitrosonornicotine and N'-nitrosonornicotine-1- N-oxide. Cancer Letters 20 333--340 Hecht SS, Rivenson A, Braley J, DiBello J, Adams JD and Hoffmann D (1986a) Induction of oral cavity tumors in F344 rats by tobacco-specific nitrosamines and snuff. Cancer Research 4a 4162-4166 . He.cht .SS,. Trus.hin /q, Castonguay .A and Rivenson'A (1986b) Comparative "'timiorig.em.'city and .DNA ;m. eth, ylation: ".m.. F~.44 rats by 4-(methylnitrosamino.)-l-(3- pyridyl)-l-butan.one and N'.-nitros6dimethylatnine. Can~ek Res~a~,ch 46 498:-~2. ." Hecht SS, Abbaspour A and. Hoffmahn D (1988a) A study 6f tobacco carcinogenesis XLII. Bioassay in A/J mice of some structural analogues of tobacco-specific nitrosamines. Cancer Letters 42 141-145 Hecht SS, Carmella SG, Trushin N, Spratt TIE, Foiles PG and Hoffmann D (1988b)

able ~e 66" tines :lent 73 king e.A :tion 148- ctors ,124 -four ficity ~d 4- :otine atory -1-(3- esis 4 rats ne-1- 986a) and rative -1-(3- ~nesis ~ecific 9SSb) Tobacco-specific nitrosamines 291 Approaches to the development of assays ,for interaction of tobacco-specific nitrosamines with hemoglobin and DNA. IAR C Scientific Publications 89 121-128 Heuch I, Kvale G, Jacobsen.BK and Bjelke E (19.83) Use of.alcohol, tobacco and coffee and risk ~f pancrefis cancer..British Jo.urnal of Cancer 48 637--643 Higginson J and Muir CS (1979) Environmental carcinogenesis: misconception~ and limitations to cancer control. Journal of the National Cancer Institute 63 365-384 Hirsch JM and .Johansson SL (1983) Effect of long-term application of snuff on oral mucosa: an experimental studY in the rat. Journal of Oral Pathology 12 187-198 Hoffmann D and Adams JD (1981) Carcinogenic tobacco-specific N-nitrosamines in snuff and saliva of snuff-dippers. Cancer Research 41 4305-4308 Hoffmann D and Brunnemann KD (1983) Endogenous forrn.ation of N-nitros.o.proline in cigarette smokers. Cancer Research 43 55705574 Hoffmann D and Hecht SS (1985) Nicotipe-derived N-nitrosamines and tobacco related cancer: current status and future directions. Cancer Research 45 935-944 Hoffmann D and Hecht SS (1988) Smokeless tobacco and cancer. ISIAtlas of Science- Pharmacology 2 46-52 Hoffmann D and Hecht SS (1989) Advances in tobacco carcinogenesis. In: P Grover (ed)., Handbook of Experimental Pharmacology--Chemical Carcinogenesis and Mutagenesis. Springer, London (in press) Hoffmann D and Rathkamp G (1968) Chemical studies on tobacco smoke. III. Primary and secondary nitroalkanes in cigarette smoke. Be#rage zur Tabakfor- schung 4 124-134 Hoffmann D, Raineri R, Hecht SS, Maronpot RR and Wynder EL (1975) Effects of N'-nitrosonornicotine and N'-nitrosoanabasine in rats. Journal of the National Cancer Institute 55 977-981 Hoffmann D, Adams JD, Piade JJ and Hecht SS (1980) Chemical studies on tobacco smoke LXVII. Analysis of volatile and tobacco-specific nitrosamines in tobacco products. IARC Scientific Publications 3 507.516 Hoffmann D, Brunnemann KD, Adams JD and Hecht SS (1984a) Formation and analysis of N-nitrosamines in tobacco products and their endogenous formation in consumers. IARC Scientific Publications 57 743-762 Hoffmann D, Rivenson A, Amin S and Hecht SS (1984b)Dose-response study of the carcinogenicity of tobacco-specific nitrosamines in F344 rats. Journal of Cancer Research and Clinical Oncology 108 81-86 Hoffmann D, Adams JD, Lisk D, Fisenne I and Brunnemann KD (1987) Toxic and carcinogenic agents in dry and moist snuff. Journal of the National Cancer Institute 79 1281-1286 Hoffmann D, Brunnemann KD and Venitt S (1988) Carcinogenic nitrosamines in oral snuff. Lancet i 1232 International Agency for Research on Cancer (1985) Tobacco habits other than smoking: betel quid and areca nut chewing; and some related nitrosamines. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Vol. 37 International Ag.eney for Research. on Cancer (1986) Tobacco. smoking.. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemical~ to Humans, Vol. 38 Joly OG, Lubin JH and Caraballoso M (1983) Dark tobacco and lung cancer in Cuba. Journal of the National Cancer Institute 70 1033-1039 Jussawalla DJ and Deshpande VA (1971) Evaluation of cancer risk in tobacco chewers and smokers: an epidemiologic assessment. Cancer 28 244-252

292 Stephen S Hecht and Dietrich Hoffmann I~abat GC, Die~k GS and Wynder EL (1986) Bladder cancer in nonsmokers. Cancer Kinlen I2r:ahd: Rogbt "E (t9~) L+t/kaemla a£d smoking hfibits" among .Onited "Statfis- veterans• British Medical Journal 297 657-659 .' KIus H, Begutter H and In~orp M (t987) Environmental tobacco smoke in real life situations• Abstracts of the 4th International Conference on Indoor Air Quality and Climate, p 14. Berlin, FRG Konstantinidus A, Smulow JB and Sonnenschein C (1982) Tumorigenesis at a • predetermined oral site after one intraperitonea.1 injection of N-nitroso-N- • .tn~th~lurea: Science 216"1235-1"237 " .. - .'- • ' ' . • . .. ... Ladd KF, Newmark HL and Archer MC (1984) N-Nitrosation o~ pr01ine in smokers and nonsmokers: Jou.rnal of the National Cancer Institute 73 83-87 LaVoie EJ, Prokopczyk G, Rigotty J, Czech A, .Rivenson A and Adams JD (1987) Tumorigenic activity of the tobacco-specific nitrosamines 4-(methylnitrosamino)-l- (3-pyridyl)-l-but anone (NNK), 4-(methylnitrosamino)-4-(3-pyridyl)-l-butanol (iso-NNAL) and N'-nitrosonornicotine (NNN) on topical application to Sencar mice. Cancer Letters 37 277-283 Loechler EL, Green CL and Essigmann JM (1984) In vivo mutagenesis by 06- methylguanine built into a unique site in a viral genome. Proceedings of the National Academy of Sciences of the USA 81 6271-6275 Mohr U and Reznick G (1978) Tobacco carcinogenesis. In: CC Harris (ed), Pathogenesis and Therapy of Lung Cancer, pp 263-367. Marcel Dekker, New York Morse MA, Hecht SS and Chung F-L (1989a) Dietary inhibitors of chemical carcinogenesis, 5. Effects of aromatic isothiocyanates on tumorigenicity, 06- methylguanine formation, and metabolism of the tobacco-specific nitrosamine 4- (methylnitrosamino)-l-(3-pyridyl)-l-butanone in A/J mouse lung. Cancer Research 49 2894-2897 Morse MA, Wang C-X, Stoner GD, Mandal S, Conran PB, Amin SG, Hecht SS and Chung F-L (1989b) Dietary inhibitors of chemical carcinogenesis, 4. Inhibition of 4- (methylnitrosamino)-l-(3-pyridyl)-l-butanone-induced DNA adduct formation and tumorigenicity in lung of F344 rats by dietary phenethyl isothiocyanate. Cancer Research. 49 549-553 Nahmias A and Roizman B (1972) Infection with herpes simplex viruses 1 and 2-3. New England Journal of Medicine 289 781-789 Nair J, Ohshima H, Friesen M, Croisy A, Bhide SV and Bartsch H (1985) Tobacco- specific and betel nut-specific N-nitroso compounds: occurrrence in saliva and urine of betel quid chewers and formation in vitro by nitrosation of betel quid. Carcinogenesis 6 295-303 Nair J, Nair UJ, Ohshima H, Bhide SV and Bartsch H (1987) Endogenous nitrosation in the oral cavity of chewers while chewing betel quid with .or without tobacco. IAR C Scientific Publications 84 465-469 Oesterd.ahl B;G and Slorac.h S. (1988) To.bacco-specific N-nitrosa'm~nes in'the saliva of ' habitual ~nuff diigpers: FoOdA. d~it~'ves'and Contaminants 5 58]~'-586 ' '." ~ : ""-' Ohshima H; Nair J, Bourgade MC, "Ffiegen M, Garren .L add Bartsch I-I (1985) Identification and occurrence of two new N-nitrosamino acids in tobacco px:oducts: 3-(N-nitroso-N-methylamino)propionic acid and (4-(N-nitroso-N-methylamino)- butyric acid. Cancer Letters 26 153-162 Palladino G, Adams JD, Brunnemann KD, Haley NJ and Hoffmann D (1986) Snuff- dipping in college students: a clinical profile. Military Medicine 151 342-346 Park NH, Sapp JP and I-Ierbosa EG (1986) Oral cancer induced in hamsters with herpes simplex in~ • Oral Pathology 62 : Parsa.I, Fo~,e CA, ( " " " " Biological EffeCt: pp 233-244. Cold Peto R (1986) Tobat Preussmann R and American Chemic. Prokopczyk B, Rive •... .: (Methylnit.rosami~. ... . carcinogenicity, a~ Rivenson A, Fioffm lung and exocrint derived N-nitrosa~ Roberts DL (1988) 49-81 Rodenhuis S, Sleto: Bodegam PCI-I an ras oncogene actix 5738-5741 Royal College of Livingstone, Lond Sipahimalani AT, C? N-nitrosamines in Toxicology 22 261- Tominaga S (1986) . Publications 74 12." Tricker AR and Pr~ zarda tobacco. Cat Tricker AR, Haubne tobacco-specific ni ,' under simulated g~ ! Tso TC (1972) Physi i and Ross, Stroudst ~i US Department of t Smoking: Cancer, 82-50179 US Department of t Using Smokeless T. US Department of I- i Smoking, Nicotine No. (.CDC) 88-840 "• ~" US Department of • Report.on Nutritio~ ton DC US Department of Consequences of S : DHSS Publication ' US National Resear, Nitroso Compoun6

tnd f4- md cgr :-3. :co- ine rid. ion :CO. ~ of ,85) CtS: uff- vith Tobacco-specific nitrosamines 293 herpes simplex infection and simulated snuff-dipping. Oral Surgery, Oral Medicine, Oral Pathology 62 164-168 Parsa I, Foye CA, Cleary CM and Hoffmann D (1986).Differences in Metabolism and Biological Effects of NN.K.i.n. Human Target Cells. B~n.b.ury Report No.23,.. pp 233-244..Cold Spring Harbor N'2" .. Peto R. (1986) Tobacco. UK and China. Lancet ii 1038 Preussmann R and Eisenbrand G (1984) N-Nitroso carcinogens in the environment. American Chemical Society Monographs 182 829-867 Prokopczyk B, Rivenson A, Bertinato P, Brunnemann KD and Hoffmann D (1987) 3- (Methylnitro.samino)lSropionitrile: occurrence in saliva of betel quid chewers, carcinogenicity,'and DNA methylation in F344 rats. Can.cer Research 47 467-471 Rivenson A, Hoffmann D, Prok0pczyk B, Amin S and Hecht $8 (1988) Iaada~ction of lung and exocrin~ pancreas tumors in F344 rats by tobacco-specific and Areca- derived N-nitrosamines. Cancer Research 48 6012-6917 Roberts DL (1988) Natural tobacco flavor. Recent'Advances in Tobacco Science 14 49-81 Rodenhuis S, Sletos RJC, Boot AJM, Evers SG, Moor WJ, Wagenaar SSC, Van Bodegam PCH and Bos JL (1988) Incidence and possible clinical significance of K- ras oncogene activation in adenocarcinoma of the human lung. Cancer Research 48 5738-5741 Royal College of Physicians of London (1983) Health or Smoking. Churchill Livingstone, London Sipahimalani AT, Chadha MS, Bhide SV, Bratap AI and Nair Y (1984) Detection of N-nitrosamines in the saliva of habitual chewers of tobacco. Food and Chemical Toxicology 22 261-264 Tominaga S (1986) Spread of smoking to the developing countries. IARC Scientific Publications 74 125-133 Tricker AR and Preussmann R (1988) The occurrence of N-nitroso compounds in zarda tobacco. Cancer Letters 42 113-118 Tricker AR, Haubner R, Spiegelhalder B and Preussmann R (1988) The occurrence of tobacco-specific nitrosamines in oral tobacco products and their potential formation under simulated gastric conditions. Food and Chemical Toxicology 26 861-865 Tso TC (1972) Physiology and Biochemistry of Tobacco Plants. Dowden, Hutchinson and Ross, Stroudsburg PA US Department of Health and Human Services (1982) The Health Consequences of Smoking: Cancer, a Report of the Surgeon General. DHSS Publication No. (PHS) 82-50179 US Department of Health and Human Services (1986) The Health Consequences of Using Smokeless Tobacco. NIH Publication No. 86-2874, Bethesda, MD US Department of Health and Human Services (1988a) The Health Consequences of Smoking, Nicotine Addiction, a Report o'f the. Surgeon General. DHSS Publication No. (CDC) 88-8406 US Department of Health. and Human Services (1988b) The Surgeon Genera'l.'s Report on Nutrition and.Hum.an. Health: US Government Printing Offi..c.e; W.ashing- tonDC" -_ . -" -'.• US Department of Healtli ~nd Human Services (1989) Reducing the" Health Consequences of Smoking. 25 Years of Progress. A Report of the Surgeon General. DHSS Publication No. (CDC) 89-8411 US National Research Council (1981) The Health Effects of Nitrate, Nitrite and N- Nitroso Compounds. National Academy Press, Washington DC

294 Stephen S Hecht and Dietrich Hoffmann US National Research Council (1986) Environmental Tobacco Smoke--Measuring • . Exposures and Assessing Health Effects. National. Academy Press, Washington DC Wenke G, Brunnemann KD and Hoffmann D .(1984) A study of. betel quid carcinogenesis. IV. Analysis of the saliva of chewers. A preliminary i-epo.r~. Journal of Cancer Research and Clinical Oncology 198 ~[ 10-113 Winn DM, Blot WJ, Shy CM, Pickle LW, Toledo A and Fraumeni JR Jr (1981) Snuff- dipping and oral cancer among women in the southern United States. New England Journal of Medicine 394 745-749 Wogan" GN (1988) Detection of DNA damage in studies on" cancer etiology and • prevention. IARC Scientific Publications 89 32-51 • World Health Organization (1988) Smokeless tobacco control. Report of a WHO study group..WHO Technical Report Series 773 Wynder EL and G6ri GB (1977) Contribution of the environment to cancer incidence: an epidemiological exercise. Journal of the National Cancer Institute 58 825-832 Wynder EL and Graham EA (1950) Tobacco smoking as a possible etiologic factor in bronchiogenic carcinoma. A study of six hundred and eighty-four proved cases. Journal of the American Medical Association 143 329-336 Wynder EL and Hoffmann D (1967) Tobacco and Tobacco Smoke. Studies in Experimental Carcinogenesis. Academic Press, New York Wynder EL and Stellman SD (1979) The impact of long-term filter cigarette usage on lung and larynx cancer risk: a case-control study. Journal of the National Cancer Institute 62 471-477 (The authors are responsible for the accuracy of the references.) Preforme¢ beverage,. JOSEPH H H Institute of Fo~ 14853 I Introducti II Nitrosatio III Precursor IV Occurren~ V Evidence t VI Reduction VII Estimates VIII Regulator~ IX Conclusio~ Keywords: N-n: cured meats, Summary Most western-s nitrosamines. 1: kg) amounts o: dietary volatile bacon, and bee declined in rec nitrosamines hl exposure for co person. Asian • '. preliminary dat nitrosamine cot Indirect eviden~ one to three ore Imperial Cancer ~ ,

2063628232

icotine, r 1986, ~, W1N N, and , .Guy's NI C O TINE, S MO KIN G, AND THE LOW TAR PROGRAMME EDITED BY NICHOLAS WALD and SIR PETER FROGGATT st Bartholomew's Hospital Medical College OXFORD NEW YORK TOKYO OXFORD UNIVERSITY PRESS 1989

6 Trends in mortality from smoking-related diseases in England and Wales SARAH C. DARBY, RICHARD DOLL, and IRENE M. STRATTON Abstract Data are presented on the recent trends in mortality in England and Wales from lung cancer, ischaemie heart disease, aortic aneurysm, and chronic obstructive lung disease, considered to "be the main smoking- related diseases. For the most part the trends can only be explained by changes in the real incidence of the diseases'. Furthermore, any changes • in the level of disease attributable to manufactured cigarettes must be due to changes in the constituents of cigarettes rather than the number smoked. However, the possibility of distinguishing the effects of altera- tions in the level of the main cigarette components, tar, nicotine, and carbon monoxide, purely from national trend data seems remote. 6.1. Introduction The objective of this paper is to present data on the recent trends in mortality from the main smoking-related diseases in England and Wales, and to discuss briefly whether or not the observed trends represent real changes in the level of disease, or whether they are due to changes in the efficacy of treatment, the accuracy of diagnosis, or the terminology used to describe and classify the diseases. A fuller report will be published else- where,1 but this paper should provide a basis for examining the extent to which the real trends in these diseases are due to variation in the amounts and types of tobacco products smoked, or to other agents. 6.2. Smoking related diseases The recent International Agency for Research on Cancer (IARC) mono- graph2 evaluating the carcinogenic risk of tobacco smoking to humans has grouped the various causes of death into five categories according to the way in. which they are related to smoking, as shown in Table 6.1. Category A includes six diseases where the evidence suggests that practically the whole of the difference in mortality between smokers and life-long non- Table 6.1. Im: different ways Category* t~ g ( C ( D E *A, diseases f, for which excess excess mortality which excess mo

ted ,NE nd and m, and ~oking- !ned by 'hanges ~ust be ~umber ' altera- ~e, and 9te. trends in ad Wales, esent real ges in the gy used to shed else- extent to e amounts ,C) mono- amans has ing to the Category tically the • long non- ! Smoking-related diseases in England and Wales Table 6.1. Importance of different causes of death related to different ways in England and Wales1 71 smoking in Category* Cause of death No. of deaths as % of total deaths in England & Wales, 1984 A Cancer of lung 6.3 Ischaemic heart disease 27.8 Respiratory heart disease 0.4 Aortic aneurysm 1.3 Peripheral vascular disease 0.3 Chronic obstructive lung disease 4.3 Subtotal 40.4 B Alcoholism < 0,1 Cirrhosis of liver 0.4 Poisoning 0.2 Suicide 0.8 Subtotal 1.4 C Cancer of oesophagus 0.8 Cancer of lip, tongue, mouth, 0.4 pharynx, larynx Cancer of stomach 1.8 Cancer of liver 0.2 Cancer of bladder 0.8 Cancer of kidney 0.4 Cancer of pancreas 1.1 Cancer of cervix uteri 0.3 Cancer of unspecified site 1.6 Respiratory tuberculosis < 0.1 Pneumonia 4.4 Other respiratory disease 1.4 Myocardial degeneration 0.9 Hypertension 0.8 Arteriosclerosis 1.2 Cerebral thrombosis 2.5 Other cerebrovascular disease 10.1 Peptic ulcer 0.8 Hernia 0.2 Osteoporosis 0.2 Subtotal 29.9 D Cancer of endometrium 0.2 Parkinsonism 0.6 Ulcerative colitis < 0.1 Toxaemia of pregnancy < 0.1 Subtotal 0.8 E All others 27.5 *A, diseases for which excess mortality in smokers is attributable to smoking; B, diseases for which excess mortality in smokers is attributable to confounding; C, diseases for which excess mortality in smokers may be partly or wholly attributable to smoking; D, diseases for which excess mortality in non-smokers may be preventable by smoking; E, other diseases.

72 Sarah C. Darby, Richard Doll, and Irene M. Stratton smokers is due to tobacco. Four of these: cancer of the lung, ischaemic heart disease, aortic aneurysm, and chronic obstructive lung disease, have been selected for the present study. Between them these four accounted for about 40 per cent of the total deaths registered in England and Wales in 1984. Respiratory heart disease has been omitted on account of the difficulty of obtaining reliable statistics relating to it, and peripheral vascular disease has been omitted because of its low case-fatality rate, rendering it unsuitable for study using death certificate data. Category C in the IARC monograph includes those diseases that have been positively associated with smoking, and for which the excess mortality in smokers may be partly or wholly attributable to smoking. Some of these are cancers of specific sites, and the available human data are reviewed in the mono- graph. For eight sites, namely oesophagus, lip, oral cavity (excluding the salivary gland), pharynx (excluding the nasopharynx), larynx, bladder, kidney, and pancreas, the monograph concludes that there is sufficient evidence to establish tobacco smoking as an iml~ortant cause*. Between them these eight sites of cancer accounted for just over 3 per cent of the total deaths registered in England and Wales in 1984. Data for them will be presented elsewhere.I Myocardial degeneration is included in IARC Category C, and we have included myocardial degeneration with reference to arteriosclerosis in our definition of ischaemic heart disease prior to 1968 because of the frequency with which it was used in the past to describe cases now attributed to the latter. The remaining non-neoplastic diseases in Category C have been omitted from the present study. For peptic ulcer this is because the mortality trends are complex and affected by many factors, including therapy, and for other diseases it is because the relative risks of mortality in smokers compared with non-smokers have been low in most studies of the effects of tobacco smoking. The three remaining disease categories defined by the IARC are: category B, diseases for which excess mortality in smokers is attributable to confounding; category D, diseases for which there is an excess mortality in non-smokers which may be preventable by smoking; and category E, diseases that are generally unrelated to smoking. All these have been excluded, except in so far as the trends in them are relevant to the understanding of the trends in the diseases related to smoking. 6.3. Data on smoking-related diseases Annual death rates for the selected diseases have been calculated in 5-year age-groups for males and females between 1950 and 1984 inclusive and will *In the case of renal cancer, the evidence is sufficient only in relation to the renal pelvis. Adenocarcinoma of the kidney arising from the cortex was classed as 'perhaps' caused by smoking. Smo be published in f sidered because, uncertain in earli, International Cla: the early 1950s d reliable for most 85+ age group tl population estima population within give rise to misle~ Trends in the ." plotted in Figs 6.1 Multiple of 1950 rate 8.00] (a) 4.001 °%0 2.001 oO° 1.00 - "-'=" 0.50- 0.25 1950 ' 19~0 ' 1 Ye Fig. 6.1. Trends England and 60-64 (A), 70 Males Females

aaemic .', have aunted 'ales in of the ipheral y rate, ry C in ~itively nokers :ancers mono- ing the adder, fficient ~tween of the will be IARC erence .o 1968 ~scribe ases in n most ~lisease excess iseases aay be nerally "as the in the 5-year will pelvis. ~used by Smoking-related diseases in England and Wales 73 be published in full elsewhere.1 Data prior to 1950 have not been con- sidered because, for many diseases, the diagnosis becomes increasingly uncertain in earlier time periods, and the difficulties in bridging between International Classification of Disease (ICD) codes also increase. Since the early 1950s diagnostic accuracy is thought to have been reasonably reliable for most important diseases, except in the very elderly• For the 85+ age group there is the additional problem that only an aggregated population estimate is published, and changes in the age-distribution of the population within the age group could create distortions in the rates and give rise to misleading trends. Trends in the annual death rates for each of the selected diseases are plotted in Figs 6.1-6.4 for age groups 30-34, 40-44, 50-54, 60-64, 70-74, Multiple of 195~ rate 8.00- (a) ...... oOOO°° 4.00- o o°°°° 2.00- .°~°~"" ~,,~.,~,.~ .... , ..... 13 r700 0.50- n ooo o° 0o0 o 0.25- o 1950 ' 19~0 ' 19'70 ' 10'80 ' Year Multiple of 1950 rate 8.00 ] (b) / 4.00-t ". "........... 0.50 [3 c~ 0 25 o 1950 19~)0 19'70 19'80 Year Fig. 6.1. Trends in lung cancer mortality rates in (a) males and (b) females in England and Wales, 1950-1984, at ages 30-34 (t3), 40-44 (~), 50-54 (~), 60-64 (A), 70-74 (~), 80-84 (o). Annual data except as indicated below. Males Females Age group Years Plotted in combined 30-34 1950-56 1953 80-84 1950-52 1951 30-34 1950-71 1960 40-44 1950-54 1952 50-54 1950-52 1951 60-64 1950-51 1950 70-74 1950-51 1950 80-84 1950-53 1951

74 Multiple of 1950 rate 4.0 (a) Sarah C. Darby, Richard Doll, and Irene M. Stratton Multiple of 1950 rate 4.0- (b) Multiple of 1950 rate 16.0 - (a) Smoki, 2.0 1.0 t~~t~°°° .... 00% ........ 0.5' 1950 1955 1960 1965 1970 1975 1980 1985 Year $,$. • 2.0- , , ... •=. • ~z~z~^~t-~z~z, 0.5--- 1950 1955 1960 1965 1970 197519801985 Year Fig. 6.2. Trends in mortality rates for IHD in (a) males and (b) females in England and Wales, 1950-1984, at ages 30-34 ([]), 40-44, (@), 50-54 (m), 60-64 (/x), 70-74 (e), 80-84 (©). Annual data except as indicated below. Age group Years Plotted in combined Males 30-34 1950-54 1952 Females 30-34 1950-69 1959 40-44 1950-53 1951 1.0 0.5 1950 19~0 ~ Year Fig.6.3. Trends in ao in England and Wale~ 60-65 (/~), 70-74 ( Males and 80-84 years. The ICD codes used to define each disease are indicated in Table 6.2. In the present study interest centres on the trend within each age-group, rather than the relative positions of the different age groups, and so the age-specific rates have all been divided by the corresponding rate in 1950, thus enabling trends for a wide range of ages to be plotted on a single figure. In cases where the rate in 1950 was very low and, therefore, subject to variability the divisor has been calculated from the rate for a period spanning several years, starting in 1950, and continuing until the number of deaths was at least 400; that is, sufficient to reduce the standard error of the rate to less than about 5 per cent. The age groups for which this occurred are indicated in footnotes to each figure. For some diseases and sexes, the total numbers of deaths in the entire period was less than 400, and so these age groups have been omitted from the figures for the relevant diseases and sexes. Females 6.4. Factors that r Apart from real ch~ factors that may ini (1) changes in the e length of surviv (2) sudden breaks national Classifi

801985 gland (~), icated ~ each rOUES, ,nding :ed on efore, for a til the mdard ch this es and n 400, :levant Smoking-related diseases in England and Wales Multiple of Multiple of 1950 rate 1950 rate 16.0- (a) 16.0. (b) 8.0¸ 4.0, 2.0, 0.5. 1950 1960 19'70 ' 19'80 Year 8,0. 4.0- 2.0- 1.0- 1950 ooooOO°o°Oo°°°oO° %o.... • 08~Q~°%** . .* ** ***** * 19'60 1~70 1~80 ' Year 75 Fig. 6.3. Trends in aortic aneurysm mortality rates in (a) males and (b) females in England and Wales, 1950-1984, at ages 40-44 (¢,) (males only), 50-54 (It), 60-65 (A), 70-74 (e), 80-84 (©). Annual data except as indicated below. Males Females Age group Years Plotted in combined 40-44 1950-78 1964 50-54 1950-59 1954 60-64 1950-55 1952 70-74 1950-54 1952 80-84 1950-58 1954 50-54 1950-66 1958 60-64 1950-58 1954 70-74 1950-55 I952 80-84 1950-57 1953 6.4. Factors that may produce artificial trends in mortality rates Apart from real changes in the incidence of the disease in question, four factors that may influence trends in certified mortality rates are: (1) changes in the efficacy of the treatment available may alter the typical length of survival of individuals with the disease; (2) sudden breaks in the trends may occur when revisions of the Inter- national Classification of Diseases.(ICD) are introduced; o

76 Sarah C. Darby, Richard Doll, and Irene M. Stratton Multiple of 1950 rate 2.00 ] (a) o 1.oo f'- #-~-,, .,"- ," ..... 1 0.25|4 •• * ", •4, • 0.125-~ , 1950 19'60 19'70'19'80 Year Multiple of 1950 rate 2.00 ] (b) 1 1950 19'60 ' 1~7o 1~8o Year Fig. 6.4. Trends in mortality rates for chronic obstructive lung disease in (a) males and (b) females in England and Wales, 1950-1984, at ages 30-34 (n) (males only), 40-44 (.), 50-54 (m), 60-64 (zX), 70-74 (e), 80-84 (O). Annual data except as indicated below. Males Females Age group Years Plotted in combined 30-34 1950-78 1964 40-44 1950-52 1951 40-44 1950-57 1953 50-54 1950-51 1950 (3) changes in the use of medical terms may occur, so that conditions previously categorized under one heading tend to be classed under another; (4) changes in the accuracy of diagnosis may alter the numbers of people who, dying from the disease, are not recorded as such on the death certificate, or who, dying from some other cause, have their death recorded as due to the disease in question. The first of these factors reflects an important biological change, the re- maining three are human artefacts. In the following sections, the observed trends in the selected diseases will be described, and the role of these four factors will be discussed briefly in relation to them. Smoking Table 6.2. Internatio calculating the trends Cancer of lung Ischaemic heart disease Aortic aneurysm Chronic obstructive lung disease 6.5. Lung cancer From Fig. 6.1, it appe 34, 40-44, and 50-54 at older ages the deatl 44 years the death ra groups the declining t~ were approximately ¢ 1954. At older age I menced in successive declining in all age g death rate was still inc times the male rate ir The trends in death for age group 30-34 death rates were incre this was superceded ~ tinued progressively ~ returned to its 1950-1 rate did not start to d remained at about tw death rates have risen any sign of a decreas~ death rate has been sli female rates were abe Survival after diagn cent of patients rein changes in the efficacy on trends in mortalit) nomenclature over th

980 males only), :ept as itions ander eople death death ~e re- erved four Smoking-related diseases in England and Wales 77 Table 6.2. International Classification of Disease (ICD) codes used in calculating the trends in diseases presented in Figs 6.1-6.4 ICD codes 6th revision 7th revision 8th revision 9th revision 1950-57 1958-67 1968-78 1979-84 Cancer of lung 162-3 162-3 162 162 Ischaemic heart 420, 422.1 420, 422.1 410-414 410-414 disease Aortic aneurysm 451 451 441 441 Chronic obstructive 502, 527 502, 527 491-2, 519 491-2, 496, 519 lung disease 6.5. Lung cancer From Fig. 6.1, it appears that the death rates in men in the age groups 30- 34, 40-44, and 50-54 years were reasonably stable in the early 1950s while at older ages the death rates were increasing. In age groups 30-34 and 40- 44 years the death rate began to decline in the early 1960s. In both age groups the declining trends have continued, so that by 1980-1984 the rates were approximately one-third and one-half, respectively, those in 1950- 1954. At older age groups, successively smaller reductions have com- menced in successive calendar years, so that by 1984 the death rate was declining in all age groups other than age 80-84 years. At this age the death rate was still increasing so that in 1980-1984 the male rate was over 6 times the male rate in 1950-1954. The trends in death rates among women are similar to those in men only for age group 30-34 years. In women aged 40-44 and 50-54 years the death rates were increasing during the 1960s. In the 40-44 years age group this was superceded by a decline that started in about 1970 and has con- tinued progressively so that by 1980-1984 the female death rate had returned to its 1950-1954 value. In the 50-54 yeai's age group the death rate did not start to decline until the late 1970s and by 1980-1984 it still remained at about twice the 1950-1954 value. At older ages the female death rates have risen steadily since the early 1950s and do not yet show any sign of a decrease. In these older women the rate of increase in the death rate has been slightly less than that in old men, and by 1980-1984 the female rates were about 3.5 times their 1950-1954 values. Survival after diagnosis of lung cancer remains poor, with under 10 per cent of patients remaining ali~ve 5 years after diagnosis,3 so that any changes in the efficacy of the available treatment will have had little impact on trends in mortality. Similarly, there have been no changes in medical nomenclature over the last 35 years which would have affected thelung

78 Sarah C. Darby, Richard Doll, and Irene M. Stratton cancer mortality trends appreciably, and plots of annual lung cancer death rates for males and females in 5-year age groups during the period 1950- 1984 show that there is no age group for males or females which shows a sudden break at the time of introduction of a new ICD revision. The main improvements in recognition of lung cancer happened before the 1950s, and the general level of diagnosis is likely to have been good since then. There remain, however, some specific diseases which could theoretically be a source of diagnostic confusion with lung cancer. For example, .the rise in lung cancer death rates at older ages might be due, in part, to a reduction in the numbers of deaths attributed to cancer of unspecified site, pneumonia, or senility, and the reductions in lung cancer death rates in younger males might be partly due to a preferential diagnosis of cancer of the pleura (especially pleural mesothelioma) or cancer of the mediastinum. To examine these possibilities, tabulations of the mortality rates for these diseases have been constructed for males and females for the same time period and age groupings as for lung cancer itself, and the ,levels of and trends in these diseases compared with those for lung cancer. We .have concluded that there is no evident explanation for the recent decrease in lung cancer death rates in males aged 25-54 years other than a real decrease in the incidence of the disease. Nor is there any positive evidence that the continued increases in lung cancer death rates in males aged 80-84 years or females aged 60-84 years are due to anything other than a real increase in the incidence of the disease. However, it is possible that some of the increase in the 80-84 years age group has occurred as a result of a decreasing tendency to certify deaths as due to senility, especially among females. 6.6. Ischaemic heart disease From Fig. 6.2, male death rates for ischaemic heart disease (IHD) were increasing in the 1950s in all age groups. The rate of increase was greater in men aged under 55 than at older ages, so that by the early 1970s the rates in younger men were approximately double their values in the early 1950s. A turning point occurred, however, during the 1970s, so that by the early 1980s the rate in every age group below 80-84 years was declining. The decline started slightly earlier, and its rate has been somewhat greater, in men aged under 55 than in men aged 60-64 or 70-74 years. However, by 1984 the male death rates had not yet returned to their 1950 level in any age group and in men aged under 55 they were still increased by 50 per cent over their 1950 .values. The trends in female IHD death rates in those aged under 55 are similar to those for males. For women aged 60 and above the rate of increase Smokin during the 1950s and groups has the rate i~ than its 1950 value. diminished during th approximately stable ages 70-74 years an. started to decline an, Plots of annual IN during the period 19: females which show~ revision. To examin IHD and hypertensio the choice of conditi, heart disease and pi Fig. 6.2, rates for the those for IHD. Ther, ascribing death to p~ age-specific trends i explanation for the i and the 1970s in mr changes in the incide_ availability of corona treatment of hyperte may all have contribl age groups since the 6.7. Aortic aneurys Death rates from ao above age 60 (see Fig and by the early 1980 trends are broadly sin years there is some e Before the Secont aneurysm. Syphilitic distinct, and not like: syphilitic aortic aneui whereas more recenl more likely to be sp, deaths certified as dll coded in the sixth rev the introduction of ~o

.-'ore ood ,uld For ;, in ~ of leer osis the dity for the cer. :ent in a tive ales :her fere '~r in • ~s in ~.A arly ~he :, in • , by any :ent ~ilar ~ase Smoking-related diseases in England and Wales 79 during the 1950s and 1960s was less than in men; in none of these older age groups has the rate in any year been more than about 30 per cent greater than its 1950 value. In women aged 60-64 years the rate of increase diminished during the 1970s and in the early 1980s the mortality rate was approximately stable, and about 20 per cent higher than its 1950 value. At ages 70-74 years and 80-84 years the female death rates have recently started to decline and are now approximately equal to their 1950 values. Plots of annual IHD rates for males and females in 5-year age groups during the period 1950-1984 show that there is no age group for males or females which shows a sudden break at the introduction of a new ICD revision. To examine the possibility that diagnostic confusion between IHD and hypertension, or rheumatic and other heart diseases, or a shift in the choice of condition to which death is attributed in patients with both heart disease and pneumonia, has accounted for some of the trends in Fig. 6.2, rates for these other diseases were calculated and compared with those for IHD. There was no evidence to suggest that material changes in ascribing death to pneumonia or IHD has had a major influence on the age-specific trends in IHD rates. In addition, there was no evident explanation for the increase in IHD death rates between the early 1950s and the 1970s in males and females aged 30-54 years, other than real changes in the incidence of, or fatality from IHD. However, the greater availability of coronary care units, greater use of coronary bypass surgery, treatment of hypertension and, in the last few years, better drug therapy, may all have contributed to the decreasing death rates seen in the younger age groups since the 1970s in both males and females. 6.7. Aortic aneurysm Death rates from aortic aneurysm were increasing..in the 1950s in men above age 60 (see Fig. 6.3). Since 1975 the rate of increase has diminished, and by the early 1980s the death rates were relatively stable. In women the trends are broadly similar to those in men. However, in women aged 50-54 years there is some evidence of a decrease in the last few years. Before the Second World War, syphilis was a major cause of aortic aneurysm. Syphilitic and non-syphilitic aortic aneurysm are clinically distinct, and not likely to be confused diagnostically. However, formerly, syphilitic aortic aneurysm was often i-eferred to simply as aortic aneurysm, whereas more recently cases due to syphilis have become rare, and are more likely to be spe.cified as such. Following the trend in terminology, deaths certified as due to aortic aneurysm, not specified as syphilitic were coded in the sixth revision of the ICD as if they were due to syphilis. After the introduction of the seventh ICD revision in 1958 the coding rules

8O Sarah C. Darby, Richard Doll, and Irene M. Stratton specified that such deaths should be coded as if they were not due to syphilis. To overcome these changes, mortality rates for cardiovascular syphilis have been calculated for the period 1950-1984 and added to those for aortic aneurysm. In men aged 70-74 and 80-84 years, and in women aged 70-74 years the rates for this wider disease group had increased more than three-fold during this period, indicating that mortality rates due to non-syphilitic aortic aneurysm have increased by at least this amount. In women aged 70-74 years there was a two-fold increase, and smaller increases were observed for both sexes at age 60-64 years. Below age 60 it is hard to draw any firm conclusion as the numbers of deaths are so few. In recent years some cases of aortic aneurysm have been treated surgically before rupture, but they are unlikely to have been on a large enough scale to have had a material impact on the death rate. 6.8. Chronic obstructive lung disease In men, mortality rates from chronic obstructive lung disease (COLD) at ages 40-44 and 50-54 years were relatively stable in the early 1950% see Fig. 6.4. From about 1960 they started to decline. These declines have continued progressively and in 1984 the rates in these age groups were, respectively, 13 and 22 per cent of their 1950 values. In age groups 60-64 and 70-74 years male COLD mortality rates increased in the late 1950s so that in the early 1960s they were increased by 20 and 50 per cent, respectively, compared with their 1950 values. Since then the rates in these two age-groups have declined progressively, and in 1984 were 40 and 90 per cent of their 1950 values. At ages 80-84 years COLD mortality rates in men increased by nearly 40 per cent between 1950 and the late 1970s, and since then have remained relatively stable. In women the trends differ from those in men. At ages 70-74 and 80-84 years the rates have declined progressively since 1950, and in 1984 were approximately one-half their 1950 values. At ages 50-54 and 60-64 years the rates have remained approximately stable and in 1984 were just below their 1950 values. At ages 40-44 years the rate remained approximately equal to its value in the early 1950s until about 1970, but since then has declined steeply, and by the early 1980s was about one-third its value in the early 1950s. There are no sudden breaks in the annual COLD death rates on intro- duction of the new ICD revisions. However, there has been a shift from the terms chronic bronchitis and emphysema towards COLD during the period 1950-1984, and so our definition of COLD includes all three. Emphysema was classed with 'other respiratory disease' up to 1968, and COLD was classed with 'other respiratory disease' until 1978. Therefore 'other respir- atory disease' has also been included in the definition throughout the whole time period. Smokin Diseases that migh unspecified bronchiti when trends in morta COLD, no obvious e mortality rates were severity of the diseas, by stronger advice t otherwise have made 6.9. Conclusions There have been su from many smoking 1950-1984. For the n than changes in the r changes are due to i fuller discussion. Ho manufactured cigare during the period 19 changes in the level manufactured cigaret the cigarettes, rather in the tar and nicoti~ closely correlated,5"' carbon monoxide le,~ effects of alterations from a study of natic Acknowledgements The trends in mortal using computerized Office of Population References 1. Wald, N., Doll, R., Trends in smoking r in preparation. 2. IARC (1986). Tobo_ carcinogenic risk oi Research on Cance.

t due to ,vascular to those I "~¥om en ed more s due to ount. In smaller age 60 it so few. treated a large )LD) at 50S, see es have were, 60-64 .950s so cent rates in 0s, and er from eclined If their mained ~es. At ~ in the I by the ~ intro- om the period ~ysema D was respir- whole Smoking-related diseases in England and Wales 81 Diseases that might be confused diagnostically with COLD are acute and unspecified bronchitis, bronchiectasis, asthma, and pneumonia. However, when trends in mortality from these diseases were compared with those for COLD, no obvious explanation for the large changes in COLD age-specific mortality rates were found, other than real changes in the incidence and severity of the disease. A lower fatality rate could in part have been caused by stronger advice to stop smoking, although changes in treatment will otherwise have made little difference. 6.9. Conclusions There have been substantial changes in the age-specific mortality rates from many smoking related diseases in England and Wales in the period 1950-1984. For the most part, these cannot be explained in any way other than changes in the real incidence of the disease. The extent to which the changes are due to manufactured cigarettes or to other agents deserves fuller discussion. However, the average weekly consumption by men of manufactured cigarettes at each age group in Britain remained stable during the period 1948-1975,4 and this provides good evidence that any changes in the level of disease in men at least that are attributable to manufactured cigarettes have been caused by changes in the constituents of the cigarettes, rather than the number smoked on average. As the changes in the tar and nicotine yields of manufactured cigarettes have been quite closely correlated,5'6 and also somewhat correlated with changes in carbon monoxide levels,5'6 it is likely to be impossible to distinguish the effects of alterations in the level of any one of these components purely from a study of national trends in smoking related diseases. Acknowledgements The trends in mortality rates presented in this paper have been Calculated using computerized historic mortality data files made available by the Office of Population Censuses and Surveys. References 1. Wald, N., Doll, R., Darby, S., Kiryluk, S., Peto, R., and Pike, M. (ed.) (1989) Trends in smoking related diseases in Britain. Oxford University Press, Oxford in preparation. 2. IARC (1986). Tobacco smoking. IARC monographs on the evaluation of the carcinogenic risk of chemicals to humans, Vol. 38. International Agency for Research on Cancer, Lyon.

82 Sarah C. Darby, Richard Doll, and Irene M. Stratton 3. Cancer Research Campaign (1982). In Trends in cancer survival in Great Britain. Cancer Research Campaign, London. 4. Lee, P.N. (ed.) (1976). Statistics of smoking in the United Kingdom (7th edn). Tobacco Research Council, London. '5. Wald, N., Doll, R., and Copeland, G. (1981). Trends in tar, nicotine and carbon monoxide yields of U.K. cigarettes manufactured since 1934. British Medical Journal, 282, 763-5. 6. Fairweather, F.A., et al. (1981). Changes in the tar, nicotine and carbon monoxide yields of cigarettes sold in the United Kingdom. Health Trends, 13, 77-81.

' 2063628247

Journal of Epidemi.o(ogy and Community Health, 1989, 43, 168-I 72 Predi Whitehall. Study . andTh~tt • . quesii KRISTIE L EBI-KR~;STON " comb: ' " " subcb Fr~mtheDepartmef~.t~fEpidemi~l~gy.~L~nd~n~ch~fHygieneat~dTr~pica~Medi¢ine;.K.eppe~Street~L~.nd~n.~. ~. ": Comi~ ABSTRaCt Fifteeri year chi'onic bronchitis mortality was investigated among 17 717 male civil servants aged 40-64 yea. rs par.ticipating in the Whitehall Study..Associations.were assessed.between " mortality and Medical Research Council standardised questions about chron.ic phlegm production and breathlessness, and a measure of lung function. Low FEVl was the most powerful single predictor of mortality; controlling for age, smoking habits and employment grade, the relative hazards ratio (RHR) was 20. Using mortality rates standardised for age and smoking, the proportion of mortality in the total population statistically attributable to low FEV1 (population excess fraction) was 57%. Breathlessness while walking on the level was the best predictor among the questions and combinations of questions; the relative hazards ratio was 12 and the population excess fraction, 39%. A Medical Research Council definition of chronic bronchitis including chronic phlegm production and breathlessness was also strongly associated with chronic bronchitis mortality (RHR= 13); however, the population excess fraction was only 20%. This definition identified only 30% of the 64 deaths, and added almost nothing to prediction by FEV1 alone. The results suggest that although the combination of chronic phlegm production and chronic airflow limitation is strongly associated with mortality from chronic bronchitis, the presence of chronic phlegm production alone is not associated with mortality. One of the definitions of chronic bronchitis recommended by the B~onchitis Research Committee of the Medical Research Council in 1965~ a includes report of both chronic phlegm production and breathlessness, where breathlessness was used as an indication of chronic airflow obstruction. This definition was based on the belief that individuals reporting chronic phlegm production would later become disabled by chronic airflow obstruction.3 Other definitions of chronic bronchitis have been proposed, including chronic phlegm production alone, and cough with expectoration not attributab.le. 14 to othe.r.lu~.'g dis.eases... 'R~.c.ent's.tudi~s :l~ave sho~v.~- th/~t ~h~b'nie ~l~tilegmproduction knd chronic airflow obstrq.cfion.are two.separate disease pmcessesJ-m Fletchdr and Pride recently advocated that the term chronic bronchitis be used "only to denote chronic or. recurrent bronchial hypersecretion." While the .design of future epidemiological investigations can benefit from ctirrent knowledge, analyses of data already collected must be limited to the questions asked and the niea~urements'taken. SuCh" analyses have. provided: and 6ontinue to p.roviite, valuable insights ifito the detdrminants of .chronic diseases. Many .studies have used• the Medical .Research Council (MRC) question~ about el-ironic phlegm production and breathlessness walking on the level, or other questions derived from or comparable to them. This study was designed to compare those standardized questions, a standard lung function measurement, and combinations of questions and measurement, to determine which provides the most satisfactory prediction of 1'.5 year mortality from chronic bronchitis in the Whitehall Study of male civil servants. • ". '.•.'-)c"--. ....'.cox. '.:-" .... "'.':" " )... i !:'" .'dev l'Si , . .: of iesl5 In "~967-6~, l~'~03~ale .civil servants aged 40-64 ~ofifou years wer~ examindd and questioned iri the Whitehall Study. In t.hepresent arialysis, smokers of only pipes .or cigars (n=640) and men with unknown smoking habits (n=46) were excluded, leaving a total population of 17 717. Death certificates were provided • for each subject who died within the United Kingdora 'by the Central Regist.ry of the National Health Service)Underlying causes of death were coded to' the 8th" revision. of the International Classificatioil. of "1"68 .... " " : :" '" "" .- , .'. ,: , , ";: .: :.., ..- and w be rt bronc; mor.ni • si~pk . increa who r regula increa and a lastin~ consid breath increa: definit Puh Three expirm from ~ capaci~ obtain, nien .r~ abgut. questk breathi calcula used a: < 65% onlym phlegrr persist~ for m~ • withou grade. ! current were s "persisl low FE 'shown-) i.h.'ablts ~ .r~ number O grade ~

Predicting 15 year chronic bronchitis mortality in the Whitehall Study Diseases. There were 64 deaths from chronic bronchitis (ICD8 49i..0-491-9) in ..the' 15 years of "follow up: Analyses ~¢re r~p~te'd ,for the 76 deaths." • fr6m ~hronic'airway~ .obst~'t.'i~n (ICD8 490.0~492'.9.) • . and the results 'were similar (not shown). The four questior/s in the shorter MRC bronchitis questionnaire12 were analysed individually and in combination. These questions were formulated by a subcommittee of the Medical Research Council's Committee on the Aetiology of Chronic Bronchit.iS,.~3 and wet~ based'on questions which had.bebn'shownto be reliable and valid in diagnosing chronic bronchitis.14 The questions about phlegm most mornings for 3 .months (persistent morning phlegm or simple bronchitis) and about a recent period of increased cough and phlegm were asked only of men who responded affirmatively to the question about regular phlegm in the. winter. "Persistent and increased phlegm" includes both persistent phlegm and a recent period of increased cough and phlegm lasting 3 weeks or more. This definition does not consider the responses to the question about breathlessness walking on the level. "Persistent and increased phlegm and breathlessness" is an MRC definition of chronic bronchitis. Pulmonary function was assessed by spirometr3'. Three blows were recorded, and the mean forced expiratory volume in one second (FEV~) was obtained from the two blows with the highest forced vital capacity (FVC). Predicted values of FEV~ were obtained from a linear regression of age and height for men responding negatively to the MRC questions about winter phlegm and breathlessness, and to questions about wheeze and weather affecting breathing (n=2806). Percent of predicted FEV~ was calculated, and values below 65% ("low FEV~") were used as a measure of pulmonary impairment. "FEV~ < 65% predicted, without persistent phlegm" includes only men with low FEV~ who did not report persistent phlegm production. The results for the 2353 men with persistent phlegm without FEV~ are similar to those for men with "persistent and increased phlegm without breathlessness" and so are not reported. Cox proportional hazards regression models were develooed usi.n~BMDP-2L~ s for the various measu~'~s of respi" ratory impai'rmeni,e~trollihg (or potential • -confounding by age, smoking habits and employment grade. Mean ages were slightly higher and proportions currently smoking or with high employment grade were slightly lower for men with breathlessness, "persistent and increased phlegm and breathlessness;" low FEVI, or any combination of these measures (not • shown). Age was entered as si.ngle years; smoking habits as never smoker, ex-smbker, or smoker, and • tlumber of cigarettes smoked daily; and employment grade was entered as low (clerical or other) or high 169 (administrative, ex.ecutive,. 9r professional, scientific arid ~echnical). The magnitudes of associations, adjasting for..eovariates~.were estimated by calculaiing relativ.e hazard~ raiioi from ~h~ r~gr~s~ion Population excess rates and population excess fractions~6 were calculated using mortality rates standardised for age and smoking. Results Pre~al'efice rat~ of'tl4~'v~ohs'~aeasti~e~ of respiratory impairment ranged from a high of 23. t % for "usually .have phelgm in winter" to a low of 0"9% for "low FEV~ and persistent and increased phlegm and breathlessness" (table 1). Approximately half of the men who reported breathlessness or had a low FEVt also reported persistent phelgm production. The numbers of deaths attributed to chronic bronchitis were small. None of the measures identified all 64 deaths; the majority of deaths were identified by the presence of low FEV~ alone (n=45) or by low FEV~ plus breathlessness (n= 50). Adding "persistent and increased phlegm and breathlessness" identified only one additional death to those identified by low FEV~. Smoking was a major risk factor for chronic bronchitis mortality. Only one death occurred among never smokers; there were 14 deaths (21.9%) among ex- smokers and 49 (76.6%) among smokers. All of the measures except "persistent phlegm without breathlessness" were significantly associated with chronic bronchitis mortality in Cox proportional hazards regression models (table 1). The relative hazards ratios were highest for low FEV~ alone or in combination with breathlessness or "persistent and increased phlegm and breathlessness" (RHR= 19-26). The ratios were lower, but still large (RHR = 12) for "breathlessness walking on the level" and "persistent and increased phlegm and breathlessness." Excluding men who also reported persistent phlegm production markedly reduced the relative hazards ratios for breathlessness (I1.7 to 3.4) and low FEV~ (19.9 to 4.4), which suggests that the presence of both persistent phlegm production and airflow limitation predict, higher .mortality from chronic' bronchitis. ~pw~ver; persistent,..'phlegm .. production• withot~t br'eathldssness or low FEVt was. not significantly associated with mortality. Of men reporting breathlessness, .319 (32-6%) also had low FEV~; and 153 (52-6.%) of men with "'persistent and increased phlegm and breathlessness" also had low FEV~. Further Cox proportional hazards regression models were developed to assess the associations of breathlessness and "persistent and increased phlegm and breathlessness" with mortality rates while controlling for FEVa percent predicted (ngt shown). The relative hazards ratios were

170 Tabie I Prevalence rates and estimated parar~etersojf Cox proportional hazards r.egfession models for "15 year chronic bronchitis mortality by various measures of.respiratory impairment, males 40-6~l years, .Whi~t~hall Study. . ~. ~.. " at ~a)" Model " * " .... ' CoeffiCient . SE RHR* " • 25 Age (single years) 0' 1572 Smoking habits Ex-smoker v non-smoker 1,7167 , Smoker ~ non-smoker 2'1055 Number of cigarettes smoked daily 0'0352 Employment ,,grade (low v high) . 1.2727 ,'. (If.) Me, asares of raspiratdry impairment ~aitded individually:to model i~t~ave)'" " ," "'. " " " Prevalence :%) Coefficient 0.0260 1.0355 1.0372 0.0120 0.2750 SE 1.17 5.57 8.21 1,04 3.57 Kristie L Ebi-Kryston Predict Usually have phlegm in winter 23.1 1.3807 Pea'sistent morning phlegm 15.8 1.3700 R~c~nt period of increased cough and phlegm 9.7 0.6794 Persistent and increased phlegm 7.2 1.4213 Breathless walking on the level 5.5 2.4594 Persistent phlegm without breathlessness 13.6 - 0.2508 Breathless without persistent phlegm 3,3 1.2082 Persistent and increased phlegm and breathless 1.7 2'5216 FEVI <65% predicted 6,4 2.9883 FEVI < 65% predicted without persistent phlegm 3'9 1"4691 FEVI <65% predicted or breathless 10.2 2.9665 FEVI <65% predicted and breathless 1.8 3.1737 FEVI <65% predicted or phlegm and breathlcss~" 7.2 3.2873 FEVt <65% predicted and phlegm and breathless 0.9 2'9726 RHR* Deaths 0.2792 3.98 44 0.2628 3,94 37 0" 1305 1.97 27 0.2f~54 4.14 24 0.2586 11'70 33 0.3262 0.78 12 0.3795 3"35 8 0"2891 12.45 19 0.2839 19'85 45 0'3058 4'35 14 0,3111 19.42 50 0.2659 23.90 28 0,2833 26.77 46 0.2975 19"54 17 Figure for men Whiteha men in the chr~ high rel (1"6 per * Relative hazards ratio t Adjusted for age only markedly reduced, to 2.89 for breathlessness and to 2.25 for "persistent and increased phlegm and breathlessness." Although airflow limitation confounded the associations between these measures and mortality, breathlessness and "persistent and increased phlegm and breathlessness" remained significantly associated with mortality. Using chronic 'bronchitis mortality rates standardised for age and smoking, the population excess rates per 1000 were calculated (mortality rates in the total population statistically attributed to the measure) together with the population excess fractions (proportion of mortality in the total population statistically attributable to the measure) for selected measures of respiratory impairment. These results are shown in table 2. Both statistical measures were highest for low FEV~ and low FEVI plus either breathle, ssness or "persistent and inc.reased phlegm agd breathlessnegs" .(population.e~ee~s.rates = 1.6-1-8 pe¢ 1000, populafi.on excess fractions ~ 57-67%).. ' Persistent phi&gin and breat.hlessheg.s had intermediate values of population excess rates (1"!-1"2 per 1000) and population excess fractions (34--39%), with the other measures having smaller values. As shown in the figure, the distribution of values of FEV~ (percent of predicted) for men who died of chronic bronchitis was much lower than the distiSbutio~a for men who did "not. Using.a cut off of less than 65°/'0 predicted captures most of the deaths .. without irlcluding-too many false positiyes. The distributions of values were similar for smokers and ex-smokers; 70% of the deaths (34 of 49 smokers and I 0 of 14 ex-smokers) occurred in men with a low FEVI. Discussion These analyses of a group of male civil seryants suggest that low FEV~ is the best single predictor of 15 year mortality from chronic bronchitis; the 6-4% of Table 2 Population excess rates and population excess fractions for 15 year chronic bronchitis mortality by various measures of respiratory impairment, males 40-64 years. Whitehall Study* the dis predictt Adding number excess r fraction subgrm specific and inc effect o fraction FEV~ ~ increase number combin Relat "persist Measures of respiratory impairment Population E.~:cess rate (~. f. ~ooo) and bre Population excess f Exce~a fraction ¢ *./, ) - "persist " ,-.., . whiehs • ' pr"edietc ." few dea "incredst remaini much s bronchi identifie breathk fraction C~ was ab~ O~ 8ugg.esL, ~o predict~ Persistent. and increased phlegm 0.78 23.8 " Breathl.es~. wa~.ng on ~e level .1.20 38.6 Persistent and increased phlegm' and breathless 0-64 19.8 FEVI <65% predicted 1.57 .56.7 FEVI <65% predicted or breathless 1"83 67' I FEVI <65% predicted and breathless 0.91 30. I FEVI <65% predicted or phlegm and breathless 1'62 5'9,6 FEVI <65% predicted and phlegm and breathless O' 53 16"6 * .Based .... r~ality rates a~lj.ust'ed foz a.g.e and s ,moldng •

onchitis rants off5 % of excess arious years, Predicting 15 "year chronic bronchitis mortality in the Whitehall Study 171 The relatively poor ability of one M RC definition.of chronic bronchitis to predict mortality from chronic / ~ o--o o~ [ __-. bronchitis is i.nteresting. Including in the.definitibn . z~ both chronic phlegm pr.o.duction,.which ~as.not lethal ~2~ in this study, and breathlessness, Which Was~~nay.h~veI ~ weakened its predictive abilit3/... Nearly all of the deaths • ~ from chronic bronchitis occurred among smokers. • ~ ~ Mortality rates from chronic bronchitis have been decreasing,17 which may in part be due to lower rates ~ 10 of cigarette smoking and to changes in cigarettes, specifically lower average tar yields.18 Lower..tar ~ cigarettes may not havethe.same influence or/.chronic phlegm production and airflow obstruction,19 further ~o~ z~.~ 3~.a~ ~s.~ ~.u ~s.~ ~s.~ ~.~ ~s.lo, lo~.,~ ,1,a diminishing the predictive ability of the definition. ~v~ ~ ~,~,,~ In summary, these analyses suggest that a measure Figure D~stribution of values of FEV~ per cent of predicted of reduced lung function provides the best prediction for men who died from chronic bronchitis and men who did not, in these data of subsequent mortality from chronic Whitehall Study bronchitis. If lung function measurements are not available, the next best choice is the question about men iri the study with low FEVt experienced 70% of breathlessness walking on the level. "Persistent and the chronic bronchitis deaths. This is reflected in the increased phlegm and breathlessness" predicts high relative hazards ratio (20), population excess rate mortality through its association with airways (1.6per 1000), population excess fraction (57%), and obstruction. the distribution of values of FEV~ (percent of predicted) for men who died and those who did not. The author gratefully acknowledges the helpful Adding men with breathlessness slightly increased the suggestions of Professor G Rose. number of deaths identified (to 50), the population The study was supported in part by Training Grant excess rate(to 1.8 per 1000)and the population excess No 5 T32 HLO7337-09 from the National Heart, fraction (to 67%), while increasing the size of the Lung and Blood Institute. subgroup by 37%, resulting in a lower subgroup specific mortality rate. Adding men with "persistent Address for correspondence and reprints: Dr Kristie L and increased phlegm and breathlessness" had little Ebi-Kryston, Department of Environmental and effect on prediction. The population excess rates and Preventive Medicine, The Medical College of St fractions were smaller for combinations of both low Bartholomew's Hospital, Charterhouse Square, FEV~ and either breathlessness or "persistent and London EC1M 6BQ. increased phlegm and breathlessness" as were the number of deaths identified, making these combinations less useful prediiztors. References Relative hazards ratios were large (12) for both "persistent and increased phlegm and breathlessness" t Medical Research Council. Definition and classification of and breathlessness. The population excess rate and chronic bronchitis for clinical and epidemiological purposes. Lancet 1965; i: 775-9. excess fraction was higher for breathlessness than for z Medical Research Council. Questionnaire on respiratory "persistent and increased phlegm and breathlessness," symptoms and instructions for its use. London: Medical which suggests that breathlessness may be the better Research Council, 1966. predictor of chronic bronchitis mortality. Relatively 3 Fletcher CM. Terminology in chronic, obstructive lung diseases. J Epidemiol Community Health 1978; 32:'282--8. few deaths occurred, among men with "persistefit and 4 Ciba ~ Fofindation Guest Symposififa. Tei'minology, increased phlegm and breathlessness" (19 of'64). The definitior/s and classifications of chronic pulmonary remaining measures of respiratory impairment had emphysema and related conditions. Thorax 1959; 14: much smaller relative hazards ratios with chronic 286-99. 5 Higgins MW, Keller JB. Predictors of mortality in the bronchitis mortality. Persistent morning phlegm adult population of Tecumseh. Respiratory symptoms, identified approximately the same number of deaths as chronic respiratory disease, and ventilatory lung breathlessness, and the population excess rates and function. Arch Environ Health 1970; 21: 418-24. fractions were similar. However, the prevalence rate 6 Fletcher CM, Peto R, Tinker CM, Speizer FE: The natural history of chronic bronchitis and emphysema. Oxford: was about three times that of breathlessness, which Oxford University Press, 1976. suggests that breathlessness is a more specific 7Fletcher CM, Peto R. The naturalhistory of chronic predictor of mortality, airflow obstruction. Br Med J 1977; i: 1645-8.

172 s Kauffmann F, Drouet D, Lellouch J, Brille D. Twelve years' spirometric.changes amo.ng Paris area workers. Int J Epidemiol 197.9; 8: 201-12. 9 Peto R, Speizer FE, Cochrane AL, et al. The relevance in adults of air-flow obstruction, but not of mucus hypersecretion, to mortality from. chronic lung disease. Results from 20 years of prospective observation. Am Rev Respir Dis 1983; 128: 491-500. i0 Ebi-Kryston KL. Respiratory symptoms and measurements as predictors of 10-year mortality from respiratory disease, cardiovascular disease, and all causes • ifitheWhitehallstudy.JC.linEpidemiol1988;41:251-60. l~Fletcher CM, Pride NB. 'Definitions of emphysema, chronic bronchitis, asthma, and airflow obstruction: 25 years on from the Ciba syhaposium. Thorax 1984; 39: 81-85. 12 Reid DD, Brett GZ, Hamilton PJS, Jarrett RJ, Keen H, Rose G. Cardiorespiratory disease and diabetes among middle-aged male c~vil servants. A study of screening and intervention. Lancet 1974; i: 469-73. Kristie L Ebi-Kryston Journa ' 13 Medical Research Cour~cil, committee on the Aetiology of Chroiaic Bronchitis. Standai'dized q~estiohhaires on • ' ". 14 resfiratory symotoms. Br'Med J 1960; ii: .1665. • .... " ." " Cochrane AL, ~hapman PJ, Oldham PD. Obs6rvers' ' errors in tak ng medica h stories Lancet 1951;'i: 1007'-9. ' '1 1"11~ 15 Dixon W J, ed. BMDP statistical software, 1983 printing with additions• Berkeley: University of Califorma Press, ~-~r~ 1983. ~6 Last JM, ed. A dictionary ofepidemiology. Oxford: Oxford University Press, 1983. 17 Melia RJW, Swan AV. International trends in mortality . S G "I rates for bronchitis, emphysema and asthma during tl~e. From t period 1971-1980. WorMHealth Stat Q 1986; 39: 206-17. 18 Capell PJ. Trends in cigarette smoking in the United Kingdom. Health Trends 1978; 10: 49-54. 19 Higenbottam T, Shipley M J, Rose G. Cigarettes, lung cancer, and coronary heart disease: the effects of inhalation and tar yield. J Epidemiol Community Health 1982; 36:113-7. Accepted for publication November 1988 ABSTR; Study replac Oesigt Settin; Park ~ Subje~ gener~ obtair Meas~ contrt An in inforn Mo showi [RR] prepa', used 9 risk a interp cardic ConcL major Currei eardio hormo Evidet includ~ womel sarlie 8 a surg advers for ex~ in men on th~ eontra prosp~ Howe, Whiteh~

2063628253

994 BRITISH MEDICAL JOURNAL 5 APRIL I980 BgtTIStt .~t~ weighted Personal Paper the propo; ettes has ~ extent to, other factt that a lo~ Comment on the Hunter Committee's second report M J JARVIS, M E.H RUSSELL ' " reduce tar low nicoti smokers. I in this dir . thzt.a cot dangerous (again Dr the problc The Independent Scientific Committee on Smoking and Health (the Hunter Committee) was first established in 1973. Its terms of reference were wide--to advise on the "scientific aspects of matters concerning smoking and he.alth,'.', in particular: (1) "To receix~e in confidence full data about the constituents of cigarettes and other smoking materials and their smoke and changes in these; and to release to bona fide research workers for approved subjects such of the above as is agree.d by the suppliers of it.'.' (2) "To review the research into less dangerous smoking and to consider whether further such research, including clinical trials and epidemiological studies, needs to be carried out." (3) "To advise on the validity of research results and of systems of testing the health effects of tobacco and tobacco substitutes and on their predictive value to human health." After seven years and at least 350 000 more deaths due to smoking the second report of the Hunter Committee has just been published.1 It seems an appropriate time to consider the contribution that this committee has made. It has to be said at the outset that, in the face of the size and the urgency of the problems posed by smoking, its record is disappointing in the extreme. Its first report~ in 1975 was concerned largely with estab- lishing guidelines for testing tobacco products containing tobacco substitutes. Cigarettes containing tobacco substitutes now account for less than 1% of the market so that the first report has turned out to be largely irrelevant to the immediate problems of smoking in Britain. Indeed, a poignant paragraph in the second report notes that a proposed long-term study of the effects on human smokers of cigarettes containing sub- stitutes could not be implemented because there were not enough smokers smoking them. We cannot blame the Hunter Committee for that, but it does point to a failure on the part of the tobacco industry at that time adequately to consider the smoker as well as the product smoked. The Hunter Committee has not learned from this failure, but has repeated it in its second report. The second report, published five years and 19 meetings after the first one and after skting foi" 13 months on'the desks of the UK Health- Ministers, fs jn some ways a remarkable document. It is remarkable for its brevity, for what it does not contain as well as what it does, and for the way it totally fails to measure up to the urgency of the smoking problem (as noted by Dr J Donald Ball in his dissenting minority report). Most of all it is remarkable for the stunning naivety of its implicit model of smoking behaviour. In addressing the issue of the development of "lower risk" Addiction Research Unit, Institute of Psychiatry, Maudsley Hospital, London SE5 8AF M J JARVIS, Bsc, lVlPHIL, clinical psychologist M A H RUSSELL, MRCP, MRCPSYC.~t~ senior lecturer and consultant psychiatrist cigarettes the report, with some complacency, pats the industry on the back for achieving a reduction in average tar yield from 31"4 mg per cigarette in 1965 to 17"4 mg in 1977 with parallel decreases in..nicotine yields, and rdcommends "further sub-.- stantial reductions in tar yields" in the future. Lower risk cigarettes are equated with cigarettes with lower tar and nicotine yields. If people smoked cigarettes in the same way that smoking machines do, this would indeed be the case. But there is much evidence that they do not)-~ Smoking machine to smoker The tendency for smokers to regulate their smoke intake has been ignored by the Hunter Committee. The committee's thinking appears to be dominated by an obsession with machine- smoked yields. On the basis of machine-smoked yields the smoking of large cigars should be the most deadly form of tobacco use, but epidemiological studies show them to be far less harmful than cigarettes. One would have hoped that this discrepancy would have made the Hunter Committee more cautious about extrapolating too directly from smoking machine to smoker and that it would have made it place as much emphasis on measurements of the smoke intake of smokers as it has placed on the smoke output of cigarettes. Nowhere in the report is there any reference to the numerous published studies on the tendency of smokers to modify their smoking pattern in response to changes in the tar and nicotine yields of their cigarettes. More serious is the omission of any reference or recommendation which shows any awareness whatsoever of the importance of measuring the smoke intake of smokers using blood nicotine,~ v blood carboxyhaemoglobin,~ 9 or blood thiocyanate~° n concentrations, or the well-established, simple, and inexpensive indirect measure of expired air carbon monoxide3~ A recent study of 330 cigarette smokers who had been smoking their usual brand in their usual way showed that blood nicotine .concentrations were similar in smokers of high tar plain cigarettes (nicotine yields 1 "9" mg), middle tar unventilated fill~er cigarettes (1.3 mg nicotine), and" lo~ tar cigarettes v,~ith'v~ntilat~d fill}e'rs (0"8 m~ nicotine).~-" Since tar and nicotine yields .are highly correlated (> 0"9) it may be inferred that the intake of tar to the lungs of these three groups was also sirrlilar. Such results come as no great surprise to anyone who has been reading the tobacco-smoking publications. These suggest that an approach aimed simply at further reductions in tar and nicotine deliveries will do little to reduce the dangers of smoking. This is not only because smokers compensate by increasing inhalation so as to leave their smoke intake relatively unchanged, but also because a point is reached where reduced deliveries meet with reduced consumer acceptance (Lord Hunter indeed touches on this point in a covering letter to the secretaries of state). There is not much point in providing cigarettes that no one, other than non-inhalers, ~vill smoke. There is evidence that we are already approaching such a barrier. The average sales- 2063628254 Catch-22 If the l behaviour indt/s'try" i suggested to ultra 1¢ acceptable response t It comme~ among sw precisely cigarettes force. Th people wt desire to cigarettes strenuous desire." 2 runs som dangerou~ sired me, carmot gix The co~ is equally and "oth, required I containin~ toxicity t~ at least as and is ine Publlcat Since mended that this mitte,e, wI to imaglm terms of workers "agr.eed I~ for some as the ta~ appear tc Indeed, t noted tha The ~ of CO ~ regularly have bec

weighted nicot'ne, t~ yield has shown no decline since 1974, and the proportion of smokers smoking low tar, low nicotine cigar- likewise become stuck at around a mere 11%. The to which this is due to lack of nicotine, or tar, or some r factor is still unknown but crucial. It has been suggested" ~a that a low tar, low CO, but medium nicotine cigarette might reduce tar and CO intake more than occurs with low tar, low CO, low nicotine cigarettes. It might also be more acceptable to smokers. Present evidence supports the view that a new approach in this direction would be worth investigation. It is astonishing that a committee appointed to "review the research into less dangerous smoking" should largely ignore this crucial area (again Dr Ball in his minority report shows some awareness of the problem). Catch-22 If the Hunter Committee is unaware of the importance of behavioural factors and of thb role of nicotine .the tobacco industry is not. The report notes that some companies "have suggested that the addition of natural nicotine or nicotine salts to ultra low tar and nicotihe products would produce a more acceptable smoke for dependent smokers." The committee's response to this could hardly be more lukewarm or disappointing. It comments that "if this resulted ih an increased dependence among smokers, then it would be difficult to approve it." Yet it is precisely because so many smokers are highly dependent on cigarettes that the argument for lower risk cigarettes gains its force. This is in fact recognised earlier in the report: "Many • people who feel they cannot yet give up smoking have a strong desire to smoke less dangerously either by smoking fewer cigarettes or lower risk cigarettes. The committee believes that strenuous efforts should continue to be made to meet this The position is, in effect, Catch-22. The argument something like this: many smokers want to smoke less because they cannot give up, but must not be given medium nicotine low tar cigarettes if these mean they cannot give up. The committee's othercomment on nicotine-spiked cigarettes is equally fatuous. It is stated that "toxicity testing in animals" and "other studies in man" (unspecified) will probably be required before such cigarettes are permitted. Why a cigarette containing, say, 4 mg tar and 1 mg nicotine should require toxicity testing when currently available cigargttes containing at least as much nicotine and far more tar do not is unexplained and is inexplicable. Publication of CO yields Since the publication of CO yields of cigarettes is recom- mended by Dr Ball in his minority report, we must assume that this step ~vas considered and reiected by the whole com- rnittee, which nevertheless gives no reasons for this. It is difficult to imagine what those reasons might have been. The committee's terms of reference empower it to "release to bona fide research workers for approved subiccts" data on smoke constituents "agreed by the suppliers of it." The Government Chemist has for some time been routinely measuring the CO yields as well as the tar and nicotine yields of cigarettes, and there do not appear to be any undue technical difficulties of measurement. Indeed, the first report of the Hunter Committee specifically noted that the method has "proved very satisfactory." The World Health Organisation has urged the publication of CO yields in t~vo reports2~:' CO yields are published arly in many other countries. Medic~ scientists in Britain been clamouring for CO yield data)" ~: There do not 995 seem to be any grounds for withholding publication. One is forced to the conclusion that the Hunter Committee's decision not to recommend publication did not stem from technical considerations. Finally, what about other harmful constituents ? The com- mittee's response on this issue is to procrastinate. It states briefly that it "proposes to ask the industry to provide full relevant data to assist in reviewing and evaluating the constituents of smoke so that the committee is better able to advise the Secretaries of State about the desirability or otherwise of settirfg levels for some of these constituents. Although it li~ts about a dozen potentially harmful constituents including, for example, oxides of nitrogen, hydrogen cyanide, benzo(a)pyrene, polycylic aromatic hydrocarbons, and nitro- samines, no recommendations are made for the publication of these data. Yet current cigarettes show enormous variation in their yields of some of these products. Current brands, for example, have an eight-fold variation in delivery of oxides of nitrogen,TM which the committee acknowledges "contribute to the over.all pathological changes induced by smoke in th.e lung 'parenchyma leading to emphysema." How much longer must we wait for the committee to release the data, let alone give advice on control and regulation of all these harmful substances ? If its shilly-shallying over CO yields is anything to go by, we are in for a long delay. References x Independent Scientific Committee on Smoking and Heatth. 2nd report. Developments in tobacco products and the possibility of lozoer-risk cigarettes. London: HMSO, 1979. (Released to the public 12 February 1980.) ~ Independent Scientific Committee on Smoking and Ilealth. 1st report. Tobacco substitutes and additives in tobacco products. London: IIMSO, 1975. z Creighton ]DE, Lewis PFL In: Thornton RE ed. Smoking behaviour: physiological and psychological influences. London: Churchill Living- stone 1978:289-300. a Russell MAIl. In: Krasncgor NA ed. Cigarette smoking as a dependence process. NIDA research monograph 23. Washington DC: US Govern- ment Printing Office, 1979:100-22. ~ Ashton H, Stepney R, Thompson JW. Self-titration by cigarette smokers: Br MedJ 1979;ii:357-60. ~ Russell MAH, Sutton SR, Feyerabend C, Saloojee Y. Smokers' response to shortened cigarettes: dose reduction without dilution of tobacco smoke. Clin Pharmacol Ther 1980;27:210-8. ~ Russell MAH~ Wilson C, Patel UA~ Cole PV, Feyerabend C. Plasma nicotine levels after smoking cigarettes with high~ medium~ and low nicotine yields. Br l~Iedff 1975 ;ii :414-6. ~ Russell M2G~I, Wilson C~ Patel L!A, Cole PV, Feyerabend C. Comparison of effect on tobacco consumption and carbon monoxide absorption of changing to high and low nicotine cigarettes. Br MedJ 1973;iv:512-6. ~ Wald NJ, Idle M, Smith PG. Carboxyhaemoglobin levels in smokers of filter and plain cigarettes. Lancet 1977;i:110-2. ~0 Pettigrew A_R, Fell GS. Simplified colorimetric determination of thio- cyanate in biological fluids and its application to investigation of the toxic amblyopias. Clin Chem 1972; 18:99~-9. t~ Vogt TM~ Selvin S, Widdowson G, Hulley SB. Expired air carbon monoxide mad serum thiocyanate as objective measures of cigarette exposure. ~tm ff Pub Health 1977;67:545-9. ~z Russell MAH, Jarvis MJ, Iyer R, Feyerahend.. C. Relation of nicotine yield of cigarettes to blood nicotine concentrations in smokers..Br Med~Y 1980;280:972-6. ~ Russell MAIl. Low-tar medium-nicotine cigarettes: a new approach to safer smoking. Br ,~fedJ 1976;i:1430-3. ~ World Health Organisation. Smoking and its effects on health. Technical Report S~ries No 568. Geneva: WHO, 1975. ~ World Health Organisation. Controlling the smoking epidemic. Technical Report Series No 636. Geneva: WHO, 1979. ~6 Ball KP; Carbon monoxide yield of cigarettes. Br MedJ 1979;ii :731. x~ Wald N, Doll R. Carbon monoxide yield of cigarettes. Br Medff 1980; 280:646. xs Diamond L. In: Banbury Report 3. New York: Cold Spring Ilarbor Laboratories, 1980. (Accepted 18 s~larch 1980)

2063628256

ir~g behavior as diapausing C. pipiens, JH may initiate biting behavior in many of these insects. Ti4e possibility that JH initiates biting behavior in Anopheles freeborni has al- ready been suggested by Case et al. (3). They found that the JH mimic Meth- oprene stimulated both biting behavior and diapause termination. However, a second JH mimic (6,7-epoxygeranyl-4- ethylphenyl ether) failed to elicit biting despite diapause termination. Case et al. postulated that a natural JH might initi- ate biting in Anopheles freeborni, and that the discrepancy in results was due to differences in molecular structure be- tween the mimics tested and natural JH's. Recent studies suggest that JH is not the only hormone involved in the regula- tion of mosquito biting behavior. In Aedes aegypti and Anopheles freeborni, ovaries with developing eggs secrete a hormone that suppresses host-seeking or biting behavior between gonotrophic cycles (15). Whether this ovarian hor- mone influences JH synthesis to pre- vent biting during egg development is unknown. ROGER W. MEOLA ROtqALD S. PE'rRALIA Department of Entomology, Texas A & M University, College Station 77843 References and Notes 1. R. W. Gwadz and A: Spielman, J. Insect Physiol. 19, 1441 (1973); A. O. Lea, ibid. 14,305 (1968): R. W. Gwadz, L. P. Lounibos, G. B. Craig, Gen. Comp. Endocrinol. 16, 47 (1971). 2. M, M. Lavoipierre, Nature (London) 181, 1781 (1958); S. S. Chen, Diss. Abstr. B30 (9), 4187-B (1969); J. G. Stoffolano, Jr., in Experimental Analysis oflnsect Behavior, L. B. Browne, Ed. (Springer-Verlag, New York, 1974), p. 32. 3. T. J. Case, R. K. Washino, R. L. Dunn, Ento- mol. Exp. Appl. 21. 155 (1977). 4. This long-day, warm-temperature regimen was selected to avoid any diapause-related effects on the corpus allatum of C. pipiens which might confuse interpretation of results. C. quinquefas- ciatus, a nondiapausing species, was also reared and maintained in this regimen. 5. Mosquito larvae (lOOper 350 ml of tap water) were reared in covered plastic pans (27 by 19 by 6 cm) on a diet consisting of equal parts Brew- er's yeast, lactalbumin, and finely ground labo- ratory animal chow. Daily rations varied with larval stage; newly hatched (day 0) and day 1, 150 rag; day 2 and day 3,250 rag; day 4,450 rag; and day 5, 250 mg. 6. A. O. Lea, J. Insect Physiol. 9, 793 (1963). 7. R. Meola and A. O. Lea, J. Med. Entomol. 9, 99 (1972). 8. A. Spielman, ibid. 11,223 (1974). 9. According to A. Clements [in The Physiology of Mosquitoes (Macmillan, New York, 1963), p. 691, egg maturation in mosquitoes is a two-statue process. The resting stage represents the initial phase of follicular development to a point where the oocyte may be distinguished from the nurse cells. The second phase of development in- volves deposition of protein yolk which occurs after the mosquito feeds on blood. A. O. Lea, J. Insect Physiol. 15, 537 (1969). Juvenile hormone-I (methyl-3,11-dimethyl-10, I 1-cis-epoxy-7-ethyl-trans,trans-2, 6-tridecadi- 10. 11. 324 (1961); S. Kawai, Trop. Med. 11,145 (1969); L. Sandburg and J. Larsen, J. Insect Physiol. 19, 1173 (1973); A. Spielman and J. Wong. J. Med. Entomol. 100 319 (1973). 13. B. F. Eldridge, Science 151,826 (1966), 14. _ and C. L. Bailey, J. Med. Entomol. 462 (1979). 15. M.J. Klowden and A. O. Lea, J. Insect Physiol. 25, 231 (1979); R. Beach, Science 205, 829 (1979); M. J. Klowden, ibid. 208, 1062 (1980). 16. A. N. Franzblau,A Primer of Statistics for Non- Statisticimts (Harcourt, Brace'and World, New York, 1958). 17. Supported by U.S. Army Research and Devel- opment grant DAMD 17-79-C-9103. Approved as T. A. No. 15987 through the director of the Texas Agricultural Experiment Station. We thank C. Bailey, H. M. Kaska, L. L. Keeley. M. J. Klowden, and A. O. Lea for helpful com- ments. 9 April 1980; revised 17 June 1980 Have Tar and Nicotine Yields of Cigarettes Changed? Abstract. In official assays of the tar and nicotine yields of 12 popular brands of cigarettes, smoking machines took fewer puffs, on the average, in 1974 than in 1969. The decline in puffs appemw to have been a major cause of the reported reductions in tar and nicotine yields dttring this period. Since 1967 in the United States and 1969 in Canada, the governments have sponsored regular assays of the "tar" and nicotine yield of cigarettes by means of smoking machines (1). These assays show that the tar deliveries of most ciga- rette brands have declined in the last 10 years (2). The value of the published fig- ures has been criticized repeatedly, how- ever, on the grounds that smokers can compensate for reduced deliveries by al- tering the way they smoke, for instance by taking more or larger puffs (3). Clear- ly, the advantages of switching to "mild- er" cigarettes depend on the degree to which smoking behavior remains un- changed. Similarly, fair comparisons of tar and nicotine deliveries require the be- havior of the smoking machines to be held constant. We believe that a loop- hole exists in the standard smoking-ma- chine procedure in that it does not speci- fy the number of puffs to be taken. The number of puffs taken per cigarette for some brands declined significantly from 1969 to 1974, and we believe that this change has contributed to the reported reductions in their tar and nicotine con- tent. Many people may assume that, puff for puff, newer versions of popular brands (4) have been becoming weaker in tar and nicotine. The standard proce- dure, however, fixes neither the number of puffs taken on different brands during the same test nor the number taken on the same brand in subsequent tests. The procedure prescribes that a smoking ma- chine (essentially a motorized syringe) take a 2-second 35-ml puff once each minute until a fixed butt length is reached (5). Number of puffs is determined, then. by the bum-time of the cigarette. Burn- time can be influenced, for example. by the porosity of the cigarette paper or the amount of tobacco in the ciga- rette. The Federal Trade Commission (FTC) laboratory has not saved records of the number of puffs taken per cigarette in its tests (6), but in Canada such information has been kept, although it has never been studied systematically or published. Wc report here an analysis of puffdata for of the best-selling Canadian filter ciga- rettes, which accounted for 60 percent of the cigarette market in 1970 and 70 percent in 1974 (7). Our analysis was limited to the I I Canadian sur- veys between 1969 and 1974, in which assays were done on the same machine and with the same analytical procedures (e, 8). For the 12 brands as a group. creases in tar (actually "wet ta,"') were strongly associated with decreases in the number of puffs taken by the smoking machine (r= .97, P < .01. d.f. = 4) (Fig. 1). There is a similar asso- ciation between puffs and tar for each 2063628257 Table 1. Comparison of the yields and weights (mean ± standard deviation) of 12 popt~mr brands of Canadian cigarettes in survey 1 (1969) and survey I l (1974). Paired t-tests (two-tailed' are used. Item Survey 1 Survey 1 ! Range of (1969) (1974) differences Tar(mg) 21.8 ± 1.98 18.6 +-- 2.24* 1.3 to 6.9 Nicotine(mg) 1.31 --- 0.14 1.15 + 0.14~" -.07 to .49 Weight(g) 1.12 ± 0.08 1.06 ± 0.08' .03 to .11 8.8 ± 0.98* .4 to 1.9 2.12 ± 0.22~ -.04 to .30 0.131 ± 0.012~ -.013 to .025 enoate) was purchased from ECO Chemical In- Puffg 9.8 .4- 1.1 termediates, ECO-Control, Inc., Cambridge, Tar her huff 2 2,1 + fl 21 Mass. --"~- ~7- r ........... 12. P. Tate and M. Vincent, Parasitology 28, 115 N~cot~ne per puff 0.135 ± 0.011 (1936); R. Wallis, Proc. Entomol. Soc. Wash. 61. 219 (1959); H. Chapman, Mosq. News 21, *P < .002. fP < .0I. ~tNot significant. ~00926~0 Copyright © 1980 AAAS SCIENCE, VOL. 209, 26 SEPTEMBER

bra~d across the 11 surveys. For the nine brands that showed a tar decrease of at :least 1.8 mg from survey' 1 to survey 11, !the co,rrelatior~coefficlents (d.f. = 9) for puffs and tar ranged from .60 to .94 i(mean = .79), all P's -< .05, two-tailed; for the three brands that showed less than a 1.8 mg decrease in tar, the correlations ranged from .22 to .48 i(mean = .35), none statistically signifi- [cant. Two U.S. brands that were includ- !ed in some or all of the surveys showed isignificant correlations between puffs and tar (Winston, r = .99, P < .01, d.f. 3; Kool, r = .93, P < .01, d.f. = 9). Table 1 shows the changes from the survey in 1969 to the last in 1974. Overall, the cigarettes weighed less in 974 than in 1969 and may have contain- less tobacco. Tar dropped by 14.7 ~ercent. One might expect that the ~le index of tar per puff would cor- the problems of comparing assays on differing numbers of puffs. Un- the chemistry of cigarette ~moke does not permit this easy solu- tion. Tar delivery increases with each raft; therefore omission of the last few rUffS can markedly alter the total tar de- and the tar per puff (9). In one of the same cigarettes,, tar per puff eight puffs was about 2.36 mg and nine puffs was about 2.50 rag; thus a of one puffcaused a decrease of 1.6 percent in tar per puff(9). Note that a lecrease of one puff from survey 1 to 11 caused a change of 5.4 percent tar per puff. In contrast, nicotine de- remains fairly constant with each on a cigarette (9). Consistent with chemical fact, nicotine per puff did change significantly from survey 1 to 11. Given the numerous ways to reduce delivery, changes in number of puffs do not account for all the re- in tar deliveries by specific over the last 20 years. Although unable to assess the relative con- of other causes of tar reduc- the conclusion seems inescapable at a reduction in the number of puffs by smoking machines has been a or factor in the apparent decrease in toxicities from 1969 to 1974. least as early as 1958, scientists the tobacco industry discussed the isleadingness and unfairness of assays on differing numbers of puffs, ar- ing that 14 puffs should be required for cigarette and that the puff in- should be adjusted so as to the proper butt lengths (10), Ob- smoking-machine conditions on a fixed .number of puffs per provide more realistic esti- 22 20 ~ • 1973 18 9.0 9.2 9 4 9 6 Number of Duffs Fig. I. Mean tar (tar = total particulate mat- ter - nicotine) and mean number of puffs per cigarette for survey years 1969 to 1974 for 12 brands of filter cigarettes. If two surveys were conducted in one year, the data are based on the mean values. Linear regression: TPM = 4.44 (puffs) - 21.85, r2 = .94. Smoking ma- chines do take fractions of puffs (0.1 puff = 3.5 ml of smoke). mates for those smokers who may be ac- customed to obtaining a certain number of puffs from their cigarettes. If a smoker does have a habit of getting a certain number of puffs (or, equivalently, a cer- tain total volume of smoke) per cigarette, the reduction in burn-time by roughly 10 percent should offer only slight impedi- ment to getting all the smoke desired. (Imagine the success of a diet which per- mitted 9 minutes rather than 10 minutes to consume an ice cream cone.) Further- more, in one study the number of puffs per cigarette taken by people in highly controlled laboratory settings had a stan- dard deviation of 1.26 (mean = 8.5 puffs per cigarette) (11). It can be argued that if these smokers had been smoking a pack of 1974 cigarettes, often they would have smoked, in effect, a 1969, often a 1972 cigarette, and so on, depending on the number of puffs they took. The results of standard smoking-ma- chine assays are used extensively and of- ten at face value by epidemiologists and other researchers interested in mon- itoring smoking habits and the hazards of smoking (12). In addition, the smoking public has been encouraged to select less hazardous brands based on tar : nicotine ratios (13), and the tobacco industry has been praised for a steady decline in these ratios (14). The reduction in these ratios for some brands, however, may be due to a decrease in the number of puffs taken during the assay; and, as a con- sequence, the value of the tar : nicotine ratio as an index of less-hazardous ciga- rettes may be more limited than has been supposed (15). It is important to be able to compare meaningfully and fairly the tar and nicotine deliveries of cigarettes from brand to brand and through the years. The present lack of attention to number of puffs per cigarette seems to violate the spirit of the assay, even though it strictly adheres to its letter~ Since smoking machines automatically count puffs, the number of puffs taken per cigarette could be published (and list- ed on packets) along with the tar.and nic- otine deliveries. This revision would im- prove data for health researchers, pol- !cY-~akers, and smokers (16). LYNN T. KOZLOWSKI Clinical Institute, Addiction Research Foundation, Toronto, Ontario, Canada M5S, 2S1 W. S. RICKERT. J. C. ROBINSON Labstat Incorporated, Kitchener, Ontario N2CIL3 NEIL E. GRUNBERG Department of Medical Psychology, Uniformed Services University of the Health Sciences, Bethesda, Maryland 20014 References and Notes I. "Report of 'tar' and nicotine content of the smoke of 176 varieties of cigarettes" (Federal Ti'ade Commission, Washington, D.C., 1979); "Tar, nicotine levels show some substantial re- ductions" (news release, Health and Welfare Canada, Ottawa, 29 March 1974), 2. FTC defines tar as total particulate matter (TPM) minus water minus nicotine; the Cana- dian Cigarette Testing Laboratory has used the FTC definition since 1972 but originally defined tar as TPM minus nicotine. In this report, "tar" or "wet tar" will refer to the original Canadian definition. 3. H. Ashton and D. W. Watson, Br. Med. J. 1970- ill, 679 (1970); S. J. Green, in Smoking Behaviors. R. E. Thornton, Ed. (Churchill Livingstone, Edinburgh, 1978), p. 380; L. T. Kozlowski, M. E. Jarvik. E. Gritz, Clin. Pharmacol. Ther. 17, 93 (1975); S. R. Sutton et al., ibid. 24 (No. 4), 395 (1978). 4. The newer, extremely low-tar cigarettes (< 5 mg tar) are not considered in this stud~', though they have contributed to th~ decline m overall tar deliveries. 5. The butt length is 23 mm in the United States and 30 mm in Canada; if the length of the filter plus its overwrap is greater than these figures (and in many brands it is), 3 mm is added to the butt length. 6, H. Pillsbury, Federal Trade Commission, per- sonal communication. 7. Belvedere king size filter, Cameo king size filter, Craven A king size filter, Du Maurier king size filter and regular filter, Export A regular filter, Mark Ten king size filter and regular filter, Num- ber 7 king size filter, Peter Jackson king size fil- ter. Players regular filter, Rothmans king size fil- ter. 8. The Canadian Cigarette Testing Laboratory was moved from the Department of Statistics, Uni- versity of Waterloo, in October 1977 to Labstat Incorporated, a private corporation. 9. R.. M. Wiley and J. G, Wickham, Tob. Sci, 18, 67 (1974). 10. E. H. Keith and J. R. Newsome. ibM. 2, 14 (1958). 11. J. E. Henningfield and R. R. Griffiths, Behav. Res. Methods lnstrum, II (No. 6), 538 (1979), 12 G. B. Gori and C. J, Lynch, J. Am. Med. Assoc. 240, 1255 (1978); E. C. Hammond and L. Garfin- kel, Environ, Res. 12, 263 (1976). 13. M. A, H. Russell, Br. Med. J. 1976-1, 1430 (1976). 14. R. Stepney, Lancet 1979-11,422 (1979). 15. Tar yields change more rapidly than nicotine yields do as a function of puffs (9). 16. Smokers might like to calculate the "'unit-price'" (price per pun of their cigarettes, 17. We thank M. Pope, R. C. Frecker, D. Zilm, A. Wilkinson, and K. Wagner for advice and assist- auce. 4 July 1980 1551 1980

2063628259

A SAFE CIGARETTE? Edited by GIO B. GORI National Cancer institute FRED G. BOCK Roswell Park Memorial Institute ~I.~SRSITY OF NORTH C~L~OLI]fJ JUL 16 1980 HEALTH SCLE~CES LIBRARY /37 /~ COLD SPRING HARBOR LABORATORY 1980

~ 11~ ................... i IIlI II IIIIIII~I]Ill]TI I II I Diminished Smoking, Withdrawal Symptoms, and Cessation: A Cautionary Note SAUL M. SHIFFMAN Neuropsychiatric Institute School of Medicine and Department of Psychology University of California Los Angeles, California 90024 We have heard a great deal from biologists and epidemiologists about the benefits of what I shall call diminished smoking, or any procedure that reduces the final chemical delivery or biological absorption due to smoking. This includes both reduction in the numbers of cigarettes smoked and reduction in the delivery of each cigarette. I use this term to make explicit the analogy between these two procedures because much of the literature I will be review- ing concerns the first and this meeting focuses primarily on the second. It falls to the behavioral scientists to voice the cautions and reservations regarding the diminished approach to risk reduction in smoking. My task in this discussion, therefore, will be to sound these cautionary notes, focusing on the concerns--some of them only suggested by hard data--that are raised by behavioral research on diminished smoking. The first concern is that a number of studies, including some from our laboratory (e.g., Kozlowski et al. 1975; Goldfarb et al. 1976; Gritz et al. 1976; Jarvik et al. 1978), have shown that smokers compensate for reduced nicotine delivery by increasing their smoking. Although this titration is embarrassingly imperfect and incomplete to those of us trying to establish nicotine as the reinforcer in smoking, it may nevertheless mitigate the beneficial effects of smoking weaker cigarettes. WITHDRAWAL SYMPTOMS Let us assume for a moment that smokers do not compensate for reduced nicotine delivery in cigarettes. Does this mean that diminished smoking is a procedure without cost, that we are getting something for nothing? I think not. To the extent that smokers do not compensate by smoking more, they may pay the cost in the currency of withdrawal symptoms. It is by now well-documented and accepted that smokers in abstinence experience a variety of symptoms of withdrawal (see Shiffman 1979). Some of these symptoms have been assessed objectively. They include:

284 / S.M. Shiffman 1. changes in EEG patterns (Ulett and Itil 1969; Knott and Venables 1977); 2. decrements in psychomotor performance, such as a driving simulation task (Heimstra et al. 1967); 3. changes in cardiovascular and neuroendocfine functioning (Murphee and Schultz 1968; Myrsten et al. 1977); 4. increases in physical symptoms such as headache and nausea (Guilford 1966; Shiffman and Jarvik 1976); 5. weight gain, which is a function of both increased calorie intake and metabolic changes (Lincoln 1969; Glauser et al. 1970). The changes that are most crucial, however, are the subjective effects, such as increased anxiety and irritability, sleep disturbances and disturbances of sub- jective activation levels, disturbances of attention, and sharp increases in craving for cigarettes (Trahair 1967; Schechter and Rand 1974; Shiffman and Jarvik 1976). Although these effects have typically been associated with complete ces- sation of smoking, they are also characteristic of reductions in smoking. This effect is evident in the early work by Finnegan et al., who performed the first titration experiment in 1945. They had subjects smoke cigarettes delivering 1.96 mg nicotine and then switched them to a 0.3 mg nicotine cigarette. They then recorded, rather crudely, the presence and severity of withdrawal symp- toms. Table 1 shows the relationship between a smoker's degree of compensa- tion and his experience of symptoms. There is a significant progression of compensation as we move toward the more symptomatic groups. Thus, those who did not compensate suffered more severe withdrawal symptoms. Table 1 Withdrawal Symptoms Following Reduction in Nicotine Content of Cigarettes Response to change Average daily consumption 1.96 mg nicotine .34 mg nicotine Low/high ratio Group I: Did not miss nicotine 26.6 30.9 1.19 (N = 6) Group II: Mild dissatisfaction 22.0 26.5 1.20 (N = 6) Group III: Definite, temporary dissatisfaction 28.3 28.6 1.0I (N = 3) Group IV: Definite, prolonged dissatisfaction 24.7 24.6 1.00 (N = 9) Data from Firmegan et ai. (1945). One-tailed t-test between groups I and II, and IIi and IV yieldst = 1.75, df = ll,p < .05. With diminishe( (Shiffman smokers I: turkey; th subjects c map the ~ marize ot and very group sh, symptoms the first. We ~ reduction abstinenc, the synd~ smokers. THE Wll D. Perli recently 1977; Sc smokers smokers MEAN RATING Figure Craving of withd

............. ,,~m-,~ i I - 1 III IIIIlllII IIIlllllll ,ables 1977); simulation task (Murphee and ,usea (Guilford 3rie intake and effects, such as rbances of sub- ~ increases in -; Shiffman and complete ces- smoking. This brmed the first .'ttes delivering zigarette. They hdrawal syrup- .' o~l~apensa- Pr(~l~ion of ~. I'Ilus, those ms. Diminished Smoking / 285 With more refined measures, we have also demonstrated the effects of diminished smoking on withdrawal under natural conditions. In one study (Shiffman and Jarvik 1976), we collected withdrawal data from two groups of smokers participating in a smoking cessation clinic. One group had quit cold turkey; the other had cut down their smoking by an average of 60%. All subjects completed a withdrawal questionnaire four times daily, allowing us to map the early course of the withdrawal syndrome. Figures 1, 2, and 3 sum- marize our findings. The group that abstained totally experienced significant and very substantial decreases in their symptoms, and the reduced smoking group shows no such reduction in craving and a lesser reduction in other symptoms. They appear nearly as deprived on the thirteenth day as they did on the first. We can draw two tentative conclusions from these results. One is that reduction of the smoker's habitual dose is sufficient to precipitate a full-blown abstinence syndrome equal to that seen in total abstinence. The second is that the syndrome so precipitated may be maintained at these high levels in reduced smokers. THE WITHDRAWAL SYNDROME IN RESTRAINED SMOKERS D. Perlick, working in S. Schachter's laboratory at Columbia University, recently reported an experiment confirming both of these conclusions (Perlick 1977; Schachter 1978). She studied withdrawal symptoms in a group of normal smokers and in a group identified as restrained smokers. The latter were smokers who had, of their own volition and out of concern for their health, t of Cigarettes :ion Low/high ratio 1.19 1.20 1.01 • and 111 and IV MEAN RATING 5.6 3.2 2.B Figure 1 DAYS IN ABSTINENCE Craving for cigarettes reported by partially (X) and totally ((3) abstinent smokers over the first 2 weeks of withdrawal. The group by time interaction is significant (Shiffman 19791.

286 / S.M. Shiffman 2,9 MEAN 2.8 x x\ ~ .x RATING 2,?' 26 2.5 DAYS IN ABSTINENCE Figure 2 Physical symptoms (e.g., nausea, headache) reported by partially (X) and totally (©) abstinent smokers over the first 2 weeks of withdrawal. The group by time interaction is significant (Shiffman 1979). reduced their smoking by an average of 50%, Usually, this involved a com- bined regimen of cutting down on the number of cigarettes smoked and switching to weaker brands. Essentially, these smokers were following exactly the risk-reduction strategy that many of us might recommend. Furthermore, they were successful at it in the sense that they had maintained this regimen for an average of 1 year, apparently escaping the pitfall of compensation. They are, thus, good test cases for studying the costs of diminished smoking. MEAN RATING DAYS IN ABSTINENCE Figure 3 Psychological symptoms (e.g., anxiety, irritability) reported by partially (X) and tota/ly (©) abstinent smokers over the first 2 weeks of withdrawal. The group by time interaction is significant (Shiffmaa 1979). Figure normal smo! comparable ers respond of diminisht the earliest to a chronic no relief. C minimize th Anoth¢ smokers gix cigarette, tl permitted tt those prod.'u that dimini: syndrome ,~ diminished withdrawal These from folio reduction x cigarettes ~ ever, sugg considered The 13 the proces~ 200 Figure 4 The effects smokers (Pe

~) abstinent significant ~d a com- toked and ag exactly re, on. They ) abstinent fignificant Diminished Smoking / 287 Figure 4 shows the response of the restrained smokers and deprived normal smokers to a stressor--simulated airplane takeoff noise. (Results were comparable on other measures.) While they are smoking, the restrained smok- ers respond exactly like the acutely deprived normal smokers. Even after a year of diminished smoking, these subjects show deprivation effects comparable to the earliest and most severe phases of withdrawal. They are, in effect, subject to a chronic withdrawal syndrome from which their diminished smoking offers no relief. One wonders whether they are likely to turn to other substances to minimize their dysphoria. Another finding evident in Figure 4 concerns the behavior of the normal smokers given a cigarette delivering 0.3 mg nicotine. When smoking this weak cigarette, they respond nearly as adversely as they do when they are not permitted to smoke at all. This acute restraint thus produces effects equal to those produced by acute total withdrawal. In sum, Perlick's findings confirm that diminished smoking is capable of eliciting an undiminished withdrawal syndrome and that the syndrome can be maintained chronically. The cost of diminished smoking not paid in compensatory smoking may thus be paid in withdrawal symptoms. These withdrawal effects are also likely to play a role in deterring smokers from following a risk-reduction program of diminished smoking. Gradual reduction was long considered a promising method of weaning smokers from cigarettes without the trauma of cold turkey cessation. Outcome studies, how° ever~ suggest that abrupt cessation is more effective, and it is now generally considered the strategy of choice in smoking cessation (see Shiffman 1979). The point of interest for our purposes comes from a few reports regarding the process of gradual reduction. Several studies emphasize that gradual reduc- A B C Figure 4 The effects of nicotine deprivation on the irritability of heavy smokers, nonsmokers, and restrained smokers (Perlick 1977).

288 / S.M. Shiffman tion regimes tend to proceed successfully until they reach a hurdle at 12 cigarettes per day (Upper and Meredith 1970; Levinson et al. 1971; Shapiro et al. 1971). Not only is it exceedingly difficult to obtain further reductions past this point, but typically many of the smokers drop out of the program at this stuck point. Thus there appears to be a definable level of smoking reduction beyond which further reductions are unacceptable to many smokers. Taking my lesson from C. Lynch, I note that these studies were conducted around 1970, when cigarettes delivered 40% more nicotine than they do today (Public Health Service 1979). Recomputing the value of the reduction threshold in terms of our current cigarettes, we find a value of 16.8 cigarettes per day. This suggests that smokers will often be unwilling to smoke fewer than 16 cigarettes per day. Since this value is higher than the no-risk threshold computed by C. Lynch (this volume), it implies that many smokers will be unable to carry out such a risk reduction program. EXPECTED BENEFITS OF DIMINISHED SMOKING I want to turn now to an examination of one of the expected benefits of diminished smoking. It seems reasonable to expect that once smokers are Table 2 Studies Showing No Effect of Cigarette Consumption on Cessation Daily cigarette' consumptionb Method of Study Population' cessation mean range Bums (1969) chest disease spontaneous ,~-~ 20 < 10 - 60 Burr et al. (1974) myocardial infarct spontaneous 3: ,~ 23 <10 - 20+ Dubren (1977) treatment ~ ~- 30.4 ? Fee and Benson (1971) treatment ,~ -~ 27.5 <15 - 50+ Graham and Gibson (1971) spontaneous ? ? Graham and Gibson (1971) spontaneous ? ? Guilford (1966) treatment ? ? Guilford (1966) treatment _v =, 23.4 < 10 - 40+ Handel (1973) medical practice spontaneous 3: = 23.2 I - 30+ Keutzer (1968) treatment ,~ = 28.3 1 - 50+ Mausner (1970) chest disease spontaneous ~ ~ 36.5 10 - 60+ Pederson and Lefcoe (1976) spontaneous ~ ~ 25.5 ? Perri et al. (1977) spontaneous 3~ ~ 31.7 <20 Ross (1967) myocardial infarct treatment ~ ~34.9 1 - 50+ Trahair (1967) spontaneous ? ~ Wilhelmsen (1968) treatment • -~ 20 14 - i5 aAll populations included males and females. bAll studies measured cigarette consumption by number of cigarettes smoked except for Graham and Gibson (1971) and Guilford (1966), which also estimated depth of inhalation.

earl2 apiro et 9ns past ~ at this ,.duction nducted o today duction garettes .e fewer ~eshold will be efits of .ers are -60 - 20+ ? - 50+ ? ? ? - 40+ - 30+ - 50+ -60+ ? :20 - 50+ ? Diminished Smoking ! 289 maintained at a reduced dosage they will find it easier to give up the habit entirely. A review of the literature suggests that this may not be true. Table 2 summarizes 14 studies in which this hypothesis was tested and failed to be confirmed. These studies sample a wide range of smoking doses and cover both spontaneous cessation and smoking clinic populations. In nearly all of these studies, the measure of dose is number of cigarettes smoked, which is unfortu- nate, since it takes into account neither the strength of the cigarette nor how it is smoked. Nevertheless, these results are surprisingly negative. Since the high relapse rates among ex-smokers make initial cessation an inadequate index of outcome, a review of follow-up studies is called for. Table 3 summarizes the relevant studies. In my own work now in progress, I am collecting data from smokers who have recently relapsed to smoking. So far, 1 have found no relationship between the latency to relapse and the smoker's habitual dose. Most surprising is the study by Berglund (1969) of smokers undergoing treatment in the Five-Day Plan. In a very thorough follow-up, she found lower success rates among the lighter smokers. Generally, over follow- up periods ranging from 3 weeks to 18 months, light smokers are no more successful than heavy smokers in maintaining abstinence. The only follow-up study reporting lighter smokers more successful in long-term abstinence was done by West and colleagues (1977). However, further analysis of the tables they presented reveals that this effect is accounted for by the smokers smoking between zero and 20 cigarettes per day. This suggests that this effect, to the extent it is present, tends to act primarily within the range of light smokers. This qualification is confirmed by the studies that do support the hypothe- sis of more successful cessation among lighter smokers. These studies are summarized in Table 4. It is noteworthy that even among these positive studies, half are ambiguous in that they report an effect only with some measures or in some subsamples. Examination of these studies will thus allow us to qualify the nature of the effect. Delarue (1972) reports that, among men, there is no relationship between dose measures and cessation. Among women, there is no effect for number of cigarettes; only carboxyhemoglobin (COHb) levels, predict outcome. There are serious problems with these data however. Castleden and Cole (1979) showed that mean COHb levels of 3.5% are found after overnight abstinence and that levels of 4.5% are achieved after a single cigarette. Delarue (1972) reports mean COHb levels of only 4.0% among the female light smokers in his study, which suggests that this group might be composed of noninhaling smokers or that the measurements were taken during a period of near abstinence. In either case, this anomaly severely limits the ability of Delarue's findings to be generalized. A treatment study by Thompson and Wilson (1966) also produced a complex outcome. They found that although the number of cigarettes the smoker consumed at the time he entered treatment was related to outcome, the number of cigarettes per day consumed 3 months prior to treatment was not. These results suggest that the treatment may have been most successful for

Table 3 Follow-up Studies of Effect of Cigarette Consumption on Cessation Effect on initial follow- Daily cigarette Study cessation up Follow-up interval consumption Other measures Berglund (1969) no reverseda 2, 6, and 18 months .~ =, 22 Eisinger ( 1971) ? no ? ? Harrington (1978) no no 10 and 15 weeks "~ == 23.5 Kanzler et al. (1976) in females only no 1 i,~ years .~ = 30 Pomerleau et al. (1978) yes no 1 year .~ = 30 Shiffman (unpubl. result) ? no 3 weeks ~7 = 32.4 Tongas et al. (1976, 1978) yes no 1 year .~= 50 West et ill. (1977) no yes 5 years ~ =, 22.18 Zeidenberg et al. (1977) in males only no 1 year ? number of cigarettes, estimated degree of inhalation latency to relapse effect only, ~< 20 measured serum cotinine "Lighter smokers showed lower success rates. 99Z~gg890Z

Table 4 Studies Showing Effect of Cigarette Consumption Study Population Sex Daily cigarette consumption Method of cessation mean range Consumption measures Effects Baric et al. (1976)a pregnant Delarue (1972) Donovan (1977) pregnant pregnant Schwartz et al. (1972) pregnant Thompson and Wilson (1966) F spontaneous very light ? F treatment '~ 10 - 40+ F treatment ? 10 - 40+ M treatment ? 10 - 40+ F spontaneous 2" ~ 15.2 ? F treatment ~ ~ 15.2 >5 F spontaneous 2. ,~ 10 <1 - 20+ M and F treatment 2" --~ 29.6 ? M and F treatment 2" ~ 32 ? number of cigarettes COHb~ number of cigarettes number of cigarettes, COHb number of cigarettes number of cigarettes number of cigarettes number of cigarettes at entry to treatment number of cigarettes 3 months prior to entry yes yes no } no yes no } yes yes no Bracketed areas segregate subgroups or measures not showing effects from those showing effects within the same studies. "No statistics given. "Light smokers: COHb£ = 4%. Castleden and Cole (1979) report ~" = 4.5% after a single cigarette. 69~9~9~90~

292 / S.M. Shiffman those smokers who began tapering off prior to treatment. This implicates motivation, rather than dose per se, as the critical factor in cessation. The results of Donovan's (1977) study of pregnant women smokers can be interpreted along similar lines. Donovan found that habitual dose was a pre- dictor of successful spontaneous cessation--that is, it distinguished the women who quit without any intervention, including medical advice. In a controlled trial, half of the remaining women smokers were exposed to an intensive program of medical information and advice to change their smoking habits. Among these treated women, the number of cigarettes usually smoked had no relationship to outcome. This qualifies on the findings of Schwartz et al. (1972) and of Baric et al. (1976). Both studies report a relationship between cessation and dose based on study of spontaneous quitting in pregnant wbmen. It is also noteworthy that the mean smoking rates in these three studies are low, suggest- ing that their findings may also be limited to very light smokers. CONCLUSIONS The bulk of the evidence reviewed here indicates that, with the exception of highly motivated light smokers, light smokers are typically no more successful in achieving or maintaining abstinence than are heavy smokers. This surprising finding may be explainable in terms of the intensity of withdrawal symptoms, which also seems to be independent of habitual dose (Shiffman and Jarvik 1976; Shiffman 1979). At any rate, this finding suggests that some of the benefits expected from diminished smoking may be illusory. In sum, behavioral research on cigarette smoking suggests that diminished smoking cannot be considered as a purely benign procedure without untoward behavioral consequences. Several studies indicate that many smokers will compensate for reduced nicotine delivery. Among those who do not compen- sate, many may be expected to suffer from a protracted withdrawal syndrome. Finally, some of the anticipated salutory effects of diminished smoking on smoking cessation rates may fail to materialize. These behavioral risks ought to be weighed with the medical benefits in any proposal to promote less hazardous cigarettes. ACKNOWLEDGMENT S.M.S.'s current work is supported in part by a grant number DA-01986 from the National Institute on Drug Abuse. The author also gratefully acknowledges the aid of Joan Rauschenberger in preparing the tables and of Rivia Gately in preparing the manuscript and the continuing mentorship of Murray E. Jarvik. REFERENCES Baric, L., C. MacArthur, and M. Sherwood. 1976. A study of health education aspects of smoking in pregnancy, lnt. J. Health Ed, (suppl. 2) XIX: 1. Berglund, E.L. 1969. A Jbllow-up study of thirteen Norwegian tobacco withdrawal clinics. Bums, Burr, ,~ Castled Delaru~ 1 Donov; 1 Dubren Eis.~ ~, Fee, ~ Finneg t GlauseJ GoldfaJ Grahan Gritz, : Gull for 1 Handel Harfin~ t Heimsl Jarvik, / Kanzle 1 Keutze Knott, Kozlo~ 0 O3

.,, ......... ,~r-----11-I~ ..... - 'If l' 111 IIII ill lil III This implicates ~ation. a smokers can be dose was a pre- ished the women • In a controlled to an intensive smoking habits. • smoked had no artz et al. (1972) ,etween cessation ~¢bmen. It is also are low, suggest- _he exception of more successful • This surprising awal symptoms, man and Jarvik of the that diminished ithout untoward ~' smokers will ~Io not compen- awal syndrome. ed smoking on al risks ought to e less hazardous 3A-01986 from Y acknowledges Rivia Gately in ay E. Jarvik. cation aspects of hdrawal clinics, Diminished Smoking/293 The Five-Day Plan. Final report. Norwegian Cancer Society, Oslo. Burns, B.H. 1969. Chronic chest disease, personality, and success in stopping cigarette smoking. Br. J. Prey. Soc. Med. 23:23• Burt, A., D. lllingworth, T.R.D. Shaw, P. Thornley, P. White, and R. Turner. 1974. Stopping smoking after myocardial infarction. Lancet 1".304. Castleden, C.M. and P.V. Cole. 1979. Variations in carboxyhaemoglobin levels in smokers• Br. Med. J. 4:738. Delarue, N.C. 1972. A study in smoking withdrawal• Can. J. Public Health (Smoking and Health suppl.) 64(2):$5. Donovan, J.W. 1977. Randomized controlled trial of anti-smoking advice in pregnancy. Br. J. Prey. Soc. Med. 31:6. Dubren, R. 1977. Evaluation of a televised stop-smoking clinc. Public Health Rep. 92:81. Eisinger, R.A. 1971. Psychosocial predictors of smoking recidivism• J. Health Soc. Behav. 12:355. Fee, V.M. and C. Benson. 1971. Group therapy; a review of 6 years' experience in a Scottish anti-smoking clinic. Community Medicine 126:361. Finnegan, J.K., P.S. Larson, and H.B. Haag. 1945, The role of nicotine in the cigarette habit• Science 102:94. Glauser, S.C., E.M. Glauser, M.M. Reidenberg, B.F. Rusy, and R.J. Tallarida. 1970. Metabolic changes associated with the cessation of cigarette smoking• Arch• Envi- ron. Health. 20:377. Goldfarb, T., E.R. Gritz0 M.E. Jarvik, and I.P. Stolerman. 1976. Reactions to cigarettes as a function of nicotine and "tar." Clin. PharmacoL Ther. 19:767. Graham, S. and R.W. Gibson. 1971. Cessation of patterned behavior: Withdrawal from smoking• Soc. Sci. Med. 5:319. Gritz, E.R,, V. Baer-Weiss, and M.E. Jarvik. 1976. Titration of nicotine intake with full-length and half-length cigarettes. Clin. Pharmacol. Ther. 20:552. Guilford, J.S. 1966. Factors Related to Successful Abstinence from Smoking• Final Report. American Institutes for Research, Pittsburgh, Pennsylvania. Handel, S. 1973. Change in smoking habits in a general practice. Postgrad. Med. J. 49:679• Harrington, N. 1978. The craving factor in the treatment of smoking. Br. J. Soc. Clin. Psychol. 17:363• Heimstra, N.W., N.R. Bancroft, and A.R. Dekock. 1967. Effects of smoking upon sustained performance in simulated driving task. Ann. N.Y. Acad. Sci. 142:295. Jarvik, M,E., P. Popek, N.G. Schneider, V. Baer-Weiss, and E.R. Gritz. 1978. Can cigarette size and nicotine content influence smoking and puffing rates? Psycho- pharmacology 58:303. Kanzler, M., J.H. Jaffe, and P. Zeidenberg• 1976. Long- and short-term effectiveness of a large-scale proprietary smoking cessation program--a 4-year follow-up of Smoken- ders participants. J. Clin. PLvchol. 32:661• Keutzer, C.S. 1968. Behavior modification of smoking: The experimental inyestigation of diverse techniques. Behav. Res. Ther. 6:137. Knott, V.J. and P.H. Venables. 1977. EEG alpha correlates of nonsmokers, smokers, smoking and smoking deprivation• Psychophysiology 14:150. Kozlowski, L.T., M.E. Jarvik, and E.R. Gritz. 1975. Nicotine regulation and cigarette smoking. Clin. Pharmacol. Ther. 17:93.

294 / S.M. Shiffman Levinson, B.L., D. Shapiro, G.E. Schwartz, and B. Tursky. 1971. Smoking elimination by gradual reduction. Behav. Ther. 2:477. Lincoln, J.E. 1969. Weight gain after cessation of smoking. J. Amer. Med. Assoc. 210:1765. Mausner, J.S. 1970. Cigarette smoking among patients with respiratory disease. Am. Rev. Respir. Dis. 102:704. Murphee, H.B. and R.E. Schultz. 1968. Abstinence effects in smokers. Fed. Proc. 27:220. Myrsten, A.L., A. Elgerot, and B. Edgren. 1977. Effects of abstinence from tobacco smoking on physiological and psychological arousal levels in habitual smokers. Psychosom. Med. 39:25. Pederson, L.L. and N.M. Lefcoe. 1976. A psychological and behavioural comparison of ex-smokers and smokers. J. Chronic. Dis. 29:431. Perlick, D. 1977. "The withdrawal syndrome: Nicotine addiction and the effects of stopping smoking in heavy and light smokers." Unpublished dissertation, Columbia University, New York. Perri, M.G., C.S. Richards, and K.R. Schultheis. 1977. Behavioral self-control and smoking reduction: A study of self-initiated attempts to reduce smoking. Behav. Ther. 8:360. Pomerlau, O., D. Adkins, and M. Pertschuk. 1978. Predictors of outcome and recidivism in smoking cessation treatment. Addict. Behav. 3:65. Public Health Service. 1979. Smoking and health: A report to the Surgeon General. DHEW publication number (PHS) 79-50066. Government Printing Office, Wash- ington, D.C. Ross, C.A. 1967. Smoking withdrawal research clinics. Am. J. Public Health 57:677. Schachter, S. 1978. Pharnaacological and psychological determinants of smoking. Ann. Intern. Med. 88:104. Schechter, M.D. and M.J. Rand. 1974. Effect of acute deprivation of smoking on aggression and hostility. Psychopharmacologia 34:19. Schwartz, D., J. Goujard, M. Kaminski, and C. Rumeau-Rouquette. 1972. Smoking and pregnancy. Results of a prospective study of 6,989 women. Rev. Europ. Etudes. Clin. Biol. XVII:867. Shapiro, D., B. Thursky, G.E. Schwartz, and S.R. Shnidman. 1971. Smoking on cue: A behavioural approach to smoking reduction. J. Health Soc. Behav. 12:108. Shiffman, S.M. 1979. The tobacco withdrawal syndrome. In Cigarette smoking as a dependence process (ed. N.A. Krasnegor), p. 158. DHEW publication number (ADN) 79-800. Alcohol, Drug Abuse, and Mental Health Administration, Rockville, Maryland. Shiffman, S.M. and M.E. Jarvik. 1976. Smoking withdrawal symptoms in two weeks of abstinence. Psychopharmacology 50:35. Thompson, D.S. and T.R. Wilson. 1966. Discontinuance of cigarette smoking: "'Natural" and with "therapy." J. Amer. Med. Assoc. 196:1048. Tongas, P., S. Goodkind, and J. Patterson. 1976. An investigation of the effects of post-treatment maintenance on the cessation of smoking within four behavioral treatment modalities. Paper presented at the Western Psychological Association Convention, Los Angeles. Tongas, P., J.L. Patterson, and S.D. Goodkind. 1978. Discrimination of long-term successful and unsuccessful quitters. Paper presented at the 86th annual convention Tra Ule Upl We: COt SHIt BAT SHIF BAT SHIF

,g elimination Med. Assoc. ~_~. Am. Rev. ~. Fed. Proc. from tobacco itual smokers. comparison of the effects of ion, Columbia .f-control and ~king. Behav. ,nd recidivism con General. ash- i77. noking. Ann. smoking on Smoking and ;rop. Etudes. ing on cue: A Z:108. rooking as a alion number ~ministration, two weeks of g: "Natural" he effects of tr behavioral Association of long-term d convention Dtminished Smoking / 295 of the American Psychological Association, Toronto, Canada. Trahair, R.C.S. 1967. Giving up cigarettes: 222 case studies. Med. J. Aust. 1:929. Ulett, J.A. and T.M. Itil. 1969. Quantitative electroencephalogram in smoking and smoking deprivation. Science 164:969. Upper, D. and L. Meredith. 1970. A stimulus control approach to the modification of smoking behavior. In Proceedings of the 78th Annual Convention of the American Psychological Association, p. 739. American Psychological Association, Washing- ton, D.C. West, D.W., S.G.M. Swanson, and G. Wilkinson. 1977. Five year follow-up of a smoking withdrawal clinic population. Am. J. Public Health. 67:536. Wilhelmsen, L. 1968. One year's experience in an anti-smoking clinic. Scand. J. Respir. Dis. 49:251. Zeidenberg, P., J.H. Jaffe, M. Kanzler,, M.D. Levitt, J.J. Langone, and H. Van Vunakis. 1977. Nicotine: Continine levels in blood during cessation of smoking. Compr. Psychiatry. 18:93. COMMENTS HOFFMANN: Behavioral science is new to me. Please explain one thing. Earlier, Dr. Hammond presented data based on the tar and nicotine delivery. He did not relate them to the number of cigarettes. However, your tables are all based on the number of cigarettes. Isn't there some behavioral difference on the number of cigarettes you smoke? Do you understand what I'm talking about? SHIFFMAN: Yes. It's hard for me to comment on this point. My impression also was that the effect was strongest among the very light smokers, and this is congruent with the data I have presented. Dr. Hammond suggested that it had to do with the number of cigarettes as well. In terms of summarizing the variable of dose, I think there's at least good reason to suspect that the number of cigarettes and total nicotine and tar delivery are similar, and I think we ought to raise the question about whether cessation is easier from weaker cigarettes. BATTISTA: What about when you use data where you provide carbon monoxide levels as a measure of dose? That number, if you don't consider when the sample was taken in reference to the smoking history of that individual, would be absolutely worthless. SHIFFMAN: I agree. That's precisely the trouble with such a study. There's no indication of when the sample was taken. It was probably taken in the morning. BATTISTA: Four percent could be low; it could be high, depending on the time from last cigarette smoked. SH1FFMAN: Well, given that after overnight abstinence the average level is something like 3.5, then it's hard for me to see how that could be high. 0

296 ! S.M. Shiffman RUSSELL: If a smoker hasn't been smoking for 12 hours, the COHb drops to levels indistinguishable from those of nonsmokers. CAIN: The term "craving" is something we all know, but I've often won- dered how to characterize the craving that occurs when you are smoking less than you want to smoke, when you've abstained, and so on. It presumably has a feeling of some sort. But has anyone tried to charac- terize it? SHIFFMAN: Well, I think Dr. Battista's joking comment "talk to him, talk to a smoker" leads us in the right direction. A lot of our research involves self report and, more recently, relatively open-ended interviews with people in abstinence--those smokers describe it as a feeling. They also do describe it as almost a physical sensation analogous to hunger. But that's about as much as I can tell you. I don't know how else to characterize it. The Cig MICH Instit~ The M Londo The qut machin intake ~ such as know 1 nicotin~ Wynde machin form o harmfu have n machin measur of ciga~ In nicotin, the cig epidem filter-ti biases quality in the nicotin more admitt~ die to ', ACCEI At the

20~3628275

A SAFE CIGARETTE? Edited by GIO B. GORI National Cancer Institute FRED G. BOCK Roswell Park Memorial Institute ~I.~F_~SITY OF NORTH JUL ] 6 1980 HEALTH ~CLE~CES LIBRARY ~V /37 /~ COLD SPRING HARBOR LABORATORY 1980

~ to see that this , the fluctuations y--well, for the ,~¢ smokers, but iother, And the .de this attempt. neetings every 4 .ustriaas, nor the o consensus, but Public Policy Issues in the Promotion of Less Hazardous Cigarettes JEFFREY E. HARRIS Department of Economics Massachusetts Institute of Technology Cambridge, Massachusetts 02139 From 1967 to 1977, the domestic market share of cigarettes with FTC tar levels of 15 mg or less increased from 2-23%. The proportion of manufac- turers' advertising and promotional expenditures devoted to these cigarettes increased from 5-49% (Federal Trade Commission 1978). By 1978, 23% of adult male cigarette smokers and 35% of adult female cigarette smokers regu- larly consumed brands with FTC tar levels of 14 mg or less (National Center for Health Statistics 1979). This discussion briefly addresses two basic questions. Should we continue to promote these trends in cigarette consumption? If so, what public policy interventions are at our disposal? IS PROMOTION OF THE LESS HAZARDOUS CIGARETTE AN APPROPRIATE PUBLIC POLICY? The promotion of lower-tar and nicotine cigarettes would be an appropriate public policy if: 1. the population of current cigarette smokers were unchanging, with no new entrants and no quitters; 2. a smoker's shifting to a lower-tar and nicotine cigarette did not adversely affect his or her style of smoking or the number of cigarettes smoked; 3. the dose-response relation between cigarette tar and nicotine delivery and cigarette-induced health damage were uniform across the population. When one of these suppositions is violated, however, the value of a policy promoting less hazardous cigarettes is not so clear. Initiation of Smoking The progressive decline of the tar and nicotine contents of cigarettes over the past 25 years may have made it easier for teenagers and young adults--

334 / J.E. Harris particularly young women--to experiment with and later become habituated to cigarettes. This possibility is particularly important in view of the marked increase in smoking among teenage women in recent years. Data presented in Table 1 indicate that lower-tar cigarettes have infiltrated the teenage smoking population and that high-tar, nonfilter cigarettes are vir- tually absent. (Most "of the teenagers reporting a brand with FTC tar over 20 mg smoked a relatively new, 120 mm filter-tipped brand.) The marked preva- lence of low-tar smoking among older teenage females is consistent with independent survey data on young women ages 17-24 reported from the Health Interview Survey (National Center for Health Statistics 1979). Although these data are not conclusive, I know of no evidence contradicting the hypoth- esis that the availability of lower-tar and nicotine cigarettes enhanced the rate of initiation of smoking among younger females. Cessation of Smoking Earlier in this meeting, Lawrence Garfinkel cited data from the American Cancer Society's 25-state study showing that smokers of lower-tar and nicotine cigarettes in 1959 had a higher probability of reporting nonsmoking status in 1972 (Hammond et al. 1976). However, cigarettes regarded as low in tar and nicotine during this period do not represent current products. Whether smokers Table 1 Distribution of Cigarette Brands of Current Regular Teenage Smokers According to FTC Tar Content, 1978 Sex and age Fraction of current smokers according to FTC tar * Overall prevalence of current 15 mg or less 16-'20 mg 21 mg or greater smoking (%) (%) (%) (%) Males 12-14 years 23 "77 0 3 15-16 years 38 62 0 14 17- 19 years 30 69 1 23 12-19 years 32 68 1 13 Females 12- 14 years 26 70 4 4 15- 16 years 30 70 0 12 17-19 years 40 57 3 28 12- 19 years 37 61 3 15 Data from U.S. National Institute of Education telephone survey in 1978 of over 3000 teenagers. Respondents reported their current brand, and not the brand of the very first few cigarettes smoked. aExcludes current smokers who did not specify brand and type. Percentages may not add up to 100 due to roundoff. quit 1- 1964 wom. pensi Staff., cigar more lowel the h, Smo The incr~ smo! cons 197( wer( Cen smo (Ha 22 [ que incr inc~ sm( stu( low fret the. eric sm~ abl aer pol fer tra m~ let ci~

IIII I I .... IF--I~iliJlii ~ ome habituated to w of the marked es have infiltrated cigarettes are vir- FTC tar over 20 he marked preva- s consistent with eported from the 1979). Although ~cting the hypoth- hanced the rate of m the American r-tar and nicotine .moking status in as low in tar and Whether smokers )retail prevalence of current smoking (%) 3 14 23 13 4 12 28 15 1978 of over 3000 the very first few s may not add up to Public Policy Issues / 335 of the new cigarettes with even lower tar and nicotine are more or less likely to quit has not been determined. Although a significant percentage of adult males have quit smoking since 1964, the rate of quitting among adult women has been less impressive. Since women currently aged 45 years and over have a disproportionately high pro- pensity to smoke lower-tar and nicotine cigarettes (National Center for Health Statistics 1979), it is possible that the availability of lower-tar and nicotine cigarettes has served as an alternative, thus deterring quitting in this group. The more intriguing possibility is that the increased publicity and availability of lower-tar and nicotine cigarettes has actually enhanced the public perception of the health risks of smoking. Smoking Frequency The continued decrease in cigarette tar and nicotine has been associated with an increase in the average number of cigarettes smoked per day among current smokers (Harris 1979). The percentage of adult male current smokers who consumed 25 or more cigarettes per day increased from 25% in 1965 to 28% in 1970 to 34% in 1978. The corresponding proportions for adult female smokers were 14% in 1965, 18% in 1970, and 21% in 1978 (Harris 1979; National Center for Health Statistics 1979). Using Gallup Poll data on the percentage of smokers and U.S. Department of Agriculture data on aggregate consumption (Harris 1979), I calculate that the average smoking frequency increased from 22 per day in 1954 to 30 per day in 1978. Possible explanations for this observed increase in average smoking fre- quency include a higher rate of quitting among lower frequency smokers; an increase in smoking frequency of those who continued to smoke; and, an increased frequency of smoking among new entrants into the population of smokers. Garfinkel's data for 1959-72, from the American Cancer Society study, tend to support the explanation that there is a higher quitting rate among lower frequency smokers. Yet, it tends to negate the explanation that smoking frequency increased among continued smokers (Garfinkel 1979). However, these inferences are weakened by possible underreporting bias and digit prefer- ence artifacts (Warner 1978). Moreover, these data say little about changes in smoking frequency among those who are now switching to previously unavail- able cigarettes, with considerably lower tar and nicotine, and different filter aeration, paper porosity, tobacco density, air resistance, and flavor. The re- ported increase in current smoking frequency among new smokers, particularly females, emphasizes that there is an increased frequency of smoking among new smokers (Harris 1979). In this respect, there is little epidemiological information concerning the tradeoff between smoking a few higher-tar and nicotine cigarettes and smoking many lower-tar and nicotine cigarettes. The findings of Hammond and col- leagues (1976) suggest that smoking a large number of low-tar, low-nicotine cigarettes may be more damaging.

336 / J.E. Harris Changes in the Style of