Tobacco Documents Online

Impact of Long-Term Filter Cigarette Usage on Lung and Larynx ~ Cancer Risk: A Case-Control Study 1, 2 Ernst L. Wynder, M.D., I and Steven D. Stellman, Ph.D. ABSTRACT-A case-control study was conducted among 1,034 white male and female hospital patients with histologically proved lung cancer (Kreyberg type I) or larynx cancer. After adjustment for duration ol the smoking habit, inhalation, and butt length, relative risks of developing lung or larynx cancer were consis- tently lower among long-term smokers of filter cigarettes than among smokers o1 nonfilter cigarettes, irrespective of quantity smoked. Relative risks in alt groups declined with increased years of smoking cessation. The observed risk reduction among cur- rent smokers ol filter cigarettes was consistent with that e.- pected, considering that these persons had smoked the older high-tar nonfiller cigarettes for a large proportion of their lives.- J Nall Cancer Inst 62: 471-477, 1979. Although ctgarette smoking is recognized as the major causative factor of lung and larynx cancer in both men and women (1), elimination of cigarette smoking does not at present appear to be a social possibility, Therefore, :ts Gori (2) has pointed out, alternative strategies must be sought as a humane but interim approach to the cornplex problem nf preven- ting tobac(o-related disease. "I-wo such strategies con- tinue to be explored widely: a) persuading segments of the population to reduce their consumption of ciga- rettes rettes or to quit altogether (behaeioral), and 6) modify- ing the cigarette itself to produce a"less harmful ciga- rette" (managerial). The effects of both approaches on the risk of developing tobacco-related cancer have been intensively studied for a number of years. In this paper we examine the impact that the less harmful cigarette has thus far had on the risk of developing lung or larynx cancer, and we attempt to predict the future disease patterns that may emerge from continued appli- cation of the managerial approach to cancer preven- tion. The rationale underlying this assessment is the dose- response effect obserced in both prospective and retro- spective studies; the greater the exposure (as n)easured, e.g., by the average number of cigarettes smoked per day or by duration of the smoking habit), the greater the risk of lung or other tobacco-related cant'er.s (J-6). Thus as exposure decreases, either by smoking fewer cigarettes or by smoking cigarettes significantly lower in yield of tmnorigenic components, a proportiomttely lower risk of developing tobacco-related cancers should be expected. In this regard, two major events have occurred over the past 20 years: Consumption of cigarettes with lower tars has increased (7-10), and the tumorigenic potential of tar has dea'eased (10-12). Earlier studies have shown that a lower lung and larynx cancer risk exists among long-term smokers of filter cigarettes than among smokers of nonfilter brands (13-16). We provide further ~evidence, encouraging in light of the growing popu- larity of lower tar cigarettes. MATERIALS AND METHODS The data corutilute a portion of that collected in an ongoing rett<rsprtrice stud}uf tobacco-related c:utcer. {Y'ith the use uf a standard questionnare, interviews were conducted from 1969 to 1976 in six U.S. cities by personnel who had undergone a detailed and uniform training period at our institute. Approximately one- third of all interviewed patients were at Memorial Hospital, and decreasingly smaller numbers eeere in carious hospitals in Houston, Los Angeles, New York, Birmingham, Miami, and New Orleans. Interviel.•ers visited J3~ of the coherent lung and larynx cancer patients admitted to these hospitals. The numbers of cancer patients and controls are shown in table 1 with their age distributions. There were 68-4 lung cancer cases and 350 larcnx cancer cases, All patients had histopathologic confirmation of cliag- nosis. For the analysis in this paper, only lireyberg type I lung cancer (squamous und oat cell types) was, considered beawne this type has been shown to exhibit Aausrvoeuuss rstn. t:rTS=long-term (ilter cigarrue .moker(si; NFS=non(iltet <igareuc smukcr(bl. ~ Receivttl \huth 29, 1976: ucceptrd September 3, 197R. ' Supported by Public Ftealth Service contract N0t-CP55666 and grant CA776t3 front the National Cmcer Institute. Computations were performcd in parl .u the C:ourant Computing and Mathematic's Ccnter (New York t!niversity), which is supported hy C.S. Deparo- ment of Energy contract EY-76-C-02-9077. r Dirision of P.pidcnriologY, Naylur Dana Insriwte for Disease Ptevention. :lmeriedn Ilr,lth Fuuodation, Y20 F.ant i9d St., \ew S'urk, N'.Y. 10017 . ' Dicision of Biostativics, Naylor Dana instiurte for Disease Pre.cntiun. s SS'e thunk Mr. H:ulaud Austin and Sts. IVaney Saphier for excel- lent scui,tYral and prugramming ussistance. We also acknowledge the mnnibwiuu. uf ihc fulluwiug institutloru :md indiciduals-New Ynrk. N.Y.'. llcntmLd (luapit:d, I)r. I(. lk'aniet \tount Sinai 1.h.vlic:d Crntrq 1)r. 9hridun 6up(er; hluohattan S'eteruns AtL»inistratiun Ho.spual, Dr. (:h:ules Rednnr and Dr. Surton Spritc Fraucis nrlutiold ILuspinil, kG. Esthvr Au,tiu; Mctrapuliwn Hnspitul, Dr. .Sigmwtde IlirscLe: St. L.ukeS Ituspital, Sts. Evelyn Peck. blinmi, Plu.: }liurni Ven•r:ms 1-lospilal arid Iprirersity of Miumi Fluspital Cetlter, Dr. George Butun. Ilirminghum, Ala.: tOticersity Ilo.pital, Di. Peter li. Pra<nck. Huustun, liM: \I.1). :lndcnun IluspiUd aud Tumur In.stittue, I)r. Rulun W. Rawsnn. Lus Angeles. C:alif.: Ilnieenity of Californiu at Los Angeles Hospital, Dr. Lester Breslow. :vew Urleans. Lt.: S'eter:ms :\dnrinisrraiion Ilospital and Charity I-lospitul, Dr. Alwn Ottuner. ° Tot:d particulate matter (PPAf) refers to the material separated (rom the gaseom portion of rig',rrerre smoke hr n(ambridgc tiher, w'hile "tar" is defined and me:uured by the L1.5. Fedefal'I7ade C:nnt- misaiun as 7PN1 Iess mutature and nicotine. N O O W O' W N ~ 00 VOL. 62. NO. 3. MARCIt 1979 471 J NA'l'l. ('.ANCER [NSr

472 Wynder and Sfellman TABLE 1.-Num6er of cancer and control patients Nho were either lifetime nonsmokers, rurrent smokers. or ex-smokers, b2j age and sex 1* Lung' cancer ti t Larynx° cancer i Controls Age, yr en s pa pat ents Male Female Male Female Male Female <50 72 29 41 15 1,533 1.617 50-59 191 46 108 22 1,402 1,301 60-69 215 32 93 24 1.318 1,104 70-89 79 20 44 3 582 690 Total 557 127 286 64 4,835 4,712 0 0 ° Kreyberg type I. ° Glottic and supraglottic types. the strongest dose response of all tobacco-related cancers (6, 14). The larynx cancer cases included both glottic and supraglottic types because both have been shown to be equally related to tobacco usage (I3). Controls were selected on the basis of absence of a history of tobacco-related disease. Tobacco-related LUNC CANCER tl(aEYOEAG TYPE L), t'Jil.ES NO OF CIGARETTES SMOKED PER DAY SMOKER 1-10 11-20 2N0 31-40 LI1KG CPNLER (KREY8F.0.C TYPE t). FCIALES 20 20 0 disease was defined as a cancer of lung, larynx, mout esophagus, or bladder [which comprised the cases i our prerious report (6)1; cancer of the pancreas, licl or kidney; myocardial infarction; stroke. ~Wriphll vascular disease, or abdominal aortic anI <hion bronchitis or chronic obstructive pultnonary diseasgastric ulcer; or cirrhosis of the liver. The diagnoses I 9,547 eligible controls were distributed as follon (numbers in parentheses are percentages for males ar females, respectively): cancer of the stomach 12. ~ colon or rectum_IZ. 5). prostate gland (6, 0), breast ( 13), cervix (0, 10), or skin, including melanoma (6, { leukemia, lymphoma, or Hodgkin's disease (7, 4); oth• cancers, e;g., cancer of the male or femalr reproducti1 organs (9, ll); benign neoplastic diseases (1I, II fractures (8, 6); or other nonneoplastic diseases, e.g burns, infections, or duodenal ulcers (14, 34). The major index of response to cigarette smokir was expressed as the relative risk, defined as the rat; of the incidence of lung or larynx cancer amor smokers to its incidence among nonsmokers. Poii LARYNF CANCER, MALES 30 25 1 nER I F NF F NF F NF SMo 10 II-20 21-.10 31-40 NO OF CIGARETTES SMOKED PER DAY LARYNX CANCER FEMALES 45 35 25 15 5 (9) ND OF CIGARETTES SMOKED PER DAY NO OF CIGARETTES SMOKED PER DAY Texr.eteoaz I.-Age-adjusted relative risk of lung or larynx cancer for LTFS and NFS by quantity smoked. Fraction=No. cases: No. contro F=filter cigarettes, NF=nonfilter cigaretres. A) Lung cancer, males; No. cases=143 LTFS. 150 NFS. B) Larynz cancer, mates; No. cases=l LTFS, 86 NFS. C) Lung cancer, females; No. cases=50 LTFS, 18 NFS. D) Larynx cancer, females; No. cascs=20 LTFS, 17 NF J NATL CANCER INST VOL. 62. NO. 3, MARCH 19

Less Harmful Cigarettes 473 ~ 9 estimates of relative risk were made by means of odds ratios; potentially confounding variables (e.g., age) were controlled by use of either the iVtantel-Haenszel method (17) or the iVliettinen confounder score method (18). Interval estimates of relative risk were made by the method of Gart (19). This study included all interviewed white patients who were either non.smokcrs (n(,vcr .smokcd any' tobac- eo product regularly), ex-smokers (quit at least 1 year prior to interview), or current cigarette smokers. Persons who had regulatly smuked only cig;us or pip(•s weit• excluded, but uot eigateue sniukers who alw used cigars or pipes. To ass«re that the known latency period for most tobacco-related cancers was exceeded, we restricted the smoker and ex-smoker groups to people who had smoked cigarettes for at least 20 years. RESULTS Relative Risks Among LTFS and NFS Age-adjusted relative risks were estimated separately [or LTFS (defined as present smokers who used filter cigarettes currently and for at least 10 years) and NFS (defined as present smokers whose current brand was nonfilter). Age categories used in the adjustment were: 20-49, 50-59, 60-ti9, and 70-89 years. Results of these calculations are displayed for both lung and larynx cancer in text-figure I. In these dose-response curves, the estimated relative risk is plotted against the quantity smoked (clefined as the average numbe•r of cigarettes of the (wrent brand smoked per day). Among both females and males, the risk for both lung and larynx cancer among LTFS was lower than that of NFS at each of the five quantity levels in which data were grouped. Table 2 shows the relative risks for LTFS, expressed as a percentage of the risk for NFS, by sex and quantity of cigarettes smoked daily. Percent reduction of risk is obtained by subtracting this quantity from 100%. Among male LTFS, reduction of risk ranged from I1 to 39'. for lung cancer and from 25 to 49 a for larynx cancer. Substantial lowering of risk was also calculated for female LTFS; the one exception was TABLE 2.-Retatire risk of lung or larynx cancer for LTFS.° as a percentage of the relative risk for NFS. ° . by sex and quantity smoked No. of Cancer site cigarettes smoked/ Lung Larynx day Male, % Female, % Male, % Female, % 1-10 61 38 1-20 - 11 11-20 89 69 5L 21-30 71 79 53 21+ 33 31-40 ' 66 75 31+ - 103 41+ 86 - 55 0 ' Defined as present smokers who have used filter cigarettes for ?10 years. ° Defined as present smokers whose current brand is nonfilter. TABLE 3. Eslimated relative risk of lung and larynx cancer among NFS, relatiae to LTFS, after adjvsting for various factors Sex Adjustment Odds ratio factors° 95% Confidence interval Lung cancer Male D, Q 1.19 (0.92-1.55) Male A, Q 1.26 (0.97-1.64) Female D, Q 1.29 (0.67-2.47) Female A, Q 1.37 (0.72-2.60) Larynx cancer Male D, Q 165 (1.16-2.34) D, Q, Ale 1.49 (1.05-2.10) Male A, Q 1.71 (1.21-2.41) Female D, Q 4.19 (2.66-6.61) D, Q, Ale 3.97 (2.04-7.70) Female A, Q 4.34 (2.27-8.31) ° By Mantel-Haenszel method, adjusted for duration (D): 20-29, 30-39, 40-49, 50-59, 60+ years; average No. of cigarettes smoked per day (Q): 1-10, 11-20, 21-30, 31-40, 41+; age (A): 20-49, 50-59, 60-69, 70-89 years; alcohol consumption (Ale): none, 1-6 oz/day, 7+ oz/day. For females, due to the smaller number of cases, the D categories 50-59 and 60+ were combined, as were the Q categories 31-40 and 41+, probably attributable to the small number of cases•in that category. ddjustment for duration and quantity smoked.-Al- though adjusted for age, the foregoing risk estimates may be influenced by additional confounding by dura- tion of smoking habit, even though age and duration are strongly correlated. This would be true, for example, if LTFS began smoking earlier in life than did NFS of the same age. Additional risk estimates were made in which both habit duration and cigarette quantity were con- trolled, and they were compared with corresponding estimates in which both_ age and quantity were con- trolled. Because we were adjusting for possible differ- ences between LTFS and NFS, the risks were calculated with LTFS (rather than nonsmokers) as the referent. Results are presented in table 3, in which the odds ratios and 95% confidence intervals are given. Adjustment for other smokireg intensity variables.- Because tar and nicotine levels of cigarettes are highly correlated (10), it has been suggested that persons who switch to cigarettes with a lower tar level may subse- quently adjust their smoking habits to compensate for the concomitantly lower nicotine level. This compen- sation may be manifested by an increase in the number of cigarettes smoked per day, deeper inhalation, or shorter butt length (indicating a greater portion of each cigarette smoked). Odds ratios were calculated for lung cancer by use of Miettinen's method (IS), in which the following variables were all controlled simultaneously: average daily quantity of cigarettes smoked and habit duration (continuous), educatiou (eight levels), inhalation (four levels), butt length (four levels), and city (six levels). The resulting risk es'timates, based on a sumntary over 10 strata of the confotmder score, were not significantly different from those reported in table 3. -- - -- - Adjustmenf for alcohol coruumption.-Although the risk for lung cancrt is greater than that for larynx VUI.. 1,2, NO. Y• MARCII 1979 J N,CrI-CANCE.k Ih:vl'

474 Wynder and Steilman 01 cancer among smokers (relati.e to nonsmokers) (6), table 3 shows that the reducion in risk observed among LTFS relative to NFS was actually greater for larynx cancer. Because larynx cancer is related to use of alcohol as well as to cigarette consumption, potential confounding by alcohol consumption was considered. Among controls, for example, NFS were twice as likely as LTFS to be heavy drinkers of alcohol (7 or more oz/day). The larynx cancer risks in table 3, after adjustment for duration of smoking, quantity smoked, and alcohol consumption, were reduced to 1.49 for men and 3.97 for women when this confounding was removed but were still significantly greater than I. Effect of Smoking Cessation on Cancer Risk Because maximum reduction in cancer risk is achieved by reducing tar intake to zero, i.e., quitting smoking altogether, relative risk estimates were made for ex- LUNG CANCER (RftEYBERG TYPE I), MALES YEARS CP SMOKING CESSATION LIING CANCER ()CREYBERG TYPE I), FEMALES 1n re( IBF 6 C N FfiESEN/ i-] u-6 )-10 111 NGN SM]M1ER SMGRER smokers in several age strata. In these calculations it was not possible to discriminate between former NFS and LTFS, because_few of the latter had quit smoking for a significant period of time. As observed previously (6, 13), relative risk declined with years of cessation. Because the reduction of risk was appreciably greater for people 50-69 years old than for those 70 or more years old, the relative risks for persons 50-69 only (stratified into two groups, 50-59 and 60-69, and reported as a combined risk estimate) were plotted against years of cessation in text-figure 2. This age group included more than two-thirds of all lung and larynx cancer patients. The relevance of smoking cessation, particularly in this age group, to use of the less harmful cigarette is assessed by two observations (text-fig. 2). First, male long-term cigarette smokers must have ceased the habit for at least 3 years before any decline in risk could be expected. The sizable increase in risk among males during the first 3 years after cessation may be attrib- LARYNX CANCER. MALES YEARS OF SMOKING CESSATION LARYNX GANCER. FEMALES 9 ;c 0 6 2 .t, ~ ~ YEARS OF SMOKING CESSATION YEARS OF SMOKING CESSATION D 'CRxr.n<'.rue 2.-Age-adusted risk of lung or hu.nx amier for ex-smokers (50-1i9 years old) by ye•ars of cessation. Fraetion=No. rases,No. cnntrols. A) Lung cancer. males; No. cases='106. B) Larynx cancer, males; No. cases=20I. C) Lung cancer. females; No. cases=7R. D) Larynx cancer, females: No. cases=46. J NA'fL CANCER INST VOL. 62. NO. 3. MARCH 1979

• utable to the failure of smokers with emerging signs of clinical illness to seek medical attention until after quitting. Second, 10-15 years of cessation are required before the long-term smoker's risk approaches ehat of a nonsmoker. Therefore, at least this much time must elapse for a long-term smoker to receive maximum benefit from switching to lower tar cigarettes. This was our rationctle for restric'tiug ihe Cl-F;S group to persons who had continuously smoked (ilter cigare•ttes (or at least 10 years. DISCUSSION Whether calculated as a single summary estimate or as a dose-responx curve, the data consistently point to reductions in risk for lung cancer and even larger reductions for larynx cancer risk among LTFS. As stated previously, the maximum risk reduction that a long-term smoker can expect is achieved orrly through complete cessation of smoking. One who switches from nonfilter to filter cigarettes must necessarily experience a more modest risk reduction than an ex-smoker over the same time period. Average tar levels of filter cigarettes hace been abom two-thirds those of nonfilter cigarettes for over a decade, and the tar levels for both have declined roughly in parallel; thus a?0-80°0 lowering in risk may reasonably be expected at the present time and was about what was observed. The following is a diac'ussion of current and future trends in cigarette c'omposition and smoking behavior that may soon influence the expected patterns of tobacco-related cancers. Cigarette Variables Changes in tar levels mitk time.-Wlren the first major epidemiologic studies linking lung cancer and cigarette smoking appeared in 1950 (5), the average cigarette yielded about 10 mg tar. Since that time, tar levels have fallen considerably. Text-figure 3 demon- strates that tar levels of both filter and nonfilter Less Harmful Cigarettes 475 cigarettes continue to fall,.and that even today's non- filter cigarette produces one-third less tar dran the nonfilter cigarette of a generation ago (20). This reduction in tar has occurred not only in the United States but throughout most of the world (10-12, ?1-2a), as shown in studies by Kuhn and Klus (25) in Austria, Todd in England (26), Ledez in France (personal comnumic.uiun), Tinun in Germany (27), and Hoff- mann iu Canada (unpublished observations). Changes in tar aclit~ity.-Chemical and biological experimeuts have• ittdicatt•d that tars of present-day cigarettes havc• lower carcinogenicity per gram than tars of 25 years ago (!, 10, 11). This lowered activity, coupled with reduced toud yield per cigarette, has undoubtedly t:ontribtned to a reduction in cancer risk in some populations, as observed in this and other studies (24). Smoking Variables Changes in filter cigarette consuntption,-h is not sufficient that some of the .nailable cigarettes become potentiallv less harmful in order to bring about a measurable lowering of disease incidence. Smokers rnust ako switch to those cigarettes in preference to higher tar brmds. This appear. to be the trend. By 1976 nearly 90" of all cigarettes sold in the United Stutes w•ere filter brands (20, 34). The past 5 years have witnessed ihe e•mergence of low-tar (<15 mg) cigarettes, which ha~e capwred more than 16% of the market. This trend is illustrated in text-figure d in which the market share of all cigarettes yielding less than 15 tng tar is plotted for the 10-year period 1967-76. The growing usage of these cigarettes will probably have a continuing effect on reducing the average daily tar intake and concomitant disease risk. Changes an lifetime tar consumption•-Our studies have not yet included persons who have smoked lower tar cigarettes exclusively. Text-figure 5 shows a model calculation of the proportion of a person's lifetime _ EAR Texl'-F¢.t•ur Y.-S:Jt•.nrci4luert :n•erage wr delio-ery of U.S fitu•r and nonfilter rigarettrv, 1957-78. PfC=Federul Tr•tde Conmussion. J NA'rL CANCER INST VOt-. 62, NO. tl, MANCn 1979

0 0 • 476 Wynder and Stetlman 16 5 4-i I 3-V 5-i 3-i 21 67 68 69 70 71 72 73 74 75 76 YEAR TexrFtot'Re {.-Percent m:rrket share for low-tar cigareues, 1967-76. spent smoking filter cigarettes, on the assumption that a typical smoker began at age 20 and switched to filter cigarettes in 1960 when these brands achieved wide- spread popularity. About two-thirds of lung cancer patients currently fall between ages 50 and 69 and, as text-figure 5 indicates, could have used filter cigarettes for no more than 50% of their total smoking years. Considering that the average lung or larynx cancer patient is between 56 and 60 years old and has smoked cigarettes for an average of 40-50 years, the risk for developing a smoking-related cancer today probably depends on a history that included both old and new types of cigarettes. The relative risks should be even lower in the future among persons who will have smoked filter cigarettes all their lives. Changes in smoking behavior.-Because tar and nicotine levels in cigarette smoke are strongly correlated (10), the simultaneous reduction of these two compo- nents might be compensated behaviorally by increased exposqre, e.g., deeper inhalation, more frequent puf- fing, and leaving shorter butts. Some short-term studies have demonstrated that all three methods can be used by the same smoker to adjust to the new lower tar J NATL CANCER INST cigarette [(10, 28-30); Hill P: Unpublished obsenations]. However, long-term epidemiologic studies, including this one, have not shown differences between LTFS and NFS, either among lung or larynx cancer patients or controls, in either the number of cigarettes smoked or subjective inhalation response (though the lung and larynx cancer patients as a group invariably consumed more cigarettes than did the controls) (6, 28). In this study, adjustment for inhalation and butt length did not bring about significant changes in risk estimates. Objective comparisons of levels of nicotine, its major metabolite cotinine, and carboxyhemoglobin in the sera of long-term smokers are clearly needed and are now in progress in our laboratory. They may help to resolve this behavioral issue by determining whether long-term compensation for a low-nicotine cigarette might lead to greater tar consumption. Alternative Strategies Although development of a less harmful cigarette has apparently had some success in lowering the risk of lung and larynx cancer, it must not be considered a complete prescription for the prevention of these and other tobacco-related diseases. Additional approaches are needed, particularly preventive education of young presmokers and cessation programs for curreni smokers. At present, antismoking education is making little headway with our youth. The number of boys who begin smoking has remained constant over 'the last decade, and the number of girls taking up the habit has doubled (31). New approaches to health education of the young are needed, such as the Know Your Body (KYB) program initiated-by our institute, whereby each child is involved in health promotion programs (32, 33). Antismoking propaganda among adults has been more successful for men than for women; among the 70 I I u z Y 0 s O 60 50 40 0 FILTER NON-FILTER 25% 38 0 50 1895 1905 1915 1925 YEAR OF BIRTH TexLFIGt'RE 5.-Filter cigarette usage as a percentage of total smoking experience, by birth cohon. VOL. 62. NO. 3, MARCH t979

0 0 more educated groups of men a significant percentage have been able to stop smoking on their own (9). Smoking cessation clinics have also had some success (39, 35). Nevertheless, more than -10 million adults con- finue to smoke in spite of all (•vidence presented to them of its adverse effects. As long as society condones smoking, }•oung people will take up the habit, and many adults Will rontinu(!. Obviously, bet[er auli- smoking propaganda, improved smoking cessation pro- grams, and development of even less h:u'mful rir;:3reaes are needed. REFERENCES I!) U.S, Dep:trtment of Ile:tlth, Hdut':uion, and \t'elfure: Smoking and Hr:dth. Report uf the Adcisory C:ommiuee to the Surg•ou Cruerul Of the Publi<' Ile:rlth Serriw. Puhlir Efe:rith Srrv liull No. 1103, lt'ushiugtnn, D.C.', U.S. Govl Print Off, 19tii (?) GuRt GB: Lotc-risk ri,K:nrues: A presrripti,rn. S<irn<r 191:1fN3- 1°_17, 1976 (3) d R.uato.cU E(:: .Smoking in ieldlion m deuth r,n<s of one• ntillion mvu and women. V:uI C:rn<er Inst Atonngr 19:I'L9-20i. 19fi(i ({) L'S. Department of Iir:dtit, Eduratinn, aud A'elf:ue: 1'Ite IIc:d11r Cun1e(jucn<'e5of Stnokiug. CD(I) 7~1-N7&1. I'.S Cotc Print DIf, 1971 (5) \1'sanrR I.L. (:It.tii.tsl E.1: Ibha<ro ~ntokin,g as it posOhlc otfoluGic LrcUor in bronrhiogeni< carciuom:t. Jd)[.l 1i3.329- Y7ti, 1950 I Lt'tmntu I.1., Slrt.ral.t~ SI): Compar:uice rpidrmioIogy Of IoL,tuuarlule•d t',m<rrs, C',mrrr Res 37:I608-1022. 1977 (7) IIS. Drpaninem of lle.ihh. Nduuuiun. autl ll'clf:ur. 1'hr :\dtdi l'sr nf 1uhaun. 1975 , .\tl.mta: C:cnter for I)isr.t.r Conue,l, Iturr.iu Of fir:rhh F:durutiun. 1976 (8) l1's.vnrR hl.. }lut)\n~~ 1): l~uhauo and tob:nru smuAr.Srtnin Uneul 95-19. 1976 (9) lhsolN 4:1., Cut'ts 1.5, ,\Ltts[t,m K: C:urtont snwAing h:r6ils L)' scleurd barkKround suiahles: ['heir e[fe<t un fuune rlisrase ^rnds. :\m J k:pidrmiui I00:IIi8-177, 1971 (10) l1'tuta I:H: Retent ehangrs in wbeuu pruducn and Iheir aueplMI<r b)thr eunsmnrt. br Plaeevdings of the~ Siaih IWrr- n:niuned -IulrUeto S,ieutifit' ('ungn~ss, f'ok}'o. J,rpan, 1976. Cart•sta Infonnauun Hull (sp<•tizl):15-30, 1976 (11) \9vsmtt F.1., IiwtsuNN ^: Ibbnero and Iohano .Snmkc: Studies in Lkprrimrmal (Lueiuuyeue'..is. \'rt<\'urk::\nrdrmie Presa, 191i7 (12) : }'.sllcrimcnarl to6:rrco r:urinogenesis. Scirnce I62:802- 871, 191ik- (I3) lVrsnrx E1., (bcis I.S. Yttxtuu K. et :tl: Hmironmeul:d Luuns iu e:uurr of tht• Iarsm'. :\ sceond luuk. C1mu.r 38:151J1- I601, 1971i (1I) \\lsurx 4:L, \t%nr<.ut K. Rt.ltru EJ: -I'he epidcnriology of hwg am<rr: Roeent Irends. JAMA 213:2211_2228. 1970 (15) Bsoss If, (:rusos R: Rik of lung rnutrr in unok( .re whu>witth to fihea ei4nirnrs. Am J Publie IlrAth 58:1:396-I102, t)I,x (!e) Il.t.~tstmn Y.(I. (aRrrnntt. L...Stmal,ts II, rl .il', I;u tnd 11i<1- VO1..62, NO. S. htARCli 1979 Less Harmful Cigarettes 477 tinc contt-m of (iG:rrrur smoke in retation to drnh rams. Emiron Res I9 :21i3-27 I. 1976 117) \[%NItt N. II~twH. l\': Swtisxiral uspects of the analysis of data 6ont retruspettise studi<•s of disease. J N:nl C:anmr Insl °:719-i18. I9G9 (18) .Ailt I ILAtN OS: lUdtlfiUrliorr by' .t multiRdriatf cunfuuntft'r CtOrt'. Arn J L'.pidemiol I01609-620, 197fi (19) (aet JJ: Poiul dnd unen•rI rsum.tuon of the (ommou odds r.mu iu thr eumLm:uiuu of 2X2 teWrs Kith fise<I mergina6. Rmmrurk.r 57:171-175. 1070 (20) Atsstccrr JC Jx: l'he 1976 Alaxwell report. Tohucco Reporter I03a16-17. Si-;i. 19;1i - (21) Iluttolt~s I), St.uantv 1. Ilrctn SS, ci al: Ch"mical swdirs uu ruLue<o smokr. XXXIX. On Ihr idenufitution or err<iuo- ge•ns, tumor pmmotrrs and e.rrrinogens in lobacco smoke. In Prueredii~q, of tho Third \1'orld Conleremr nn 9moking and Ilndth i1C)ardrr F.l., Ihtifm,wn Or ('ori GB, cdsl• Iul I. lt:ulun4lun. 1)C.'. 1'.S. Gusl Prim Ofl. 1976. pp 125-I15 (22) lt't:snra tl., HottsrtuaD: L.ess hurmful rcuys Of emoking. J Nail Caneer Inu 49:I719-1758, 1972 (23) IA'rNntk EL, [ie)tt+ltNa n, Onrtt RG, eds: ProceedinGS of the 7hird AVorld Cungrem out Smukiqw, and Health, sol I. Vt'ushington, D.Cl: U.S. Govt Print Off. 1976 (24) l4sIutt< t:L• tit<,tn S. rds: Lune caneer. t'[CC 'Cedtnieal Reporl Series, vol 25. Genec:r. Union Internation:tle Cuntre Ic 1976 (?s) Kt'u:y EL Kt.rs }{: Ke<luedun of smoke n[cotinc and snmke eoudcns'ate in :Aurura dwrnR the pasi fifteen evrs Farhl Alitt Orsiru LLb:rkreGie Iti.3211-329. 1975 (26) Tuuu (IF: ChsnGrs in anoAiug pa4erns in the ('.F.7ri Pro- umdiqKs of dre EJesrnth IntenmtiuNd C.m<'er (bngress, 4lorrnu•, Imlc, 1971, or<.,sionW paPer I. C.undmr. I-obzrCo Rrsedtth Cuunril, 1975 i27l Il?rnt 4'J: [lrndumd)seu tum Prnblent des t'erbrau<h<•s an \'ikutin und Itaueld.udens:rl in drt RuudesrepuLlik Deutuh- I.md IUu <lie J:rhre 1961 his 1975z Britr'[Cub:rkfurarlrtl:101-Ibl, B7(i (_^•Y) :\us\n PI: ChHUges in peranral ~mukrng huhits brouKht about In thzn.Rrn in rigzreue smuke )irld. In Procer<lings of Ihr Slrrh Intemutiunul Cub'.ntu .S<irutific Cnngrrsv. Cnkyo, J:tp:nr, 1976. Curesta lufnrnuni<rn Rull I.spe(i,rl):102-I08, 1976 (24) Fattusus S. I9ttcutn C\. Changrs of stnokmg Irxhiu xnd (uugh iu men anohing eigorrurs with 30% NSM tohmto suh- sliuuc. Br AIrd I 11127-I 130, 1976 70) Rtsrrl NLl, lCU.sou C. Rettt. l':\, it al: Phurna nieotine Iesrls dt,rr smakin,K <ig:veurn wilh high, medium, and low niecninr 1ieIds. lir \led .I 2:111-116. 1975 (If) Drp,uunrrn if Ile:dth. 6:duauiun. and lt'rlLrre: te•eoage• ,umkmgr A'atiorral p.nterne if igurrtt<• smol+ing. agcs 12 thu,ugh B• 1972 and 1971. DIIE11' Publ No. I[SA1 76-931. lC,nlriutruo. D.C.c U.S Gott Prinr Off, 1976 172) 11u I t eus CL. lYSNutK t:L: A blind npul iu presrmive ntedi<'ine. I.\)1.\ 23Ii:2196i ?197. 1976 (77) lt'utisvs ('.L. :\ItNotn (ai. l\'tNutx t:l.: Primary preventiun ,if luoui< disrue he;Giuning in ehildhoud'. The 'Anow your hndy' pml;r:un: nrsign ul stud). PRrc Ned 6::361-357. 1977 Ii{{ Stutstt<v JL: A<rniral rreirtc and rvalerntion Of srnoking ,,,nuul mrrhuds. Public Ilcalilt Rep 81:d87-19I• 1969 t7it tim kt tir <f, f..it Smukin8 rrssvtiuu prugcuus of tir amrriaur llodth Euuud;niun. 19ct \Ied 5.151-171, 19711 J NATI. (I:\NCF.R INSr