Tobacco Documents Online

. S1VlOKTNG _ AND HEALTH Sumniary and Report of The Royal College of Physicians of London on Smoking in relation to Cancer of 'ihe Lung t and Other Diseases PITMAN PUBLISHING CORPURATION' NEW YORK '1'ORONTO LONDON First published in Great Britain by Pitntan Medical Pnb7isliing Co. Ltd. @ 1962 by The Royn1 College of Physicimar of London CoPyrigltt © 1962 by Pitmnn Publishing Corporation 2 West 45 5•treet, New York 36, N Y. Library of Congress Catalogue Card Number 62-15794 M8EUS8UZ I. , I

SUMMARY Introduction Scveral serious diseases, in particular lung cancer, affect smokers mrne ofien than non-smokers. Cigarette smokers have the greatest risk of (lying from these diseases, and the risk is greater for the heavier smokers. 'fhe many deaths caused by these diseases present n challenf;e lo medicine, in so far as they are due to smoking they ,Ix)nld be prevcntnble. 7'hi.s report is intended to give to doctors and oth,t<; evidence on the hazards of smoking so that they may deciile vh:u should he done (Paras. r_3). History of Smoking rlficr its inttodnction to Europe in the r6th century, tobacco smokiug, mostly in pipes, rapidly became popular. It has always had its advocates and opponents, but only recently has scientific study produced valid evidence of its ill-effects upon health. Cigaretics have largely replaced other forms of smoking in the past sevent y years, during which time tobacco consumption has steadily increased. It is still increasing. Women hardly ever smoked before 1920: since then ihey have smoked steadily increasing numbers of rigareues {Pig,rne 1, p. 3) (Paras. 5-6). Present Smoking Habits 17irce-quarters of the men and half of the women in Britain sninl=c. A4en stnokc more heavily than women. Smoking is now t:•idesprcad atnong schnolchildren, especially boys. (Pigreras 2 and _7, pp. S mrJ 7) (Parae. 7-8). Many doctors have given up smoking vinte ih(, dangots of'thc habithave become apparcnt: only half of S2 them now smoke and less than a third smoke cigarettes (Figrfres 4 and s, pp. 9 and 7t) (paras. 9 and ri). Advertising of Tobacco. 'Phere has been a steep increase in expenditure on advertisements of tobacco goods recently. Over n million pounds was spent on such advertisemetns in i96o (Tab(e 1, p. 6; Pigrrre 6, p. 13). The increase has mostly been devoted to advertising cigarettes and many recent advertisements have been aimed at young people. It cannot, however, be assumed that advertisements arc responsible for the continuing increase in tobacco consumption today (Paras. ro-i71. Chemistry and Pharmacology :of Tobacco Sraoke Tobacco smoke is complex in composition. Its most important components are: nicotine which acts on the heart, blood vessels, digestive tract, kidneys and nervous system; ntinute amounts of various substances which can produce cancer; and irritants which chiefly affect: tlte bronchial tubes. Thc anrounts of carbon tnonoxide and arsenic in the smoke are probably too small to be harmful (1laras. 12-22). Smn)ing and Cancer of the Lung There has been a great increase in deaths from this disca,c in many countrics during t:fte past 45 years (Pip,rrre 7, P. 15). Some of this increase may be due to better diagnosis, but much of it is due to a real increase in incidence. Men are much more often affected than women. (Table II, P. i¢) (Paro,c. 23-24,) Surveys. Many comparisons have been made in different coun- ' tries between the smoking habits of patients with lung cancer a.ud those of patients of the same age and ses with other diseases. All have shown that more lung cancer patients are smokers, and more of them heavy smokers than are the controls. 'fhe association between smoking and lung cancer has been confirmed by pr'ospec- tive studies in which the smoking habits of large numbers of men have been recorded and their deaths frorn various diseases observed subsequently. [l.ll these studies have shown that death rates from lung cancer increase steeply with increasing consumption of , cigarettes. Heavy cigarette smokers may have thirty times the death rate of non-suiokers. (Pigrrre 8, p. 17). They have also shown S3 £Z48£0£80Z

~ that cigarette smokees are much more affected than pipe or cigar smokers (Figure 9, p, rg) and that those who had given up smoking at the start of the surveys had lower death rates than those who had continued to smoke (Figrrre zo, p. 21). Various criticisms, based on possible errors of selection and of diagnosis, which might have_ caused a spurious association between smoking and lung cancer in these studies, are discussed (paras. 25-29). 1'athology of Smokers' Lungs. Of three types of lung cancer, only the two commoner types are associated with smoking. The lungs of smokers without cancer show changes of chronic irritation, of the sort which might precede cancer, more often than the lungs of non-smokers (paras. 3o-3r). Interpretation of the Evidence. The association of lung cancer with cigarette smoking is generally agreed to be true but various i ossible explanations of this association other than that of cause and effect have to be considered These are (para. 32).:- (i) that people who are going to get lung cancer have an increased desire to smoke throughout their adult lives: (ii) that smoking produces cancer only in the lungs of people who are in any case going to get cancer somewhere in the body, so that smoking determines only the site of the cancer: (iii) that lung cancer affects people who would have died of tuberculosis in former times but have now survived with lungs susceptible to cancer: (iv) that smokers inherit their desire to smoke and with it inherit a susceptibility to some other undiscovered agent that causes lung cancer: that smokers are by their nature more liable to many diseases, including lung cancer, than the "self-protective" minority of non-smokers: (vi) that smokers tend to drink more alcohol than non- smokers so that drinking and not smoking may cause lung cancer: (vii) that motor car exhausts, or- (viii) that generalised air pollution may render the lungs of smokers more liable to cancer. None of these explanations fits all the facts as well as the obvious one that smoking is a cause of lung cancer. There are other causes, including air pollution and substances which may be met in a few S4 occupations, but none of them is of such general itnportance as smoking (para. 33). There are a few facts which may be considered to conflict with this conclusion namely:- (i) that lung cancer occurs in only a minority of smokers: (ii) that death rates from this disease are lower in some countries than would be expected from their cigarette consumption: (iii) that there is some conflicting evidence on the effects of inhalation of smoke: (iv) that no animal has yet been given lung cancer by exposure to cigarette smoke. :Conclusion. These facts are discussed (paras. 33-40) and none of them is found to contradict the conclusion that cigarette smoking is an important cause of lung cancer. If the habit ceased, the number of deaths caused by this disease should fall steeply in the course of time (para. ¢z). Smoking and Other Lung Diseases Chronic bronchitis is a common and distressing disease in Britain and causes many deaths, especially in middle aged and elderly men. Smokers, particularly cigarette smokers, are much more often affected than non-smokers (Figure 11, p. 29). Other agents, of which generalised air pollution is the most important, are involved and it may be that damage done to the bronchial tubes by cigarette smoke makes them more susceptible to these other agents. Many men and women who are now disabled by chronic bronchitis might have remained well had they not smoked (para.s. 42-50). Smoking may possibly contribute to the development of pulmon- ary tuberculosis, especially in the middle-aged and elderly (paras. 51-52). Smoking and Diseases of the Heart and Blood Vessels Coronary heart disease is a more frequent cause of death in smokers, particularly cigarette smokers, than in non-smokers, although the latter are also commonly affected (Table III, p. 3q). Those who give up smoking have a reduced death rate (Figure 12, p. 33). Many other factors, such as mental strain, sedentary occupation and diet, may explain some of the association of this S5 tiZ48£0£80Z I

disease with smoking, but cigarette smoking probably plays a significant part in rendering men in early middle age more liable to its serious effects. (paras. 53-57). Smoking appears to play a part in causing other arterial diseases but not high blood pressure (paras. 58-59)• Smoking and Gastro-intestinal Diseases Smoking affects the movements and secretion of the gut in many ways and may cause symptoms such as nausea and discomfort. It depresses appetite and may reduce weight. It does not.appear to cause gastric or duodenal ulcers but interferes with their healing (paras. 6o-65). Cancers of the mouth, throat and gullet occur more frequently in smokers than in non-smokers (para. 66). Smoking and Other Conditions Several relatively uncommon diseases occur more often in smokers than non-smokers (paras. 67-69). Smokers may be more liable to accidents than non-smokers (para. 70). Women who smoke tend to have babies that are underweight (para. 7r). Smoking may impair athletic performance (para. 72). The Psychological Aspect of Smoking Vcry little is known about why people smoke. Children tend to R llmw their parents' smoking habits. Intelligent children smoke less than duller children. Adults claim that smoking gives a sense of relaxation, helps them to concentrate and gives them relief when they are anxious, but these claims are difficult to test. Psychologists have suggested various unconscious motives for smoking (paras. 73-78)• Smokers tend to be more restless, less dependable and more neurotic than non-smokers. Cigarette smokers are more extra- verted than non-smokers, pipe smokers are more introverted. That the tendency to smoke tnay be partly inborn is shown by studies of the smoking habits of twins (para. 79). Smokers may be addicted to nicotine. They may wish to stop smoking for a variety of reasons, chiefly because of expense or fear of ill health. It appears that social factors play a bigger part in S6 determining smoking habits than internal drives or needs (paras. 80-82). Conclusions The benefits of smoking are almost entirely psychological and social. It may help some people to avoid obesity. There is no reason to suppose that smoking prevents neurosis (paras. 83-85). Cigarette smoking is a cause of lung cancer, and bronchitis and probably contributes to the development of coronary heart disease and various other less common diseases. It delays healing of gastric and duodenal ulcers (paras. 86-89). The risks of smoking to,the individual are calculated from death rates in relation to smoking habits atnong.Iiritish doctors (Table I V, p. 44). The chance of dying in the next ten years for a man aged 35 who is a heavy cigarette smoker is i in 23 whereas the risk for a non- smoker is only i in 9o. Only 15 per cent (one in. six) of men of this age who are non-smokers but 33 per cent (one in three) of heavy smokers will die before the age of 65. Not all this difference in expectation of life is attributable to smoking (paras. 9o-gr). The number of deaths caused by diseases associated with smoking is large (Table V,;p. 47) (para. 92). The need for preventive measures. Reduction in general air pollution should reduce the risks of cigarette smoking; but it is necessary for the health of the people in Britain that any measures that are practicable and likely to produce beneficial changes in smoking habits shall be taken promptly (paras. 93-9S). Preventive Measures Since it is not yet possible to identify those individuals who will be harmed by smoking, preventive measures must be generally applied (para. g6), . , The harmful effects of cigarette smoking might be reduced by efficient filters, by using modified tobaccos, by leaving longer cigarette stubs or by changing from cigarette to pipe or cigar smoking (paras. 97-102). General discouragement of smoking, particularly by young people, is necessary. More effort needs to be expended on dis- covering the most effective means of dissuading children from starting the smoking habit (paras. roj-io7). There can be no doubt of our responsibility for protecting future generations from S7 9Z48£0£80Z

w developing the dependence on cigarette smoking that is so wide- spread today. Most adults have heard of the risks of cigarette smoking but remain unconvinced. Doctors, who see the consequences of the habit, have reduced their cigarette consumption. Some evidence of concern by the Government is needed to convince the public. The Government have so far only asked local health authorities to carry out health education in respect of smoking, but little seems to have been achieved. The Central Council for Health Education and Local Authorities spent less than £5,000 on anti-smoking education in t955-60, while the Tobacco Manufacturers spent Cg8,ooo,ooo oa advertising their goods during this period (paras. ro8-rrr). Possible Action by the Government Decisive steps should be taken by the Government to curb the present rising consumption of tobacco, and especially of cigarettes. This action could be taken along the following lines (paras. rr2- n 9):- (i) more education of the public and especially school- children concerning the hazards of smoking: (ii) more effective restrictions on the sale of tobacco to children: (iii) restriction of tobacco advertising: (iv) wider restriction of smoking in public places: (v) an increase of tax on cigarettes, perhaps with adjustment of the tax on pipe and cigar tobaccos: (vi) informing purchasers of the tar and nicotine content of the smoke of cigarettes: (vii) investigating the value of anti-smoking clinics to help those who find difficulty in giving up smoking. Doctors and Their Patients I'here are good medical grounds for advising patients with bronchitis, peptic ulcer or arterial diseases to stop smoking. Even a smoker's cough may be an indication that the habit should be given up. Doctors are better able to help their patients to stop smoking if they do not smoke themselves. They have a special responsibility for public education about the dangers of smoking (paras. 120-121). sa 9Z48£0£80Z

0 22. Arsenic. This is of interest because it is a carcinogen. It used to be present in tobacco smoke in very variable amounts, being derived from arsenical insecticides used in tobacco plantations. The use of these substances has declined1ea and the arsenic content of cigarettes is now infinitesimal. There has never been enough arsenic in tobacco for this to be likely to cause cancer by itself'a• 4a• 60 but it might have had an adjuvant (or co-carcinogenic) action, whose significance cannot be dismissed. SMOKING AND CANCER OF THE LUNG Increasing Death rates from Lung Cancer 23. During the past 45 years lung cancer has changed from an infrequent to a major cause of death in many countries. This increase has been most serious in men and women in late middle age, whenn family and professional responsibilities are at their height. Table II, p. 14 shows the total number of deaths that have occurred in men and women between the ages of 45 and 64 since 1916; .and Figure 7, p. 15 presents the age-standardised death rate for men in these age groups during the same period from cancer of the lung, other forms of canccr, tuberculosis of the lungs and bronchitis. While death rates from lung cancer have been increasing, those from other forms of cancer, and other respiratory diseases have been declining or, like bronchitis, remaining stationary. 24. The experience of chest physicians and surgeons in the past 30 years leaves no doubt in their minds that there has been a very large and real increase in incidence of lung cancer, though some patholo- gists consider that the disease used to be more common than mor- tality figures suggest and that much of the increase may be due to improved accuracy of diagnosis on death certificatest". if there has been no increase it is difficult to see why cancer of the lung alone among all cancers should have become so much more frequently diagnosed in so many countries, and the much faster rate of increase in men than in women (Table 11, p. 14) cannot be due to improved diagnosis. T7iere must have been a notable increase even though it may not be so great as mortality figures suggest. To account for this in- crease it is necessary to postulate some causative agent to which human lungs have been newly and increasingly exposed during the present century. Cigarette smoke is such an agent and there is now a great deal of evidence that it is an important cause of this disease. Retrospective Surveys 25. At least 23 investigations in nine countries"'• s' have shown by retrospective study that among sufferers from lung cancer there is a 12 LZ48£0£80Z EXPENDITURE ON AGVERTISING, U.R. 0 PRESS \ t.y, N COMBINEU ir CIGARETTES a 6 a U USA I'IGUAi 6 EXPEN~~L V SION NnTHF.TIJNITED 1QNGOOMrt1954,1960~a pnF55 AND ON Three quarters of the money spent on tobacco goods pays for advertisements in tile press or on T.V. (Table I, p. 6). Much of the recent increase in expenditure has been devoted to adverlisements of cigarettes on television. Between 1954 and 1960 there was a fivefold increase in advertising of cigarettes and only a lhreefold increase in advertising pipe tobaccos and cigars. 11 TOBACCOS, CIGARS. ETC.

TABLE II AVERAG6 ANNUAL NUMBERS OF DEATHS FROM VARIOUS CAUSES DURING FiVE YFAn PERIODS FROM 1916'r0 1959IH MEN At~p WOMEN AGED 45-64. ENGLAND AND WALE4. j Disease ' ~ a Cancer of Lung (excluding mediastinum and trachea) b ~ Cancer other than Lung ! (excluding Hodgkin's ` Disease and Leukemia) c Bronchitis (all forms) . (including bronchiectasis) : d Tuberculosis of Lungs Period Men I Women . Men Women I Men i Women i, Men Women 1916-20 146 i 87 I, 8,876 10,881 1 4,708* ~ 3,504* I. 6,607 I 3,225 1921-25 255 121 10,325 ~ 12,034 3,804* ~ 2,776* ~ 5,689 ! 2,635 1926-30 481 177 ~ 11,005 ~ 12,940 3,053* ~ 1,947* j 5,766 ~ 2,413 1931-35 1,158 324 II 11,185 13,718 2,339* I 1,222* , 5,488 ~ 2,134 1936-40 2,020 463 i 10,985 I 14,212 2,757* i 1,086* 5,271 1,826 1941-45 ~ 3,090 566 i 10,458 14,284 5,644 1,954 ; 5,146 1,522 1946-50 i 5,031 ~ 761 i 10,121 i 13,984 5,649 1,658 4,785 ~ 1,346 1951-55 j 7,348 i 980 i 10,027 I 13,831 6,238 i 1,614 i 2,862 742 ' I956-59t I 9,108 i 1,202 j 10,265 j 14,119 6,437 j 1,526 ' 1,484 'I 345 A a. 161.1 - 163; b. 140-200, 202, 203, 205; c. 500-502; 526; d. 001-008. Currentlnternational Statistical Classification: i 1960 figures not available, * Figures not comparable with figures for later years because of changes in allocation of certified causes of death. Figures kindly supplied by Dr. R. .4. M. Case of the Chester Beatty Research Institute. l= 7 G- SO a 3~„Ro 71 B 5 Z 0 ANNOAL NUMBER OFOEA7H5 PER700,000 29 I ae .

• higher proportion of heavy smokers and a lower proportion of light smokers or non-smokers than in comparable control groups. Not only have these studies all shown the same association, but among those dealing with larger numbers it is quantitatively similar, even though the investigations have been made in different countries. 26. The methods of these investigations have varied but in essence the answers to questions about smoking habits given by patients with lung cancer were compared with those given by individuals, usually patients in the same hospital, without lung cancer. Such methods are open to criticism because of several possible ways in which bias might have been introduced in spite of precautions which were taken. In one investigation, that of Doll and Hill"• ss, the criticism that the amount smoked by cancer patients might have been over-estimated because the patient or the interviewer suspected the diagnosis (although interviewers were, in fact, not informed of the diagnosis) was met by the findings in a small group in whom lung cancer had been wrongly diagnosed. Patients in this group were at the time of the interview thought to be suffering from lung cancer, but subsequent investigation showed them to be suffering from some other disease: their smoking habits fell into line exactly with those of the control group. Another criticism was that the control group, which was usually composed of other patients in the same hospitals as the lung cancer patients, might not have represented a fair sample in respect of the smoking habits of the population from which the patients eame. But comparison of the smoking habits of the hospital control group who lived in the Greater London area with those of the general population of the same area, showed that the hospital control group actually smoked more. This was to be expected in view of the associa- tion of smoking with several other diseases, and would actually lead to underestimation of the effect of smoking in predisposing to lung cancer in these retrospective studies. Prospective Surveys 27. The results of retrospective studies have been fully confirmed by prospective studies in which, first, the smoking habits of a defined population group have been ascertained, and then the causes of death during several years' observation have been recorded. Four indepen- dent groups in three countries have conducted investigations of this sort'e, se, 57, sz• sa They all show a steady increase in numbers of deaths from lung cancer with increasing cigarette consumption, and are in close quantitative agreement not only with each other but also with most of the retrospective studies. The results of the first three of these investigations are summarised in Figure 8. The rather higher mortality found in the British study compared with the American 16 6ZL8SUS8UZ • RATIOS OF LUNG CANCER MURTALIIY BETWEEN CIGARETTE SMUNERS& NON-SMOKERS 30 z 1 I'7GURR LUNR(',EC NOCERtiDEA R T SUNaT6{REE PROSPECTIVSESSOUOr~ PER DAY ND The itgure shows how much the risk of getting lung cancer is multiplied in those who smoke various numbers of cigarettes per day compared with the risk of non-smokers. The first horizontal line in the figure indicates ten times the risk of non-smokers, and so on. The figures are derived from:- Dolt and Hill's study of British doctors aged 35 and over" (_ Hammond and Horn's study of American men aged 50b9°6 (-- o--)Dom's study of American ex-service men aged 30 and over"-(-* -)• The similarity of the steady increase in lung cancer risk with increasing cigarette smoking found by these three independent studies is impressive. The higher British rates may be due to the British habit of smoking cigarettes to a shorter stub length than the Americans and to the greater exposure of British men to air polluted by domestic and industrial smoke, 17 I 4 v

. ',f i studies, may be explained partly by the observation that the British smoke more of each cigarette than do the Americans, thus receiving a larger dose of smoke and losing the filtration effect of a long stubsa, sr (see paras. 97 and 101), and partly, perhaps, by the greater expo- sure of the British to air polluted by chimney smoke88. These investiga- tions in which estimates of relative risks for different forms of smoking were possible have all shown that pipe smokers incur a considerably smaller risk than cigarette smokers. The American investigations have also shown that the risk in those who smoke only cigars is even smaller (Figure 9) and may be no greater than that for non-smokers. 28. An important finding in all of these prospective investigations has been that the risk among those who have given up smoking for several years is less than among those who continue to smoke (Figure 10, p. 21). 29. The possibility of continuing observation of a selected popula- tion in a prospective study is particularly valuable since it provides an answer to the criticism that even in a prospective study initial selection bias may affect the results. The subjects in such a study are selected by the fact that they have replied to a questionnaire or have been chosen for interview, and bias might be introduced by inclusion of more or fewer smokers than non-smokers who are in ill health at the beginning of the observation period. In all these studies, however, the association between deaths from lung cancer and smoking was more evident in the later than in the earlier part of the observation period, which is the reverse of the trend that would be expected if the association was even in part due to initial selection bias. Another possible criticism of these prospective studies concerned accuracy of diagnosis, since in three studies the certified cause of death was accepted. Bias might be introduced, for example, if there were a special tendency for lung cancer to be diagnosed as the cause of death in heavy smokers. But the total death rate was found to increase with the amount smoked, the excess deaths among smokers being attribut- able principally to disease of the cardiovascular system, especially coronary thrombosis, and to certain respiratory diseases as well as to lung cancer. Hence if some of the deaths among smokers were being attributed falsely to lung cancer, the effect of smoking in increasing mortality from other diseases was underestimated. In the investigation of Doll and Hill and of Hammond and Horn, moreover, the associa- tion with smoking was actually greater for those cases of lung cancer in which the diagnosis had been established by the most certain method, i.e. by microscopic examination of diseased tissue, than for those in which it was dependent on clinical evidence alone.* ' For further discussion of the validity of the evidence provided by these surveys see references 13, 14, 41, 47, 51, 53, 128, 163, 18 0£48£0£80Z DEATH RATES FROM LUNG CANCERIN MEN IN RELATION TO TYPE OF TOBACCO SMOKED NON- SMOKERS CIGARS ONLY SMO&ERS OF:- PIPES ONLY CIGARETIES g PIPES/CIGARS CIGARETIES ONLY 135 90 67 39 39 0 U.K. FIGURES FROM OOLI& NILL;1956 \ U.S.A.FIGURES FROM NAMMONO & NORN;1956 127 FIUURE 9 DEATit RATES PROM TOBACCO SMOKEt~MEN ACCORDINtl TO TrrE TYPE I'F These figures are taken from the prospective study of British doctors aged 35 and over by Dolt and Hilt (Dalt©) and of American men aged 50-69 by Hammond and Horn.BO only in the U.S.A. wcre there enough pure cigar sntokers to estiniatc their death rate which was the same as for non-smokers. Pipe smokers had three times, smbkers of cigarettes with pipes or cigars five to eight times and pure cigarette smokers about ten times the mortality of-non-smokers. The similarity of the rates in both studies is impressive. The differenees between cigarette smokers and other tobacco smokers may be due to the 'greater tendency of cigarette smokers to inhale the smoke (see para. 89). 19

0 Pathology 30. There are three principal pathological types of lung cancer, and precision in microscopical diagnosis shows that smoking is associated specifically with two of these. When Kreyberg in Norway made an independent classification of pathological sections from British cases investigated by Doll and Hill without knowing the smoking habits of the subjects, he found a close relationship between the daily amount smoked and the development of cancers described as squamoust and undifferentiated (which are now the commonest pathological types) but little or none with the less common cancers described as adeno-carcinomas f e". 31. Several studies of non-cancerous changes in the bronchial epithelium in relation to smoking history have been published. Auer- bach and his colleagues in New Jersey'• 6 studied nearly 30,000 sections from the bronchi of 83 men who died of causes other than lung cancer, and 34 men who had died of lung cancer, all of whose smoking his- tories were known. There was a quantitative relationship between cigarette consumption and the frequency of microscopic changes suggesting chronic irritation. Such changes are possible precursors of some types of cancer and were most frequent in the men with lung cancer. Similar findings have been reported by other patho- logislsau, ton, iva Interpretation of the Evidence 32. Various independent authoritative bodies* have been set up to examine the evidence of the relationship between cancer of the lung and smoking and have all agreed that it is established. The most obvious explanation of this association is that it is causal. There are, however, other possible explanations which must be considered. (i) That many years before lung cancer becomes manifest some early process in its development may produce the desire to smoke. This hypothetical process must he postulated to begin to act as long as 40 to 50 years before the onset of clinical disease, to produce a desire to smoke a number of cigarettes daily in proportion to its liability to mature into cancer, to have become suddenly more prevalent within the past few decades, and to cause a desire for cigarettes rather than pipes or cigars. These postulates appear highly improbable. (ii) That smoking may not cause cancer but only determine the site t In squamous cancers the cells bear some resemblance to those found in the skin. In adeno-carcinoma the cells retain the appearance of those in glands. * British Ministry of Aealth1de British Medical Research Conncil'•', National Cancer Institute of Canada'53, Netherlands Ministry of Social Affairs and Public Health°60, U.S. Study Group of Smoking and Health 1957, U.S. Public Health Service°p• 1°', World Health Organization'°". 20 4£68£0£80Z DEATH RATES FROM LUNG CANCER IN RELATIDNTG GIVING DPSMDKING STOPPEH ShiORING FOU~- NON-SMBRERS hI0flE1HAN t0VEdR5 VEAAS ID t-m ussniAN SIILLSMORING YEABS U.S.A. Nammond6Horn;1958 U.K. Uoll6Hill;19§6 158 SMDK[f15 OF LESS 1AAN 20 CIGARE11E5 PEA flAY , „ 20 ofl MaOE • ® All LIGAAETIE SMOKERS T~'tOt3aE It1 THE EFFECT OF GrvINOGANCER.KINO ON DCATn RATES rROM LrINO In this figure rates given fo r American men relate only to cases in wluch the death the diagnosis was eslablished mtcroscopically, so that these [ates ue lower than thosa illustrated from the same source in Figure 9, p. t9. The British figures are from Doll". Only in the American study were hcavier and lighter smokers separated. There was a similar, much reduced death rate in those wlw had givcn up smoking, especially if the period without smoking had bcen for more Ihan ten years before the a highergmortal ty than the ligl ter s~m kersswho contgnued to s nokeS retmned 21