Document V3w3YO2byjpqr0OD7bY38y36Z

FILE NAME: Smoking (SMOK) DATE: 1952 DOC#: SMOK040 DOCUMENT DESCRIPTION: Journal Article - A Study of the Aetiology of Carcinoma of the Lung V4 ( t t f ilf A STUDY OF THE AETIOLOGY OF CARCINOMA OF THE LUNG BY RICHARD DOLL, M.D., M.R.C.P. Member of the Statistical Research Unit of the Medical Research Council | AND fi A. BRADFORD HILL, C.B.E., Ph.D., D.Sc. p- Professor of Medical Statistics, London School of Hygiene and Tropical Medicine; Honorary Director o f the ( Statistical Research Unit of the Medical Research Council jj a previous paper (D oll and H ill, 1950) w e reported ae first results o f a large-scale investigation undertaken t determine whether patients w ith carcinom a o f the iing differed materially from other persons, either in aeir sm oking habits or in som e w ay which m ight be elated to the theory that atm ospheric pollution is esponsible for the developm ent of the disease. We hncluded that sm oking is a factor in the production if carcinom a o f the lung, and this conclusion w as in nnform ity with the results o f some other investigations. )ur first observations were, how ever, lim ited to patients Irawn m ainly from L ondon and the adjacent counties. iVe have now extended the investigation to other parts if the country and have m ade m ore detailed injuiries into sm oking habits. M any further patients lave been interviewed (during January, 1950, to Febuary, 1952) in hospitals in Bristol, Cambridge, Leeds, md N ew castle-upon-Tyne, and also in eight o f the wenty London hospitals w hich co-operated in the first Dart o f the inquiry. Method of the Investigation The method of inquiry was described in detail in the previous paper. In brief, w e obtained notifications o f patients admitted with cancer o f the lung, stomach, or large bow el to each co-operating hospital, and these patients were interviewed by almoners, engaged'w holly on our research, w ho recorded the answers to a pre arranged questionary. The patients with carcinoma o f the stomach or large bowel provided one " control " group, but another, and more important, " control " was obtained by interviewing patients with diseases other than cancer. Each of these latter patients was chosen so as to match a lung-carcinoma patient-- namely, o f the same sex, within the same five-year age group; and in the same hospital as nearly as possible, at the same time. In the extension of the inquiry this same method has been used, but with modifications. First, notifications were made of patients with lung cancer but no longer of those with cancer of the stomach or large bowel. Secondly, the interviewers could not visit hospitals outside London whenever a suitable patient was admitted; they therefore visited the provincial centres at intervals and interviewed those patients suspected of having lung cancer who were then in the hospitals. At Bristol, Cambridge, and Leeds they also interviewed a few who were attending the out-, patient departments. An important modification was in the choice of the matched control patients. It was impossible to obtain at the provincial centres a group confined, as before, to patients with diseases other than cancer. Our previous analysis, however, had shown that patients with cancer other than lung carcinoma (mainly patients with carcinoma of the stomach or large bowel) gave smoking histories indis tinguishable from those given by non-cancer patients ; we therefore widened the matched control group to include, with certain exceptions, other forms of cancer. The excep tions which we continued to exclude were cancer of the lip, tongue, mouth, pharynx, nose, larynx, and oesophagus, since, it has at times been suggested that cancer of these sites may also bear some relationship to tobacco consumption. We also excluded all other cancers arising inside the chest. Even then it was still difficult m the provincial centres to find an adequately " matched " control for each lungcarcinoma patient. The provincial hospitals had been chosen so that a large number of lung-cancer patients could be interviewed at each visit. They were therefore thoracic or radiotherapeutic centres serving regions--that is, town and ,country. We could not seek control patients in ` the adjacent general hospitals, since these mainly serve the towns they are situated in, and smoking habits vary between town and countryside. On consideration the field of choice; was finally extended to other hospitals or units in the same area, but only to those serving as regional centres--that is;) like the thoracic or radiotherapeutic centres. Thus the :, cases and their controls should be drawn from equally wide-areas of town and country. We have also included . in the control series some patients interviewed as having lung cancer but in whom the condition was finally excluded (as at the Brompton and;Harefield hospitals in the first part' of the investigation). It will be shown that this procedure does not influence the results.. The classification of each patient was, as a genera) rule, based u p on the hospital dlbcharge diagnosis, ob tained from a study of the hospital records after the ? patient's discharge or death. Where that record was.; indefinite information was obtained from the practi tioner or hospital to w hose care the patient had been transferred. Occasionally evidence contradicting the . hospital discharge diagnosis became available-- foi ex- ; 4797 ,< - i ^ -__ HpPI AETIOLOGY OF CARCINOMA' OF LUNG ample, by histological examination at necropsy-- and in f i i these cases the diagnosis was based upon the best evi dence. F ive cases with no record o f any final diagnosis T able I.---Numbers of Patients Interviewed inEach flue*' ............ G..r.o..u.p. ` 1 ....... ppi have been excluded. * The Data 'V . mm i V ii M0f..- gH*. Between April; 1948, and February, 1952. (the whole, 'period o f this investigation), 3 j446 cancer/patients were notified (lung; stomach, and large bowel). O f these, 156 w ere aged 75 years or more and were not interviewed, Wince w e lim ite d th e inquiry to younger patients w ho , could give more reliable histories o f sm oking and other personal characteristics. In 82 cases,theidiagnosis was ' changed before tthe .almoner, paid her visit. O f the re maining ,3,208;patients, 85% were interviewed an d .498 (15%) were not. The reasons why patients were not interviewed were : already discharged from hospital, 213 ; too,,ill,.vl65; dead, 72 ; too deaf, 33 ; unable to speak English clearly, 14 ; while in one case the interview was. abandoned because the patient's replies appeared w holly unreliable; N o patient refused to be interviewed. With the lung-cancer group alone the proportion not inter viewed was also 15%. We can see no reason w hy failure to interview all the patients should have biased the re sults, since it was mainly due to the time that had to Carcinoma oflumi. ,, ,, .other special sites ,, n ,, uncertain primary site. Other diseases;:ATmalched controls)* . ,, ,, .B (other controls) Uncertain diagnosis (records untraced) All.groups* ; * A further 531 patients were interviewedj n rural hospitals, with rS,, smoking habits in country areas--see section on estimated risks -in'ttw country. the earlier part :of the investigation as a second coM group; (2) patients initially interviewed as having carciridS of the lung, stomach, or large bowel, but later fouhS have other diseases ; and (3) others interviewed as match controls and n o t required when the patients with whoiL they were paired were found not to have carcinoma of t lung. * In T able II the matched patients in the lung-carcinoihf and control groups are compared for sex and age distrtt bution, places o f interview, places o f residence, and, fciM males, social status. The method of selecting conffhljj elapse between the date o f notification and the date o f T able II.-- Comparison Between Lung-carcinoma Patients the alm oner's visit. Such losses were, o f course, few in the provincial centres, since here the almoner was re quired to interview only those in the hospital at the time o f her visit. The remaining 2,710 patients initially presumed to be suffering from cancer of the lung, stomach, or large Matched Control Patients A ttribute for Com parison No. of Lungcarcinoma Patients -M F No. of Control Patients MF with Other Diseases Attribute for Com parison No. of Lung- carcinoma Patients, M and F '*'** .'"Ft No.-oi Contra Patient M and] bowel and the 1,632 general medical and surgical patients Place o f inter- -- Age: view: interviewed as matched controls to the lung-carcinoma 25- 17 3 17 3 G reater London 1,035 1,035 patients form the subjects o f this analysis. Table I gives 3545- 116 15 116 15 Bristol 73 493 38 493 38 Cam bridge .. 36 73 s'vit 36/. (SSL *} the numbers in each disease group (final classifications 5565-74 .. 545 34 186 18 545 34 Leeds 186 18 Newcastle 58 58:%? 263 263 % based on the hospital discharge diagnosis, etc., as de scribed above). The 1,488 cases o f carcinoma of the Alt a g e s .. 1,357 108 1,357 108 All places .. 1,465 1,465 y :&yyV'1: SHItfe Wfilffli lung include 23 for which matched controls had not been interviewed when the investigation ended. Most o f our results therefore relate to the 1,465 cases paired with the 1,465 matched controls (" other diseases, A "). Cases diagnosed as carcinoma, or cancer, of the lung or bronchus, pleural endothelioma, and alveolar-cell carcinoma of the lung have been included in the " carcinoma of the lungi" group: Classification of the diagnoses according to the .recommendation of the International Symposium on the lEndemioIogy of Cancer of the Lung (Council for the International Organizations of Medical Sciences, 1952) gave Social class (RegistrarG eneral's categories): X II 39 -- 53 _ 165 -- 172 -- III 750 -- 720 -- IV 172 -- 198 -- V 231 -- 214 All social c'asses .. 1,357 -- 1,357 -- Place of residence: G reater London 791 O ther county boroughs .. 225 Other urban districts .. 275 R ural districts 155 Abroad 19 Ail places 1,465 900'-181 .> 213:* 164 - 7:\.K t 1,465% ^0% of the. first order of reliability (evidence from biopsyiM* o f the primary tumour, from operative, bronchoscopic, or radiographic examination of the primary tumour together .with cytological examination o f the sputum or biopsy of a ^econd^Q'tumour, or front necropsy), 29% of the second order p f t reliability (cytological examination o f the sputum alone or- operative, bronchoscopic, or radiographic exami nation without biopsy), and only 1% of the third order of w reliability (case history and physical examination alone or death certificate). The 70 cases classified in Table I as " carcinoma of other special sites " are all other cancers arising inside the chest (for example, sarcoma of lung and cancer of the medi astinum and of the trachea) and those other cancers o f the respiratory - passages, buccal cavity, and_ oesophagus for Which a possible relationship with smoking has at times been postulated. The 36 cases of " carcinoma of uncer tain primary site " include patients with carcinomatosis and some in whom there was doubt whether the lung growth was primary or secondary. patients leads automatically to an exact correspondence^ in sex, age, and p laces,of interview. Differences in socia('|i|fi status show no regular trend and are no greater t h a t f A m ight be due to chance (x2-- 5.28, n = 4 , 0 .2 0 < P < 0 .3 0 ).'.|^ On the other hand, the places o f residence reveal cot - siderable differences : fewer o f the lung-carcinoma group^ w ere residents o f Greater L ondon (x2= 17.22, n = ;4 ^ P < 0 .0 1 ). T he m eaning o f this inequality is considered'! ^ later (section on place o f residence). Since our observa- tions show that the consum ption o f tobacco tends to be greatest in L ondon the inequality w ill, if anything, have.J som ewhat reduced the contrasts we find between the'/ groups in their sm oking habits. In our previous reporWjig how ever, w e showed that the inequality w as unlikely toW'1'* be o f importance. ^ Assessment of Smoking Habits The difficulties o f acquiring and assessing accurately a,, The 1,278 " other diseases, B " include: (1) patients with smoking history and the measures taken to overcome them carcinoma of the stomach or large bowel interviewed in were discussed in our previous paper. It will be sufficient^ ijP -' ni EC. 13, 1952 AETIOLOGY OF CARCINOMA OF LUNG British -1273 Medical J ournal*. f repeat here that the patients were asked .(a).if they, had poked at any period of their liv e s; (b) the ages at which ley had started and stopped; (c) the amount they were in Ife habit of smoking before the onset of the illness which ad brought them into h o s p i t a l (d) the main changes in Seir smoking history and the maximum they had ever been i the habit of smoking ; (e) the proportions smoked in dpes and cigarettes ; and If) whether they inhaled. A test |as made of the accuracy o f the answers by cross- ixamraing 50, patients again, six months or more after their jfst interview, While there was, as expected, some varia- jility o f reply, we concluded that the data were reliable fjiough to indicate general 'trends and to substantiate hatcrial differences between, groups. , JjFortunately the difficulties of inquiry, have been reduced firough -the level o f taxation remaining almost . constant'hroughout the investigation. National figures show that the total jome consumption remained fairly steady--namely 213.7, 211.5, il3.8, 221.2 million lb. in the four years ending March 31, 1951 (Board of Trade, 1952), (The last major change in the standard fate of duty payable on tobacco was on April 16, 1947, when ft rose from 35s. 6d. to 54s. lOd. a lb.; on April 7, 1948, it fecame 58s. 2d.) I In the latter stages of the inquiry somewhat fuller Smoking histories were sought, and the last 557 lung(carcinoma patients (523 men, 34 women) and their matched controls were asked questions on: their :mse< of Is (a) different: types and brands of tobacco; (b) filter-tipped * cigarettes; (c) cigarette-holders; and (d) petrol lighters. The Amount Smoked The results of the inquiry provide a number o f ways in which the smoking habits of the patients can be, categorized ^ and compared. The simplest is the amount smoked lmmedi-^ ately before the onset o f the illness which brought -the patient into hospital, This, however, can be very mislead- ~ ing, since some persons--including heavy smokers--give.up smoking periodically, and it would be wrong to classifyThem > as non-smokers, merely because thev were interviewed. , during a period, o f abstention. Ip Table III, therefore,' ifthe comparison is made in a modified form: non-smokers are defined, as in our previous study, as persons-who diave , , never consistently smoked as much as one cigarette;,a day " for as long as one year; the smokers are subdividedbeither according to the amount they were smoking immediately before the onset of their illness or, if they had previously >r stopped smoking, according to the amount they i.were , smoking before they last gave up. This is-described: as - " the most recent amount smoked." t Table III shows that in both men and women there were fewer non-smokers and considerably more of the heavier (Table III.--Most Recent Am ount of Tobacco Smoked Regularly Before the Onset o f the Present Illness: Lung-carcmoma i - Patients and Matched Control Patients with Other Diseases . iv > Disease Group JMen: ^ . I; 1,357 lung-carcinoma patients (99*9%) .. p 1,357 control patients with other diseases (100%) -Women: ; 108 lung-carcinoma patients (100%) v 108 control patients with other diseases (100%) .. No. of Nonsmokers 7(0-5%) 61 (4-5%) 40 (37-0%) 39(54-6%) 1 Cig.~ 49(3-6%) 9,1 (6-7%) 14(13-0%) 18 (16-7%) 5 Cigs.- No. Smoking Daily* - 15 Cigs.- 25 Cigs-- 516(38-0%) 615(45-3%) 445(32-8%) 408 (30-1%) 299 (22-0%) 162(11-9%) 30 (27-8%) 22 (20-4%) 12(11-1%) 8 (7-4%) 12(11-1%) 1 (0-9%) ' 50 Cjg?. 41 (3-0%): 20(1-5%):: ( 0 0 [ Difference between proportions o f non-smokers and smokers-- M en: *=43-99, n = l, P<0*000001. W om en: x2=6-73, n = l , P<0*01. !; Difference between proportions o f smokers smoking different amounts---Men: 9=69*74, n = 4 , P < 0-000001. W om en: *= 8*99, n = 3 , 0*02<P<0*05. ; * Ounces o f tobacco have been expressed as being equivalent to so many cigarettes. There is. 1 oz. of tobacco in 26*5 normal-size cigarettes, so that the ^conversion factor has been taken as: 1 oz. o f tobacco a w eek= 4 cigarettes a day. > T able IV.-- M ost Recent Am ount of Tobacco Smoked Regularly Before the Onset o f the Present Illness: Lung-carcmoma Patients and Matched Control Patients with Other D iseases, Subdivided by Place and Date of Interview \ Disease Group Place and Date of Interview Carcinoma of lung /- (men) y Greater London, 1948-9 .. .Greater London, 1950-1 .. Bristo', 1950-1 ........................... Cambridge, 1 9 5 1 ............................ Leeds, 1950-1 ........................... Newcastle, 1950-1 .. .. 1 Whole investigation .. l: Greater London, 1948-9 .. Greater London* 1950-1 .. .. Y C ontrol patients with Bristo'. 1950 -1 ................... Ti other diseases (men) Cambridge, 1951 ... .. Leeds; 1950-1 .. .. . v i' Newcastle, 1950-1 .. Whole investigation .. i - - & & C arcinom a of lung (women) l; l>- Greater London, 1948-9 .. . . G reater London, 1950-1 .. .. Bristol, 1950-1 .. . . . ... C am bridge,1951 - . , . . Leeds, 19S0-1 ................. .. Newcastle, 1950-1 .. fev . , Whole investigation .. Control patients with other diseases ' |v ." (women) k- " .:M' Cireater London, J948-9 ... .. Greater London, 1950-1 . . Bristol, 1950-1 ........................... Cambridge, 1951 .. . . . Leeds, 1950-1 Newcastle, 1950-1 .. Whole investigation .. Percentage Non-smokers 1 Cig.- 0-3 1*0 2-8 0*0 0*0 .0-0 5*1 3*3 0*0 0*0 . 1-9 2-1 0-5 3-6 4*2 ,4*2 4*2 2*9 5*7 ' 5*3 8*5 4*9 2-8 8-8 7*5 t49 4 5 6-7 31*7 25*0 (50-0)t (50-0) (20-0) 68*4 : 11 7 15 0 -- (20-0) 15-8 37-0 : ' 13-0 ' - 53*3 55-0 130-0) (50-0) (60-0) 57-9 20 0. 10 0 (50-0) :' -. -- " 15*8 54-6 1 16-7 ' Percentage Smoking Daily* 5 Cigs.- 15 Cigs.- 38*5 38-2 22*5 - 50-0 47-2 37*3 30-2 31*4 ' 40*8 . 29*4 34-0 ' 39*3 ; 38*0 . 32*8 - 45*1 1 46*1 49*3 52 9, 39-6 " 43*9 29-3 28:4 . 28*2 . 29-4 35 8 34 0 . 4 45*3 30*1 31*7 15*0 20 0 10*0.: s- (50-0): - . - -- . - -- .......... (50-0) (60-0) -- 15 8 , T* 27 8 u*i 167 20 O . --- (50-0) . ,,(40-0) . 26 3 :: 20-4 10 0 10-0 -- -- - ---.a,-- 7-4 25 Cigs.+ . 25*9 26*1 33*8 20*6 17*0 21*3 . 25-0 12*9 16*3 15-5 5 9 11 3 11*9 13-4 100 30 0 --- . - ~ -- 11-1 , 0-0 t 5-0 -- --- *> * 0-9 See footnote to Table 111. T l he percentages m parentheses are based upon very small numbers and have no reliability. No;' ' Interviewed 649 706 71 34 53 244 1,357 649 306 i Y?* !'M Sw 53L3'. 'i i 244 I , 1'? (.n 20 i 2 *<' 2 ** 71 -v 5 ip.- (0 % 12 -' ,* ' 5 *\ 19 % ^ 108 - Km. 1274 D e c . 13, 1952 AETIOLOGY OF CARCINOMA OF LUNG British Medical Journal smokers among the lung-carcinoma patients than among the control patients. Amongst 1,357 men w ith . carcinoma of the lung, only 7, or 0.5%, were non-smokers; there were 61, or 4.5%, among the same number of men with other diseases. At the other end of the scale, 25% of the men with lung carcinoma had smoked 25 or more cigarettes a day (or the equivalent in pipe tobacco); the proportion in the control patients was only 13.4%. Similarly, amongst the 108 women with carcinoma of the lung, 37.0% were nonsmokers against 54.6% of the women with other diseases. Among women with carcinoma 11.1% had smoked 25 or more cigarettes a day ; among those with other diseases this proportion was 0.9%. These results are essentially the same as those recorded in our preliminary report. In Greater London, indeed, the results for the two periods 1948-9 and 1950-1 are remark ably alike (see Table IV, where they are set out separately and alongside figures from the provincial centres). Table IV also shows that the contrast between the smoking habits of the men in the two disease groups was observed quite con sistently at the provincial centres. In each case the propor tions of non-smokers and of men smoking less than five cigarettes a day were lower, and the proportion smoking 25 or more cigarettes a day was substantially higher, among the lung-carcinoma patients. In women similar results were obtained during the two periods of inquiry in Greater London, while at Bristol, Cambridge, and Leeds the num bers are too small to warrant attention. At Newcastle, however, the smoking habits of the female lung-carcinoma patients did not differ appreciably from those of their matched controls--in fact, the proportion of non-smokers was slightly higher in the lung-carcinoma group. Though the number interviewed was small (19 cases), the divergence from the experience of the lung-cancer groups elsewhere is sufficient to be statistically significant. Adding the pro vincial centres together gives the following figures for women: time they last gave up. Its advantage as a criterion is that ij information is easily obtained and likely to be reasonably accural Its disadvantage is that smoking habits vary over a iifetl$ sometimes considerably, and previous habits, which may be re) vant, are being ignored. We have therefore calculated oth quantitative estimates of the amount smoked as revealed in tl patient's history. These are (a) the amount smoked immediate before the patient's illness, (5) the maximum amount ever sniokt regularly, (e) the total amount smoked since smoking was begiS and (d) the average amount smoked daily over the 10 years pr< ceding the patient's illness, over the penultimate 10 years, ari over the whole of the patient's life since the age of 15, taking in) account recorded changes during these periods. " Qualitatively similar results are obtained whichever of thS estimates is used. The sharpest differentiation between the luh| carcinoma and control patients, for both men and women, appear, to be given by the average daily amount smoked over the %( years preceding the patient's illness. The results of this calcufii! tion are shown in Table V and differences between the groups .^ patients--particularly in the women--are more pronounced than those in Table III. i For men, the amount smoked immediately before the patient*! illness is equally good, but the estimates of the total amount consumed throughout life, the average daily amount since the age; of 15, and the average daily amount over the penultimate 10 year$ give differences only of the same order as those shown by " the? most recent amount smoked ''; the maximum daily amount ever,: smoked differentiates the groups less clearly. For women, all thei estimates except the amount smoked immediately before they patient's illness are more discriminating than " the most recent* amount smoked," though the maximum daily amount ever smoked') gives only a slightly increased divergence between the two groups;1;' In view of these results with varying measures of the:?smoking history we have used in subsequent tables the'; average amount smoked daily over the 10 years preceding;;; the patient's illness as the most appropriate criterion. A whole life history should perhaps be a truer measure off the " exposure to risk," but, as we pointed out in our ptevi-? ous paper, too much inaccuracy may result from requiring'' the patient to remember habits of many years past. Lung-carcinoma group Control group .. Nonsmokers . ! 16 . | 16 Smoking 1-14 Cigs. * 11 12 Smoking 15 4- Cigs. 1 -- Total 28 28 The presence of only one woman with lung carcinoma who had smoked 15 or more cigarettes a day is not, per haps, surprising; it seems that very few women in the provinces smoke so much. Of ail the 58 women inter viewed at provincial centres, and who suffered from " other diseases " (other than lung carcinoma, carcinoma of other sites possibly related to tobacco, and carcinoma of un certain primary site) none gave a history of smoking 15 or more cigarettes a day. In contrast, this amount was smoked by nearly 8% of 553 similar women with ail other diseases interviewed in Greater London. There is, however, an absence in the provinces of any difference between the two groups in the numbers of non-smokers and of women smoking fewer than 15 cigarettes a day. The " most recent amount smoked " will not necessarily give the best representation of a smoking history, even though defined, as here, to include the amount smoked by ex-smokers at the The Duration of Smoking Comparisons of the ages at which the patients reported that they began to smoke, the number of years they had smoked, and, when appropriate, the number of years since they last gave up are shown in Table VI. The lung-carcinoma patients are seen, on the average, to have begun smoking rather earlier, to have continued longer, and to have been rather less inclined to stop. In men, these differences are all statistically significant. In women they are not significant, but they are in the same . direction, and no less distinct, so that it is reasonable to ' accept them as real. The most pronounced difference appears in the number of years since smoking had last been given up. Since me control group contained more light smokers (Table 111) the higher proportion of ex-smokers in it here seen might, it was thought, be due to the fact that it is light smokers who more readily give up. In fact, the opposite appeared to be true. Of the 124 male control patients who had given up smoking nearly one-third (31.5%) were smoking 25 or more cigarettes a day when they gave up ; of those who continued to smoke, only 12.2% consumed as much. T able V.--Average Am ount of Tobacco Smoked Daily Over the 10 Years Preceding the Onset of the Present Illness; Lungcarcinoma Patients and Matched Control Patients with Other Diseases Disease G roup Men: 1.357 lung-carcinoma patients (99*9%) 1.357 control patterns with other diseases (100%) W om en: 108 lung-carcinoma patients (100%) 108 control patients with other diseases (100%) . . No. of Nonsmokers Less than 5 Cigs. No. Smoking Daily Average* of 5 C igs.- | 15 Cigs.- 25 C igs.- j 50 Cigs. 4- 7 (0-5%) 61 (4-5%) 40 (37-0%) 59(54-6%) 5 5 (4-0%) 129 (9*5%) 16 (14-8/0 25 (23-1%) 489 (36'0%) 570 (42-0% ) j 24 (22-/.%) 18(16-7%) 475 (35-0%) 431 (31-8%) 14 (13-0%) 6 (5-6%) 293 (21*6%) j 38(2-8%) 154(11*3%) j 12(0-9%) 14(13-0%) : 0 o ( 0 -0 % ) 1 o Difference between proportions of smokers smoking^ different amounts--M en ; r " ~ 93-77,_ n--4 , P<0-000001. W omen; xE=17-41, n = 3, P<0-001. * Ounces o f tobacco have been expressed as being equivalent to so many cigarettes. There is l oz. o f tobacco in 26-5 normal-size cigarettes, so that the conversion factor has been taken as : i oz. o f tobacco a week -- 4 cigarettes a day- )EC. 1 3 , 1 9 5 2 AETIOLOGY OF CARCINOMA OF LUNG British Medical Journal 1275 sle VI.-- Age at Starting to Smoke, Number o f Years Smoked, and Number of Years Since Smoking was Given Up, Lung-carcinoma Patients and Matched Control Patients Age at Starting i [ Lung-carcinoma ! Patients : No. % Control Patients No. I % U nder 20 1 1,077 79-8 20- ! 251 18-6 30- i 40 + 18\ 1-6 ; 4j 992 264 33\ 7/ AH ages | 1,350 100-0 1,296 *= 7-95; n = 2; O'Ol < t* < 0-02 76-5 20-4 100-0 No. of Years Smoking Lung-carcinoma Patients No. % Control Patients No. % 1- 1 2 \ 3*4 J 5 \ 6-2 10- 34/ 65 f 20- 746 55-3 725J 55-9 40 + 558 41-3 491 37-9 All periods 1,350 100-0 1.296 %a= 12*66; n = 2; P<0-01 1000 No. of Given Up Lung-carcinoma Patients No. % Control Patients No. 7. . 0- 1,280 94-8 1.172 90-4 1- 56 4-1 75 5-8 2io0-+ 68 \/ 1-0 22637/ 3-8 All periods 1,350 99-9 3,296 - Xs= 25 -87; n = 2; PC0-001 100-0 U nder 20 1 20 29-4 20- i 23 33-8 nen 30- 40-}- | \ i } 36-8 All ages 68 1000 12 24-5 15 30-6 .3} 44-9 49 100-0 J102040 + AH periods 11 42 \/ 38-2 36 52-9 6 8-8 68 99*9 188/\ 53-1 20 40-8 3 61 49 100-0 013020 + All periods 58 85-3 9 13-2 1-5 ?}. 68 100-0 43 83-7 6 12-2 4-1 49 100-0 .The Method of Smoking So far the only distinction we have drawn between nokers is in the quantity of tobacco consumed. There :c. however, qualitative differences which might be imporint--namely, whether the smoker inhales, smokes a pipe, ses a cigarette holder, smokes filter-tipped cigarettes, rolls is own cigarettes, or lights his tobacco with matches or petrol lighter. Inhaling.--All the smokers and ex-smokers were asked whether they inhaled (with the exception of three lungarcinoma and two control patients, in whom the question /as inadvertently omitted). Of 1,415 lung-carcinoma atients (men and women), 64.6% said yes and 35.4% said o ; of the 1,343 control patients with other diseases, 66.6% aid yes and 33.4% said no. The differences are negligible, iimilar results were obtained for men and women conidered separately. (Further consideration is paid to nhaling in the section dealing with site of tumour.) Cigarettes and Pipes.--Some persons usually smoke -igarettes, others usually smoke a pipe. Habits, however, jo not remain constant, and it has been necessary to divide nale smokers into three broad categories: (a) those who nave never smoked a pipe regularly for as long as one year (" pure cigarette-smokers ") : (b) those who have smoked cigarettes and a pipe ; and (c) those who have never smoked cigarettes regularly for as long as one year (" pure pipesmokers "). Among the 1,350 male lung-carcinoma patients who smoked, 3.9% were pure pipe-smokers and 74.4% were pure cigarette-smokers; among the 1,296 male control patients who smoked the corresponding proportions were 6.9% and 69.4%. The differences, though not striking, are statistically highly significant (% for the three groups, pure pipe, mixed, pure cigarette= 15.85 ; n = 2, P<0.001). It would appear that pipe-smoking is less closely associated with the development of lung carcinoma than cigarette smoking. Pipe-smokers, however, consume, on the aver age, less tobacco than cigarette-smokers, and this must account for some of the relative deficiency of pipe-smokers in the lungJcarcinoma group. It does not seem that it can account for the whole difference, since the proportion of pure pipe-smokers is somewhat lower at each level of tobacco consumption. The relevant figures are as follows: Percentage of Pure Pipe-smokers among all Smokers at Each Average Daily Consumption Level (Measured in Terms of Cigarettes) Less than 1 5 C igs.-- j 15 Cigs.-- 25 C igs.+ 5 Cigs. I M ale lung-carcinoma patients 9-1% i 7 2 7 1 1-37 Male control patients with i i other diseases i io-9.; . 10-5*.; 3-5% 2-)7 2-4% On the other hand, studying the pure pipe-smokers alone we find that 9.4% of those with Iung-carcinema smoked the equivalent of less than five cigarettes a day and 13.2% smoked the equivalent of 25 or more a day ; in the control group the proportions were 15.1% and 4.3%. In other words, a higher proportion of the pure pipe-smokers with carcinoma of the lung fall into the higher smoking categories --as with the cigarette-smokers. We conclude, as in our earlier report, that the method of smoking is of importance and that smoking a pipe, though also related to carcinoma of the lung, appearg to carry a smaller risk than smoking cigarettes (see also section on estimated risks). Cigarette-holders.--A possible explanation of this lower risk of pipe-smoking is that the pipe-stem acts as a partial filter of a carcinogenic agent. If that were so, we might expect that fewer of the patients with carcinoma of the lung had used cigarette-holders. We sought information on this point in the latter stages of the inquiry--from the last 523 pairs of male lung-carcinoma and control patients to be interviewed. Judged by the proportions of nonsmokers and pure pipe-smokers in the two groups, these last patients seem to be a representative sample of the total. Table VII shows results obtained from them. Few Table VII.-- Use of Cigarette-holders: Male Lung-carcinoma and Matched Control Patients. (Information Obtained During the Last Part of the Investigation Only) Disease Group 523 lung-carcinoma patients .. 523 control patients Never Non- Smoked smokers Cigar ettes . 4 35 26 30 Cigarette-smokers. Holders Used Total Cigarette- Never Occa sionally Regu smokers larly 479 15 10 504 413 27 27 467 9 patients had ever used holders, but the proportion of cigarette-smokers who had done so was significantly smaller in the group with carcinoma of the lung (5%) than in the control patients (12%) (x" = 14.74, n = 2 , PCO.OOl). This difference might merely be due to an association between using a holder and light smoking, but the avail able evidence suggests not. Among the 54 cigarettesmokers in the control group who had used holders five (9%) smoked an average of fewer than 15 cigarettes a day and six (11%) smoked an average of 25 or m o re; the corresponding figures for the 413 cigarette-smokers who had not used holders were 32 (8%) and 55 (13%). Further evidence that an association with light smoking cannot account for the relative deficiency of users of cigarette- 1276 D jx ^aS&M&w^rM ti\ O V / % ^ ' ^ rl ^ '`m-:,? r >r > JJl ft holdeu, in the lung-carcinoma. group is the -finding that the propci'lion who had used holders was lower at each level o f tobacco consumption--namely; Percentage o f Cigtirette-smokers who had -ever used H olders a t Each Average Daily "c, C o n su m p tio n 1Level vL5esCs itghsa.n . 5CigS; -- lS C ig s - 25 Cigs *+ Mpalawuenltuonwg huo rsumnookemdcaig% i r arettLS M ale control piticiiES .with- 5-9% r oihci diseases who smoked ugurclli& - * 13-5% 4-0% 12 6% , 't 6-0% 10 4% 4-5% 9 8% 2 \ p a o f O garc/to.--Cigarette-smokers were asked whether thc\ bought manufactured )cigarettes or bought tobacco and lolled their own. Of the 1,297 male iung- carcmonu. patients who had ever smoked cigarettes 20.7% smoked m oslk hand-rolled cigarettes; of the; 1,203 similar contiol patients the proportion was 19.1%. Evidently there is no spcufic association of manufactured, as opposed to hand lolled, cigarettes and carcinoma of the lung. It can also be concluded that the different risks associated with cigarette-- and pipe-smoking are unlikely to be the result of sthe .different types of tobacco consumed, as a number of men who roll their cigarettes use pipe tobacco. In view o f the presence of arsenic in American tobacco and -its almost complete absence from Oriental tobacco (Daff and Kennaway, 1950) it was clearly of interest to determine whether there was any difference in the propor tions o f American and " Turkish" tobacco smokers in the lung-carcinoma and control groups. The results o f such inquiry; were inconclusive because nearly all smokers had habitually smoked " Virginian." In fact, only one smoker was {found who had never regularly smoked it (a man of 70,(-under treatment for an enlarged prostate). Of the 504 male -lung-carcinoma patients who had smoked cigarettes only 3.8% said that they had, at some period, regularly smoked Turkish tobacco; among the 467 control patients thi figure was 4.5%. insjgniflcant. The difference is statistically During the last part of the investigation inquiries were disc made about the brands of cigarettes smoked and the uie- o f filter-tipped cigarettes. The results (Table VIII) show that- n o n e ;o f the four main brands recorded was more closely associated with carcinoma of the lung than another. The-proportions are remarkably similar in the two groups of patients. On the other hand, very few of the men with lung carcinoma had ever regularly used filter-tipped cigarettes Tabue-VIII.--Brands of Cigarettes Smoked and Use of Filter tipped:- Cigarettes: Male Lung-carcinoma and Matched Control Patients. (Information Obtained During the Last Part of the Investigation Only) - - 3 in 504 -compared; with 15 m ; the 467 controls- difference is significant (x2=7.74, n = l , P<0.01), b So few. observations the conclusions to be drawn muS highly speculative. The explanation may be that7? tipped - cigarettes are smoked predominantly by 'i smokers, ; but - we ha ve insufficient data to -test* possibility. I 1 U se o fP e iro lL ig h te rs.--In the two groups- of 523* patients last interviewed inquiry was made into the-iu, petrol lighters, Of the 504 male patients with carciM of the lung who smoked cigarettes,42.9% reported tbl some period they had regularly used petrol lighters -1 - 468,. similar control patients, -the proportion was,:yfflJ The difference is negligible, and the evidence is agil the hypothesis--often put forward-rrthat petrol lightersthe responsible carcinogenic agent. Use of Tobacco for Chewing and as Snuff--Althovj extraneous substances will be brought into contact', the bronchial mucosa more readily by smoking than, chewing or by snuff, it is possible that particles of tobab are inspired into the bronchial tree by these latter mea and that these uncombusted particles are carcinogenic, possibility is suggested by the frequent occurrence of caiill o f the buccal cavity m Eastern countries where the inhaS tants have the habit of chewing quids of tobacco mixef with flavouring agents. Questions about chewing tobacco and the use of snbff were asked of. 1,209 male patients with lung carcinoma aM of the 1,209 corresponding control patients.* The resulfiS show (Table IX) that fewer patients with lung carcinon|jj| Table IX.-- Use of Tobacco for Chewing and as Snuff: M a ftit Lung-carcinoma and Matched Control Patients. (Informaiifflm Obtained During Part of the Investigation Only) >ia8 Disease Group Tobacco Never Chewed Tobacco Chewed Occasionally Regularly Less than 10 Less than 10 10 Yrs. Yrs. 4- 10 Yrs. Yrs.-f* ToWSlff wfcefsjHB C hew M |j| 1.209 lung-carci noma patients.. 1.209 control patients 1,169 1,145 Tobacco Never Taken as Snuff 8 4 11 17 6 9 17 32 Tobacco Taken as Snuff 40 'iff -- Total i m who had^jpf. TakcaX^pi Tobaocp^^: as S nulT ^jl 1.209 lung-carci nom a patients.. 1,176 8 1.209 control patients ; 1,166 U 7 10 8 33 ' 'f f s 5 16 n 43 " rr{- Significance tests of differences between lung-carcinoma and control patients: 7 -S |fR Chewing: n l, 0-0KPC0-02. Snuff-taking: *2 = 1-36, n - 1, 0 -2 0 < P < 0-30. -VifiSf * ,Type o f S m o k er-4 Lung-carcinoma Patients Contro} Patients Cigarette-smpkers, smoking inanufac- itured cigarettes-- Prand,mamly smoked; Brand A ,, B ,, c .. v.t P * -Other brands .Mixed ,, ... 72 (18 2%) .. 11 (2-8%) 123(31-1%) * . 21 (5-3%) .. .... 36 (9-0%) .. v. 133 (33-6%) . All brands .. . . . . ' . 396(100*0%) - Brand n o t stated (present smokers of. hand^rolled cigarettes or pipes) 108 70 (19 4%) 14 (3 9%) 107(29 6%) 15 (4 2%) 39(10.8%) 116(32-1%) 361 (100-0%) 106 >yU l cigarette-smokers 504 - 467 Cigarette-smokers, filter-tipped cigar- dVCtteSJ'.-rV.'V':- *Eversm oked regularly* .. 3 15 N e v e r ; ,, .. .. . 501 452 ri\ - All cigarette-smokers 504 467 For one or more years. had chewed tobacco (40, or 3,3%, against 64, or 5.3%)7$ and slightly fewer had ever taken snuff (33, or 2.7%, agaimt-x^l 43, or73.6%); the differences are small though stefisticallfljii significant for chewing. - That this latter difference is likely to be real is borne out by the fact that it is more marked-dfs for those who had chewed regularly (28 to 49) than for" those who had done so occasionally (12 to 15), and mostmarked for those who had chewed regularly for more than 10 years (17 to 32). The number, of patients who,if were still in-the habit of chewing was too small for it to; be possible to assess the amount they smoked in compari-i son with others. It may be that'there were fewer lung--f, carcinoma patients who had chewed tobacco because men. who chew will smoke less. ; *After the completion, of the first part. of.the investigation these questions were temporarily omitted from the questionary, and % 148 male patients with lung carcinoma interviewed outside London and their corresponding control patients were not asked whether.,^ they chewed tobacco or used, snuff. , EC. 13, 1952 AETIOLOGY OF CARCINOMA OF LUNG British Medical-Journal 1277 ^Nature of the Carcinoma in Relation to Smoking fa Histological Type B.-. According to Wynder and Graham (1950), adenocarci- p ia o f the lung is less closely related to smoking than ! >Vother histological types of lung carcinoma. Primary : enocarcinoma of the lung is an uncommon condition, i f of particular interest, in relation to smoking, in that I ihas invariably been reported as being relatively com- I iner in women than in men. In the present series, all iients with a histologically confirmed diagnosis (approxi- ately 70% in each sex) have been divided according to me, with the following results: p 1 i|357 men .. p08 women Histological Type . Epidermoid Oat-cell or Adeno carcinoma Anaplastic Unclassi fied No Histological Confirmation 475 (52%) 303 (33%) 33 (4%) 105(11%) 441 18 (23%) 38 (48%) 10 (13%) 13(16%) 29 l l n Table X the numbers o f men and women smoking afferent amounts o f tobacco are shown separately for each histological type of growth and are compared with the lumbers expected from the experience of all male and Bl female patients in whom the diagnosis was confirmed histologically. There is no statistically significant differ|hce between the amounts smoked by patients in the different histological groups in either sex. The number pf proved cases o f adenocarcinoma is, however, too small (43) to conclude that no difference 'erat8*:N.Thef-weretfiit fact, relatively more non-smokers and very lig h t.smokers (average-consumption less than five cigarettes a day) among the patients with adenocarcinoma in both sexes, and it is possible that larger numbers would have supported Wynder and Graham's findings. Table X also shows that it is not possible to. detect any differ ence, in amount smoked, between the cases diagnosed clinically, and those in whom the diagnoses were histologically confirmed. This result suggests that the " clinical " diagnoses were generally accurate--which is not surprising, since they were based ;on find ings at thoracotomy or at necropsy in 93 (19.8%), on direct- observation of the tumour bronchoscopically in a further 96 (20.4%), while m many of the remainder bronchoscopic examina tion suggested the presence of a carcinoma though no tumour was seen. . =, Site of Tumour . ,, The site o f origin of a tumour within the lung may be of interest, since it is possible that aetiological agents 'knight reach the main bronchi but not the bronchioles and alveoli,r In the present series 1,154 (90.4%) were considered to have arisen centrally and 122 (9.6%) peripherally (in 189 it was not possible to decide). Analysis of the smoking habits o f the patients, in these two groups reveals no difference between them in the amounts smoked but a slight difference in the prevalence of inhaling. (See Table XI. The comparison, here does not allow for differences in age distribution o f the patients with central and peripheral growths, but these differencesare small and do not materially affect the results.) It will be seen that a slightly higher proportion of the males with peripheral growths inhaled regularly (62.6%) compared with |Table K= j: X. -Average Amount o f Tobacco Smoked Daily Over the 10 Years Preceding the Onset o f the Present Illness, Divided According to Histological T ype: Lung-carcinoma Patients Histological Type No- of Non-smokers ]:V-7 Average Amount Smoked Daily over 10 Years. N um ber Smoking: Less than, 5 Cigs. 5 Cigs.- 15 C igs.- 25 Cigs. -t-- " 1 A jEpidermoid carcinoma (475) .. iOat-cell or anaplastic carcinoma (303) /Adenocarcinoma (33) .. fType unclassified 0 0 5 ) ........................... f B :Histological evidence obtained (916) .N o histological evidence (441) .. C /Epidermoid carcinoma (18) Oat-cell or anaplastic carcinoma (38) ,. Adenocarcinoma (JO) .. Type unclassified (13) ............... 1 (2-4)* 2 0-8) 2 (0-2) 0 (0-6) 5 (4-9) 2 (2-1) 11 (8-6) 12(11-2) 5 (3-8) 1 (5-5) 14(19-3) 12(11-2) 2 (0-9) 8 (4-7) 36(34-5) .19(20-5) 1 (1-9) 6 (51) 2 (1-2) 2 (1-8) Men 169(166-9) 110(106-7) 7 (11-5) 36 (37-0) 322 (328-5) 167 (160-5) Women 4 (4-3) 9 (10-0) 1 (2-7) 5 (3-0) 175 (172-4) 105 (111-4) 16 (12-3) 38 (38-0) 334 (322-4) 141 (152-6) I (1-6) 7 (7-4) .0 (1-4) 4 (1-6) 116(114-13. 74 (72-0) 6 (8-2) 23 (24-8) 219(225-8) 112(105-2) 3 1 0-6) . ......... 4 . (4-3) 2 ' (1-0) V; 1 (1 1) D Histological evidence obtained (79) .. No histo'ogical evidence (29) .. 29 (28-7) 11 (11-3) 10(11-1) 5 (3-9) 20 (18-4) . 5 (6-6) 12 (10-9) 2 (3-2) 8 (10-1) 'v 6 (4-0) The figures in parentheses arc the numbers that would have occurred if the patients in the histological group in question had had, at each age, exactlyjlbe.. . same smoking habits as all the patients with which the group is being compared. T hat is, in part A of the table, ail male patients with histological .evidence. ' of the growth; in part B, all male patients; in part C, ail female patients with histological evidence of the growth; in p a rtD , all female patients. T a ble XI.- -Prevalence o f Inhaling: Patients with Carcinoma of the Lung Arising Centrally and Peripherally and Corre sponding Matched Control Patients Disease Group Regularly No o f Smokers Inhaling Occasionally Never - - Total No. of Smokers No of 3 , , Non-smokers., -.-t.... Male lung-carcinoma patients with: J,070 central growths 116 peripheral growths Male control patients corresponding to lung-carcinoma patients w ith: i 16 peripheral growths .. .. ... 558(52-4%) 72 (62-6%) 583(57-1%) , 63 (53-3%) 126(11-8%) 13(11-3%) 116(11-4%) 12(10-5%) 380(35-7%) 30 (26-1%) 322 (31-5%) - 39(34-2%) 1,064 (99-9%) . ... ...115(100% ) 1,021 (100%) 114(100%) 6 \ 'Ji' 1 1 1L ' ^4" 49 f * ' 2 Female lung-curcinoma patients with ! 84 central growths . . 24 6 6 peripheral growths -1 1 Female control patients corresponding to lung-carcinoma patients with 84 central growths .. .. . .. 14 5 6 peripheral growths .. 0 0 17 3 20 'T - ( 47 - - 5 f , 39 1 -- Significance tests. Prevalence o f inhaling (regular plus occasional): . _ _ . Mate lung-carcinoma patients, central compared with peripheral growths*. Xa = 4 `24, n = 1, 0,0 2 < P < 0 ,05... .. . ....... .... Male patients, patients with central growths compared with corresponding controls: y /~ -4-07, n.=*i, 0 -0 2 < P < 0 05 Male patients, patients with peripheral growths com pared with corresponding controls: y / --1*79, ft --1, 0*10 < !*<().20, ifift . __ 1278 D ec. 13, 1952 AETIOLOGY OF CARCINOMA OF LUNG ----- ----------- :------ :__ ___________ i____________________--------- --------- -------- - British 3'M Medical J ournai.-'l the men with central growths (52.4%)--a statistically signifi cant difference (0.05 level). The patients in each group can also be compared with their own matched controls. It is then found that the males with central growths include rather fewer regular inhalers than their controls (52.4 to 57.1%), while the group with peripheral growths contain rather more (62.6 to 53.3%). The difference is statistically significant in the former case but not in the latter, where the observations are too few to eliminate chance as an explanation. With the women the numbers are too small to warrant consideration. Estimated Risks Amount of Smoking To measure approximately the relative risks associated with different levels of smoking we need to know (a) the number of people smoking different amounts of tobacco in each age-group--that is, the numbers " at risk "--and (b) the number of people smoking different amounts in each age group who died from carcinoma of the lung. It would then be possible, because of the very high fatality of the disease, to equate the calculated death rates to the risks of developing it. The present investigation cannot provide estimates of these figures for the whole country, since the patients interviewed were drawn mainly from Greater London, and smoking habits and the lung-cancer death rate both vary between countryside and town. For example, in 1949 the report of the Registrar-General shows that the recorded death rates from lung cancer in men were 597, 521, 398, and 292 per million in, respectively, Greater London, county boroughs outside Greater London, other urban districts, and rural districts. For the Greater London area alone we may, however, proceed on three assumptions--namely, (a) that the smoking habits reported by the control patients without carcinoma of the lung who lived in Greater London at the time of their interview are, at each age and in each sex, typical of the inhabitants of Greater London generally ; (>) that the smoking habits reported by the patients with carcinoma of the lung, also living in Greater London at the time of their interview, are typical of the inhabitants who died of the disease during the period of the survey ; and (c) that the deaths attributed by the Registrar-General to lung cancer both in men and in women provide a reasonable estimate of the actual numbers of deaths due to carcinoma of the lung. On these assumptions, which are bold but, we think, not wholly unreasonable, we can calculate, for the one region, death rates for each level of tobacco consumption. The population of Greater London given by the Registrar-General for June 30, 1950, has been taken as the population at risk. The numbers of persons within this population smoking different amounts of tobacco have been estimated from the data for each sex and for each of the age groups 25-44, 45-64, and 65-74. Thus at ages 45-64 we had 932 male patients resident in Greater London with diseases other than lung carcinoma; 4.1% were non-smokers, 9.3% had smoked an average over the preced ing 10 years of fewer than five cigarettes (or their equivalent) a day, 42.6% an average of 5-14 a day, 29.9% an average of 15-24 a day, 12.8% an average of 25-49 a day, and 1.3% an average of 50 or more a day. The male population of Greater London at ages 45-64 was 937,000, and this population has been given the above proportions of non-smokers and smokers of different . amounts. (In making these estimates we ignore the lungcarcinoma patients in the total population, but their proportion is too small to make, any material difference.) In the same way the numbers of persons of each sex and age dying from lung cancers in Greater London in 1950 have been divided up on the basis of the smoking habits of the lung-carcinoma patients who were interviewed. The death rates thus obtained (Table XII) increase with both age and amount smoked. They pass from a negligible figure for male non-smokers aged 25-44 to a level of the order of 1 in 100 per year among men aged 65-74 who have smoked an average of 25 cigarettes or more a day for T able XII.--Estimated Annual Death Raies from Lung Candk per 1,000 Men and per 1,000 Women Living in Greats London; by Age Group and Average Amount of Tobaci, Smoked Daily in Preceding 10 Years* Annual Death Rate per 1,000 Persons and Age Average Amount Smoked Daily in Non- Preceding Ten Years smokers Less than 5 15 5 Cigs. Cigs.- Cigs.- 25 Cigs.- 50 Cigs.+ Lung-,)' carcinoma Inter- v viewed- Men 2 5 - ,. 0 0 0 f 0 0 3 0-13 0 1 2 0-17 0-52 45- .. 0 1 4 O'59 1-35 1-67 2-95 4-74 65-74 .. 0 0 0 2 38 2-66 3'88 6-95 10-24 Women 25- .. 0 006 0-04 0-03 6 5 - .. 0-09 0-06 0-34 45-74 . . 0-32 0-70 0-59 ____ , 013 _ J- 19 _ 2-37 -- - 61 539 a 130 . 39 v' 13 ___ * The reasons for the adoption of this measure of smoking habits isT explained in the text (see section on amount smoked.). ` A t R ates based on observation of fewer than five cases of carcinoma of theT Jung are given, in italics. .%/ the preceding 10 years. The greatest number of our sub-3 jects were aged 45-64, and the rates for this group should ,i therefore be the most reliable. In the Chart these are shown graphically against the amounts smoked. It appears, on ; the assumptions made, that the death rate increases in 7 approximately simple proportion with the amount smoked. Among women the death rate seems to rise more slowly, but the numbers are smaller and considerably less reliable. AVERACE NUMBER O f CICA RETTES SM OKED DAILY Estimated annual death rates from lung cancer in Greater London for men and for women aged 45-64, in relation to the average amount of tobacco smoked daily (measured in terms of cigarettes) in the preceding 10 years. Cigarettes or Pipe On the same assumptions, that our data are representa tive of Greater London, it is also possible to estimate mortality rates for each type of smoker--that is, cigarettes only, cigarettes and pipe, pipe only. As previously shown, the pure pipe-smokers are few and we can therefore calcu late rates only for the one broad age group 45-74, The results are; Estimated Annual M ortality Rates from LungCancer per 1,000 Men Aged 45-74. Average Amount Smoked Daily for Preceding 10 Y ears in Terms of Cigarettes* Pure cigarettesmokers Smokers of cig arettes and pipe Pure pipe-smokers All smokers ,. Less than 5 Cigs. Ml 0-87 0-951 1-04 5 Cigs.-- 15 C igs.- 25 Cigs.- 50 Cigs. + 1-71, 2-16 3-50 7-37 1-67 1-98 3-35 2-241 1-35 0-79\ 2 08t -- 1-66 2*05 3-42 5-42 * See footnote to Table III. t Based on fewer than 5 cases De c . 1 3 , 1 9 5 2 AETIOLOGY OF CARCINOMA OF LUNG Mtl),*''}" 1279 U each smoking level the estimated death rate of pipejkers is less than that of cigarette-smokers, and the differe increases with heavier smoking. In three of the four ups the death rate of those smoking cigarettes and pipe intermediate. With the amount of data at our disposal 1 the assumptions made in calculating these rates we uld be reluctant to draw any precise conclusion on the ative level of the risks. But it certainly appears that the Ics are less in pipe-smokers than in cigarette-smokers, d perhaps to the greatest extent in the heavier-smoking :egories. Town and Country There seem to be differences in smoking habits between ivnsmen and countrymen (briefly- referred to above), and :r data may be used to see whether they can wholly, or rtially, account for the reported different mortality rates >m carcinoma of the lung--though they are insufficient give more than an approximate answer. The patients terviewed in this inquiry lived in different parts of the untry. Grouping them according to place of residence, 3 can roughly estimate the smoking habits of persons dng in Greater London, county boroughs, urban districts, sons occupationally exposed to motor fumes or road dust do not appear more frequently in the lung-carcinoma group. The results will be published in full later. Social class has already been considered (Table I I ) ; it showed no significant difference between the male patients with lung carcinoma and their controls. This observation is in keeping with the Registrar-General's 1930-2 decennial supplement on occupational mortality, though in the pre sent inquiry the lack of association may be overemphasized from the fact that our control patients were usually, and deliberately, taken from the same hospitals as the patients with carcinoma of the lung. This designed equality in some respects may give an overestimated equality in social class. Place of Residence The Registrar-General's evidence that cancer of the lung is more frequent in the large towns than in the smaller towns and countryside suggests that a higher proportion of our lung-carcinoma patients would be expected to have been living in Greater London and the county boroughs. In fact, as shown in Table II, this was not the case. If, however, the patients' places of residence are analysed a b l e XIII.--Smoking Habits of Male Patients Living in Different Parts of the Country ; Divided According to Density of Population. (Standardized to age distribution of population of England and Wales, aged 25-74) Area of Residence Percentage of Non-smokers ireater London 51 otinty borough 6-8 >ther urban district ,. 8*4 lural district .. 104 Percentage Smoking Daily Average of Less than 5 Cigs. 5 Cigs.- 15 Cigs.- 25 Cigs.+ 8*3 38-3 33-7 14-6 6-6 42-7 34-0 9-9 13-3 37-1 32-3 8-9 13-7 40-8 27-6 7-7 Percentage of No. o f Patients Pure Cigarette Cigarette and Pure Interviewed -smokers Pipe-smokers Pipe-smokers 74-2 16-0 66-3 22-8 59-9 23-9 58-4 21-5 4-8 1*393 4-J 240 7-8 439 9-8 327 nd rural districts. There were, however, relatively few latients living in the country, and a special survey was herefore made (in February, 1950) of the smoking habits if 531 other patients, aged 25-74, admitted to hospitals in ural areas of Dorset and Wiltshire. To facilitate comparisons the male inhabitants of England md Wales between the ages of 25 and 74 have been taken is a standard population, and the smoking habits in this lopulation have been estimated from the incidence rates if smoking actually observed in the age groups in each of he four areas (Table XIII). It appears that as the place >f residence becomes more highly urbanized the proporion of non-smokers and of pure pipe-smokers decreases md the proportion of heavier smokers and pure cigarettemokers increases. Thus the changes are in the direction vhich would lead to a higher death rate from carcinoma >f the lung in the towns. Whether they can account for he observed differences in mortality is difficult to say. On the assumption that the estimated death rates of Table <11 should prevail equally in all areas, it would seem that he recorded differences in mortality between town and :ountry are greater than could be attributed wholly to the iifferences in smoking habits. In other words, the differinces in smoking habits shown in Table XIII are not lufficient to lead to a rural mortality rate which is only ibout half that of the large towns. Other Aetiological Factors The inquiry here reported was designed to throw light m any aetiological agent in carcinoma of the lung--for ixample, on substances which pollute the atmosphere. The questions on smoking were merely one facet of the investi gation. We now turn to other aspects. Occupation and Social Class Occupational histories were taken from all patients, but these reveal no gross association between any type of occupation and lung carcinoma which might indicate an aetiological agent of general significance--for example, per- separately, for those interviewed in Greater London and those interviewed in the provinces, a different picture is obtained (Table XIV). Of patients interviewed in Greater London, fewer with lung carcinoma lived there (76%) compared with the con trols (87 % ); more lived in each of the other types of area. Of patients interviewed in the provinces, more with lung carcinoma lived in the county boroughs (45% against 38%) and fewer lived in rural districts (22% against 28%). The differences in Greater London can reasonably be explained on the grounds that patients with cancer living outside London tend, more than patients'with other diseases, to come to London for treatment. In the provinces, how ever, the control patients with other diseases were inter viewed in hospitals which were deliberately chosen because, like the thoracic units, they also served as regional centres. Consequently there should not be on demographic grounds any deficiency among them of patients living in the smaller . towns or in the countryside. It would be reasonable to suspect that any difference between the places of residence T able XIV.--Place of Residence: Lung-carcinoma and Matched Control Patients, Subdivided by Place of Interview ~ i Place o f Interview Place of Residence Greater London Lungcarcinoma Patients Control Patients Provinces , :% i Lungcarcinoma Patients Control Patients % Greater London Other county boroughs .. Other urban districts Rural districts Abroad All places 791 31 . 133 62 , 18, 1,035 900 16 71 42 6 1,035 0 194 142 93 i 430 0 \ 165 , 142 122 1 430 ' Difference between places of residence: Patients interviewed in G reater London: x2=40'50, n = 4 , P < 0 ',001.. / ;i .: Patients interviewed in provinces (excluding patients residing abroad) l .- s *2=6-25, n = 2,`0-02< P< 0-05. U 1280 '-Dec. .13, 1952 v < .. i 2/ . g... *,r . AETIOLOGY OF CARCINOMA OF LUNG Medical JouRfdi, of patients in the lung-carcinoma, and .control groups ..would reflect differences in the relative Incidence o f the conditions in the different areas. The observation in the provinces that a smaller proportion o f the lung-carcinoma than o f the control patients lived in the country supports, therefore, the contention that lung carcinoma is less common in rural than in urban areas. Evidence can also be obtained by comparing, for patients living in a given type o f area at the time o f interview, the ptoportion who had previously lived, for any long time in the countryside (see Table XV). Among the lung-carcinoma patients living in Greater London at the time of interview, 4.3% had previously lived for 10 or more years in a rural district; among control patients living in Greater London T able XV.--Residence for 10 or More Years in the Countryv . .. aside: Lung-carcinoma and Matched Control Patients, Sub- . divided by Place of Residence at the Time of Interview .Residence at / Time of Interview Greater London.. Other county boroughs O ther urban, dis- . tricls .. Rural districts .. Abroad .. All places Lung-carcinoma Patients No. Inter viewed 791 Lived for 10 or More Years in a Rural District No. / 34 4-3 225 13 5-8 275 37 13-5 155 136 87-7 19 I -- 1,465 221 (15-1) Control Patients No. Inter viewed Lived for 10 or M ore Years in a Rura! District No. % 900 62 6-9 181 12 6-6 213 32 15-0 164 149 90-9 7 0 -- 1,465 255 (17-4) the proportion was 6.9%. For each type of area this proportion is lower among the lung-carcinoma patients. The differences, though slight, are consistent and in con formity with the previous conclusion. They are not, how ever, statistically significant (x!=6.45, n = 4 , 0.10<P<0.20). Residence Near a Gasworks Gasworkers have been reported as specially liable to carcinoma of the lung (Kennaway and Kennaway, 1947; /Doll, 1952). and it was therefore thought possible that residence near a gasworks with inhalation of its fumes might conduce to the disease. All the patients were asked whether they had ever lived near a gasworks and, if so, for how long. The results revealed no difference: 23.0% (337/1,465) of the lung-carcinoma patients and 21.5% (315/ 1;465) of the control patients had lived near a gasworks for a year or more (x2= 0.95, n = l , 0.30<P <0.50). This result agrees with that obtained by McConnell, Gordon, and Jones (1952). Exposure to Different Forms of Heating ' Further information on the possible effects of exposure V to coal-gas was sought by asking the form o f heating used in the houses in which the patients had lived. This ques;h tion also related to the possibility that exposure to benz- pyrene in the soot of domestic fires might be conducive* the development of carcinoma. Analysis was made ofsf kinds of heating used in the living-rooms of all the hoiii in which the patients had resided for three or more yea and the numbers of years were calculated that each patie had been exposed, in his living-room, to a coal fire, a j fire, an electric fire, an anthracite stove, a radiator, or ott form of heating. The results (Table XVI) reveal very lif difference in the histories given by the two groups : patients. Previous Respiratory Illnesses A large number of the control patients had some respii tory disease, and we clearly cannot assume that the histe of previous respiratory illnesses given by these patiei would be characteristic of other patients generally. ;< would not, therefore, be proper to compare the luti carcinoma patients with the general control groups patients to determine whether previous respiratory illness play any part in the aetiology of carcinoma of the lur We have accordingly compared lung-carcinoma patien with patients with other forms of cancer (mainly stomai and large bowel, and excluding those in whom the site origin of the growth was in no doubt and also those wi growths elsewhere in the chest, upper respiratory passage and mouth). The general cancer group was not selected to be of the sail sex and age distribution as the lung-carcinoma group, and it therefore necessary to allow for sex and age differences betwes them. For this purpose we have first calculated for all the cancpatients put together--lung, stomach, large bowel, etc.--tl reported incidence of previous respiratory illness in each sex ar 10-year age subgroup. These rates we have then applied to tl numbers of patients of corresponding sex and age with (a) care noma, of the lung, and (b) other forms of cancer, to calculate ho many cases of previous respiratory illness would have occurrc in each subgroup if both types of patients had had these san rates of attack. The total number expected for each illness w; then readily obtained by summing the numbers in each subgrou] The numbers, " expected " on the basis of equality, can be con pared with the histories actually recorded. To avoid bias due t any confusion between an earlier independent respiratory illnei (in which our interest lay) and an illness induced by the present of the tumour, we included in the analysis only such illnesses e had occurred at least five years before the interview. Occasional! illnesses occurring more than five years previously may have bee due to a slow-growing tumour, but the number is unlikely, w think, to be important. Questions were asked about the past occurrence of pneu monia, pulmonary tuberculosis, pleural effusion, chroni bronchitis, asthma, and chronic nasal catarrh. The result (Table XVII) show that the lung-carcinoma patients morfrequently had a history of preceding pneumonia or chronibronchitis, while other respiratory illnesses were referre< to with approximately equal frequency by the two groups The differences in the incidence of pneumonia and chronibronchitis are statistically significant (particularly the latter) though the actual proportion of lung-carcinoma patient: with positive histories is not large (with each disease 17%) r \ _ Table X VI.--Exposure to Different Forms o f Heating: Lung-carcinoma and Matched Control Patients. obtained during part o f the investigation only) (Informatioi I Type o f Heating in Living -room of Patients' Residence Disease Group . r Coal fire . . .. \ " 'S > r 4<3as fire . . .. \ Electric f ir e ........................... { R a d i a t o r ........................... 4 f r O ther heating .. " \ Lung carcinoma O ther diseases Lung carcinoma O ther diseases Lung carcinoma O ther diseases Lung carcinoma Other diseases Lung carcinoma Other diseases Lung carcinoma O ther diseases No. of Patients Exposed for Different Durations of Time Never l Y r.- 30 Y rs.- 50 Yrs.-f- 9 93 434 735 11 105 429 726 1,261 6 1,258 - 12 3 1 1 0 1,192 59 13 7 1,169 74 21 7 1,184 62 14 11 1,172 71 21 7 1,191 66 8 6 1,169 85 12 5 1,192 55 14 10 1,189 59 16 7 Total No. of Patients Test o f Significance of Difference 1,271 1,271 1,271 1,271 1,271 1,271 1,271 1,271 1,271 1,271 1,271 1,271 ;^ = 1`01, n = 3, 0-70<P<0*80 Combining all over 1 yr.: tfSs-0`39, n = l,0 -5 0 < P < 0 -T Combining all over 30 yrs.: jt2=3-25, n=2> 0-10<F<Q '2l Combining all over 30 yrs.: *2=0-84, n=2,0-50<P<0-7< Combining all over 30 yrs.: *2 = 2-89, n = 2 , 0-20<P<0-3( Combining al! over 30 yrs.: ' *2=0-17, n=2,0-90<P<0-95; ) hc. 13. 1952 AETIOLOGY OF CARCINOMA OF LUNG British M edical Journal 1281 jijBLE XVII.--Frequency of Occurrence of Respiratory Illnesses in the Past History; Lung-carcinoma and Other Cancer I? Patients* i, X Type of ;t Respiratory Illness tfhma ironic bronchitis ,. tironic nasal catarrh iural effusion ieumonia .. ilmonary tuberculosis No. of Patients 1,465 1,465 1,465 1,465 1,465 1,465 Lung-carcinoma Patients History of Illness 5 or More Years Previously Observed Expected No. % No. 19 1-3 19^ 254 17-3 222-9 198 13-5 189-2 25 1-7 25-7 250 17-1 224-8 11 0-8 13-1 No. of Patients 853 853 853 853 853 853 Other Cancer Patients History of Illness 5 or More Years Previously Observed Expected No. % No- 13 1-5 12-5 94 11*3 125-1 90 11-0 98-8 14 1-7 13-3 83 10-1 108-2 8 1-1 5-9 Test of Significance of Difference *2=0-033, n = l, 0-80<P<0-90 *2 = 14-13, n = l, P<0-001 *2 = 1-35, n = 1, 0-20< P<0-30 *2=0-057, n = 1,0-80< P < 0-90 *2 = 10-03, n = l,P < 0 -0 1 . *2=1-09, n = l , 0-20< P < 0-30 * Excluding cancer patients notified incorrectly as having lung carcinoma (see Table XIX). able XVIII.-- Frequency of Occurrence o f Previous Pneumonia and Chronic Bronchitis: Lung-carcinoma and Other Cancer Patients* Subdivided by Age and Sex j Age 5- .. 5- .. 5- . . 5- . . 5-74 Total .. Men Lung-carcinoma Patients Other Cancer Patients* No. of Patients History o f Illness 5 or More Years Previously Pneumonia Chronic Bronchitis No- of Patients History o f Illness 5 or More Years Previously Pneumonia Chronic Bronchitis No. % No. % No. % No. % 17 7 116 19> 18 i s } 16 46 493 83 17 72 15 109 545 103 19 107 20 213 186 26 14 30 16 151 15 14 20 9 18 12 < 11 10 27 13 22 15 1.357 236 -- 230 -- 526 61 -- 62 -- Women Lung-carcinoma Patients Other Cancer Patients* No. of Patients History of Illness 5 or More Years Previously Pneumonia BCrohnrochniitcis No. of Patients History of Illness 5 or M ore Years Previously Pneumonia BCrohnrochniitcis No. % No. % No. % No. % 3 7 15 3 } 22 3 } I2 26 2 } 12 0} 0 38 4 11 10 26 81 4 5 5' 6 34 3 9 7 21 101 5 5 13 13 18 3 17 4 22 112 9 8 14 13 108 14 -- 24 -- 327 22 -- 32 -- * See footnote to Table XVII. jetailed figures for the previous occurrence of these two iiseases are given in Table XVIII. The lung-carcinoma atients show a uniformly higher incidence of each disease n both sexes and in all age groups, though in men many )f the differences are quite small. In neither group, how:ver, does the frequency of a past history,, of pneumonia ncrease with age as would naturally be expected. It is not inlikely that some older patients forgot their earlier attacks >f pneumonia. This being so,' the differences found letween the lung-carcinoma and the other cancer patients nay merely have arisen because patients with a respiraory disease recall more completely their previous respiraory infections--that is, for the very reason which led us o exclude other respiratory illnesses from the control ;roup. This possibility can, however, be tested by com>aring the histories given by patients thought to have `arcinoma of the lung when they were interviewed but n whom the diagnosis turned out to be erroneous (Table <IX) with the histories given by the lung-carcinoma patients ind by the other cancer patients (Table XVIII). The incorrectly notified patients gave, in nearly every sex md age group, a history of previous pneumonia and chronic ironchitis of similar frequency to that given by the patients vith lung carcinoma and of greater frequency than that of he patients with other forms of cancer. For all sex and ige groups taken together the proportions with a positive ustory w ere:--pneumonia: incorrectly notified patients 20%, lung-carcinoma patients 17%, other cancer patients 10% ; chronic bronchitis: incorrectly notified patients 21%, lung-carcinoma patients 17%, other cancer patients 11%. When age and sex differences are allowed for (Table XIX) the incorrectly notified patients and the lung-carcinoma patients do not differ significantly, while the differences between the incorrectly notified patients and the group of other cancers are highly significant. It would seem, there fore, that the more frequent history of a preceding attack of pneumonia or of frequent attacks of chronic bronchitis in the lung-carcinoma group may well result from patients with respiratory symptoms recalling their previous attacks Table XIX.--Frequency of Occurrence of Previous Pneumonia and Chronic Bronchitis: Patients Incorrectly Thought to have Lung-carcinoma, Subdivided by Age and Sex Age 25- .. 35- .. 45- .. 55- .. 65-74 .. All ages Patients Incorrectly Thought to have Lung Carcinoma Men Women No. of Patients History of Illness 5 or More Years Previously: Pneumonia BCrohnrochniitcis No. of Patients History of Illness 5 or More Years Previously: Pneumonia BCrohnrochmiteis No. % No. % No. % No. % 17 48 .1} 2= ?} 14 14 9 79 11 14 10 13 25 98 25 26 25 26 15 25 3 12 9 36 5 4 } 22 4 16 3 } 20 5' 20 20- 267 55 -- 53 -- 68 13 -- 16 -- Comparison with patients proved to have lung carcinoma Patients incorrectly thought to have Jung carcinoma: No. giving a history of previous pneumonia .. No. expected to give such a history . . .. *2=1-64, n -1 , P -0 -2 0 ' No. giving a history of previous chronic bronchitis No. expected to give such a history . . .. *2= 1-20, n = l , 0-20< P < 0-30 .. .. .. .. .. .. .. . ... Comparison with other cancer patients Patients incorrectly thought to have lung carcinoma: No. giving a history of previous pneumonia . .. .. No. expected to give such a history .. .. .. *2= 13-84, n*=l, P < 0-001 No. giving a history of previous chronic bronchitis .. .. - No. expected to give such a history ... .. .. .. *2=21-43, n = l, P<0-001 68 59-9 69 61-99 68 48-22 ~ 69 44-5. of certain common respiratory illnesses more readily than other patients (though it is possible that the Iung-cancet and the incorrectly diagnosed groups both have suffered a greater frequency of such attacks in the past). On the present evidence we feel unable to deduce any aetiologies! relationship between lung carcinoma and previous respira tory illness. 1282: Dec, 13, 1952 AETIOLOGY OF CARCINOMA OF LUNG - ,, mmTM, Y " ----------------- --------- -------- --------------- --______________ ___ _______________ ____________ ____________________________ MSDICAL JODKN^i Validity of the Results The larger and more detailed data that we have pre sented here confirm those in our preliminary report, and . support our conclusion that there is an association between smoking and carcinoma of the lung. Other explanations ,/of the.figures might, however, be possible, and we care fully considered several alternatives in our earlier paper-- namely, () that our group with carcinoma of the lung was unrepresentative of patients with that disease ; (b) that the : patients in the control group were not truly comparable . with the lung-carcinoma patients ; (c) that the method of . selection of the control patients had led to the choice of : patients who smoked less than the average; (d) that the lung-carcinoma patients tended, because of their disease, to exaggerate their smoking habits; (e) that the interviewers .. tended to scale up the smoking habits of the lung-carcinoma patients ; and (/) that the individual interviewers might have obtained different results and have interviewed different proportions of the patients in the various disease groups. The main points of the arguments that we put forward against these- alternative explanations are illustrated in Table XX, which, utilizing the various groups of control patients, shows the proportions who smoked different amounts of tobacco, standardized according to the age distribution of the population of England and Wales, between the ages of 45 and 74. In this standardization the incidence rate of smoking in a given age group in a given sample of patients was applied to the number of men--or women--in that age group in England and Wales to give the numbers of smokers in the genera) population. The summation of the figures thus derived from each separate sex/age group gave the total distribution in the standard population of smokers of different amounts--which were then converted into a percentage distribution as shown in Table XX. The two age groups under 45 years were omitted from the calculations because there were few such patients in some disease groups, and these unreliable figures would have been given undue weight in the process of standardization. Table XX shows (a) that the lung-carcinoma group con tains a smaller proportion of non-smokers and light smokers and a higher proportion of heavy smokers than any of the other disease groups, and (b) that the proportions in the group of patients incorrectly thought to have carcinoma of ' the lung and in the groups with other respiratory diseases, with cancer in other sites, and with other diseases, are similar. We would, as before, lay special stress upon the group of patients believed by the interviewers to have carcinoma of the lung at the time of interview, but who proved finally not to have that disease. This group reveals T able XX.--The Smoking Habits of Patients in Different Disease Groups, Standardized According to the Age Distribution of the Population of England and Wales Aged 45-74, at June 30, 1950 \ > Disease Group Percent age of Nonsmokers Percentage Smoking Daily Average of Less than 5 Cigs. 5 Cigs.- 15 Cigs,- 25 Cigs.+ No. of Patients Inter viewed Aged 45-74 Men Carcinoma o f lung .. 0-3 Patients incorrectly thought to have car cinoma of lung 5-3 Other respiratory dis eases ^ 1-9 Other cancers 4-6 Other diseases 5-6 4-6 35-9 35-0 24-3 1,224 9-9 35-5 37-8 11-4 202 9-9 38-3 38-7 11-2 301 9-4 47-2 26-0 12-8 473 9-0 44-8 26-9 13-7 875 Women Carcinom a o f lung . . 40-6 13-7 22-0 9-5 14-2 90 Patients incorrectly thought to have car- . emoma o f lung 66-9 16-4 12-7 4-2 Other respiratory dis eases .. .. '66-5 22-4 0-0 1M Other cancers 68-4 14-3 11-0 5-0 O ther diseases 55-9 22-1 17*5 3-6 0-0 45 0-0 25 1-3 294 0-9 157 a distribution of smoking habits very similar to that show by the other groups but very different from that of tl lung-carcinoma patients. ' These observations make it unreasonable, we sugge's (a) to attribute the results to exaggeration by the Luni carcinoma patients, since patients with other respiratol diseases would presumably be equally inclined to exagg^i ate their smoking histories; (b) to attribute the results i bias on the part of the interviewers, since patients wh wesg believed by them to have lung carcinoma but wh were finally proved not to would have been recorded, ha there been bias, as having smoking habits similar to tti patients proved to have lung carcinoma ; (c) to attribute ,th results to some special selection of control patients wlj, were, on the average, light smokers, since there is no impot tant difference between patients without carcinoma of th! lung who were notified to us--that is, the incorrectly diag nosed group and the greater part of the other cancer groiij --and patients who were selected as controls by the inter' viewers on the basis laid down (that is, the respiratory disease group and the great majority of the " other diseases group). The further extensive data collected since the publican tion of our first report have proved to be essentially similar to the earlier data, and we have obtained no subsequent information to throw doubt on the validity of the conclu sion that there is a real association between smoking and carcinoma of the lung. Some figures for the smoking habits of the general popu lation of Great Britain in 1951 have been obtained, for other purposes, by the Social Survey of the Central Office of Information, and it is of interest to see how these com pare with the habits recorded by our hospital patients. The survey of the general population was made in May, 1951,, and was based upon interviews made by trained investiga tors with a representative random sample of the civilian population of Great Britain aged 21 years and over. The , names of persons to be interviewed were selected at random from local records in a representative sample of about 100 urban and rural local authority areas of different types t throughout Great Britain. The sample thus adequately 5 represented the population in respect of sex, age, and all other relevant factors. The questions about smoking were i part of an extensive investigation concerned mainly with | other subjects ; they were, however, drawn up in such a way as to permit the replies to be compared with those y of our inquiry. To avoid difficulties of geographical variations in smoking ; habits we have limited the comparison to patients and indi- viduals in the Social Survey sample who were resident in ' Greater London--from which the greater part of our data ,1 came. To eliminate differences due to variation in smoking " habits with age we have taken the age distribution of the y Social Survey sample as a standard and adjusted the patients to that age distribution before calculating the percentage ; distribution of the different grades of smokers (by, as usual, applying the incidence rates of smoking in the patients at each given age to the Social Survey population of that age and summing for the age groups). Persons of 75 and over were excluded. Table XXI shows that the percentages of non-smokers and light smokers among women agree closely, but ,it appears on the whole that somewhat heavier smoking habits have been recorded amongst the patients than in the sample of the general public. Differences in the dates of th e interviews (April, 1949, to February, 1952, in the one case, May, 1951, in the other) are unlikely to have influenced the results appreciably, as the national consumption of tobacco varied very little over that period. It may be argued that the differences indicate an association between smoking and many se p a ra te diseases; they may, on the other hand, result from the different methods of interviewing and from the different groups of interviewers employed. The comparison cer tainly provides no evidence that the association we have tec. 13, 1952 AETIOLOGY OF CARCINOMA OF LUNG Mtn i2'83 LU XXI.--Comparison Between Smoking Habits of Patients Without Carcinoma of the Lung, Interviewed in the Investi gation, and of the General Public, Interviewed by the Social Survey Investigators; Residents in Greater London Only Subjects Percent age of Nonsmokers Men ients with diseases ither than lung car- inoma 7-0 nple of general pub ic (Social Survey).. 12*1 Women ;ienis with diseases Mher than lung car cinoma 54-7 nple of general pub ic (Social Survey) .. 52-9 Most Recent Amount Smoked. Percentage Smoking: i 5 15 25 Cig.~ Cigs.- Cigs.- Cigs.-f 4-2 43-3 32-1 13-4 7-0 44-2 28-1 8-5 13-0 21-6 8-2 2-5 16-9 24-3 4-7 1-2 No. Inter viewed 1,390* 199 456 255 * This num ber is three less than that shown in Table X III, because the tients interviewed in the special investigation in rural hospitals are not iliided. 5served between smoking and carcinoma of the lung can ; attributed to a selective choice, for comparison with the mg-carcinoma patients, of other patients who tended to e light smokers. Discussion In discussing the data of our preliminary report we conluded that there is a real association between carcinoma if the lung and smoking, but pointed out that this is not lecessarily to say that smoking causes carcinoma of the ung. " The association would occur if carcinoma of the ung caused people to smoke or if both attributes were ;nd-effects of a common cause. The habit of smoking was, lowever, invariably formed before the onset of the disease as revealed by the production of symptoms), so that the iisease cannot be held to have caused the h a b it; nor can ve ourselves envisage any common cause likely to lead 30th to the development of the habit and to the develop ment of the disease 20 to 50 years later. We therefore conclude that smoking is a factor, and an important factor, in the production of carcinoma of the lung." Investigations in Germany (Muller, 1939 ;. Schairer and Schoniger, 1943) and in the U.S.A. (Schrek, Baker, Ballard, and DolgofE, 1950; Wynder and Graham, 1950; Levin, Goldstein, and Gerhardt, 1950; Mills and Porter, 1950) have led to very similar conclusions. In Britain, McConnell, Gordon, and Jones (1952) found no difference between the proportions of " non-smokers " in 100 lungcarcinoma patients and in 200 control patients (collected at a later date). On the other hand, they showed a con siderable difference among cigarette-smokers between the proportions smoking the larger quantities: 44.1% of the lung-carcinoma group had smoked more than 20 cigarettes a day against 23.2% of the controls. The present analysis of nearly 1,500 cases, or more than double the number dealt with in our preliminary report, supports the conclusion then reached and has revealed no alternative explanation--for example, in the use of petrol lighters. It has been suggested that subjects with a particular physical constitution may be prone to develop (a) the habit of smoking and (b) carcinoma of the lung, and that the association might therefore be indirect rather than causa! (Parnell, 1951). We know of no evidence of such a physical constitution characteristic of patients with lung carcinoma. If it does exist we should still have to find some environmental factor to account fo r th e in creased incidence of the disease in recent years. To say that smoking is a factor' in the production of carcinoma of the lung is not, of course, to say that it contri butes to the development of all cases of the disease. All observers agree that the disease occurs in non-smokers, and in the present series there were 34 such cases confidently diagnosed and histologically proved. These 5 men and 29 women said that they had never smoked at all or had smoked so little as never to have consumed as much as one cigarette a day for as long as one year. Such patients in the ordinary course of their lives must have inspired air containing tobacco smoke, and it is not possible to say whether the disease would occur in its complete absence. The reasonable presumption is that it would. Experience of cancer in other sites (for example, cancer of the skin) indicates that it is unlikely for one environmental agent to be the effective cause in all cases. Whether smoking is the sole cause of the increase in the disease in recent years is another matter. There is no evi dence to show whether there has been an increased inci dence among non-smokers. It is certain, we think, that some, if not much, of the increase is spurious" and merely the result of improved diagnosis. Rigdon and KirchofT (1952) have pointed out that in the U.S.A. the death rate attributed to lung cancer in the different States is positively correlated with the number of physicians .per 1,000 inhabi tants, and it can be shown, from their figures, that this is true independently of the correlation with cigarette con sumption--which is also related to the number of physicians per 1,000 inhabitants. The extent to which the increase is real does not, however, affect the present evidence from which the association between smoking and lung carci noma is deduced. It would be material if it were proved that there had, in fact, been no true increase of the disease following a great increase in the consumption of tobacco ; but that, in our opinion, is as far from having been proved as that the recorded increase is all real. The position, then, as we see it, is (a) that an association has been demon strated--here and elsewhere--between tobacco-smoking and carcinoma of the Jung; (b) that, independently of this evi dence, there has been a recorded increase over the years in the number of deaths attributed to the disease and an increase has also occurred in the consumption of tobacco, and particularly of cigarettes ; (c) that the increase in the number of deaths recorded is relatively greater than the increase in the consumption of tobacco, but the actual rela tion between the real increase in the number of deaths and the increase in the consumption of tobacco is entirely a matter of conjecture. Needless to say, environmental factors other than tobacco may be responsible for part of the presumed increase. The part played_ by any such factors cannot, however, be deduced merely from a contemporaneous increase in their incidence and in the death rate from lung cancer--that is, on the basis of a correlation in time: a direct association between the disease and exposure to them must be demon strated. In the present investigation some additional, but not very strong, evidence was obtained that lung carcinoma is commoner in urban than in rural areas, but otherwise we found no major or clear association apart from that with the consumption of tobacco. We should perhaps point out that we made no inquiries which would throw light upon a relationship between the development of lung carcinoma and an attack of influenza during the pandemic of 1918-19--a relationship from time to time suggested in the literature (see Lancet, 1951, 2, 737). It would be difficult to gather sufficiently accurate informa tion on such a disease after the passage of so many years. We may note (a) that no appreciable increase in lung cancer has occurred in Iceland (Dungal, 1950), though the influenza, pandemic was severe there, and \b) that, in Britain, influenza in 1918-19 affected both sexes almost equally (as judged by mortality), while deaths from lung cancer occur predomi nantly in men. Influenza (D eath R ate per Million) M -F . 1918 3,360 1919 1,350 2,967 1,101 Year 1948 1949 Lung Cancer (D eath R ate per Million) M F 422 83 " 453 86 / 1284 Dec. 13, 1952 AETIOLOGY OF CARCINOMA OF LUNG British " i Medical Jouri^ All methods of smoking tobacco do not, according to our results, carry equal risks. As in our previous report, smoking a pipe appears to be less closely associated with the disease than smoking cigarettes. In the present observa tions we have found that, contrasted with other patients, rather fewer of the patients with carcinoma of the lung have used a cigarette-holder or smoked filter-tipped .cigarettes. These observations are of interest, though it is impossible to draw any firm conclusions from them ,- since so few people have limited themselves throughout their lives to one method of smoking. It must also be remembered that any smoking technique which is differenti ally associated with a tendency to light smoking will neces sarily, according to our figures, appear to be less closely related to lung carcinoma. Nevertheless, it seems possible, from these results, that pipes, cigarette-holders, and filter . tips may, to some unknown extent, each partly separate out an active agent before it reaches the respiratory tract. On the other hand, the observation that patients who recognize that they inhale are found no more frequently in the lung-carcinoma group than in the control group appears somewhat paradoxical. We pointed out, however, in our preliminary report that until the size of the particles carry' mg the carcinogen is determined nothing can be stated about the effect which differences in depth of respiration . may have on the extent and site of deposition of the carcinogen. From the present extended observations it seems that patients with growths of central origin inhale less frequently than normal (though the difference is very small), while patients with peripheral growths may inhale rather more trequently. Such a finding could be expected if smoke when not " inhaled " were to penetrate mainly to the large bronchi while inhaling spread the deposition of smoke particles more evenly throughout the bronchial tree. From the evidence collected about each patient's smoking habits it has been possible to compute some estimates of the risks of dying from carcinoma of the lung in different age groups at different levels of tobacco consumption. These estimates, we would emphasize, are speculative and dependent on the validity of three assumptions: namely, first, that the smoking habits of the lung-carcinoma patients interviewed in this inquiry and resident in Greater London are representative of the habits of all persons dying of the : disease in Greater London in 1950; secondly, that the - smoking habits of the patients with other diseases inter viewed in this inquiry and residents of Greater London are representative of the habits of the general population ; of Greater London in 1950; and, thirdly, that the actual '- numbers of deaths from carcinoma of the lung in each ' sex in Greater London were close to the numbers recorded by the Registrar-General. The only one of these assump tions upon which we have a check is the second. Observas tions on smoking habits made by the Social Survey upon : a cross-section of inhabitants of Greater London do not 'differ radically from the observations based upon our ... patients, though they reveal rather fewer heavy smokers than we have found. Keeping these assumptions in mind, our estimates indicate that the risk of dying of lung carcivhoma increases with age, as is of course known, and in approximately simple arithmetical proportion with the '.'am o u n t'sm o k ed . A test of the truth of this conclusion is to see whether ; it accords with the observed incidences of lung carcinoma - and of tobacco consumption in different sections of the community and in different parts of the world. In this country there is a pronounced difference in the smoking - habits of men and w om en; it appears inadequate to account - for the whole excess of cases which occur in men, as is -shown by the different death rates estimated for each sex a't each level of tobacco consumption (Table XII and the -.Figure, based upon it). On the other hand, the sex ratio of the relatively few cases observed in non-smokers is not incompatible with a similar incidence in men and women in the absence of smoking--it could not be identical because of risks associated with certain industrial occupations. There is also an appreciable difference in smoking haj between men living m town and country, which shS lead to a higher death rate in the towns, but not to '! extent that is actually observed. -I It may be that other factors also operate to prod'i these different death rates in men and in women, in toy and in the countryside. On the other hand, the relatij ship between smoking and lung carcinoma is quite li to be more complex than that depicted in Table XII. I example, common observation would suggest that soi women--especially the light smokers among them--th to hold their cigarettes in their mouths less continuoui than men and not to smoke them to the end. An eqf number o f cigarettes a day may not therefore be the sat thing, from the point of view of cancer risk, when smok by men and. by women. Again, differences in consumptu of cigarettes in town and country may have been more pr nounced 10 or 20 years ago than they are to-day. We ha' no information to guide us. Stocks (1952) has, however, shown a distinct relatic between the size of a town--assessed by the number c occupied dwellings--and the mortality from lung cance It would seem likely that some agent other than tobacc (present perhaps in domestic chimney smoke or in th exhaust fumes of cars) is at least partly responsible fo the excess mortality in towns. Comparisons between the recorded death rates at differen epochs and in different countries present many difficulties ii interpretation. Varying standards of diagnosis of the causi of death, differences in methods of preparation and con sumption of tobacco, changes in consumption over the las 50 years, may all introduce gross errors. Nevertheless some correlation between national figures for cigarette con sumption and the death rate from lung cancer would be expected, though in view of the uncertainties of the data the correlation is unlikely to be high. Studying the inci dence of lung cancer in some European countries, Daff, Doll, and Kennaway (1951) conclude: " The consumption of tobacco per head has been for the last 10 years rather higher in Switzerland than in the United Kingdom, and in Norway has been about one-half that in the other two countries, while the crude death rates at the beginning and end of the period were roughly in the proportion of 10 (England and Wales) to 5 (Switzerland) and 2 (Norway). Cigarette consumption was approximately in the propor tion of 4 (England and Wales) to 2 (Switzerland) and 1 (Norway), and was more in accord with the relative death rates. The increase in the number of deaths has been about the same (twofold) in all three countries, but the increase in consumption of tobacco and cigarettes has been less. The differences in the incidence of cancer of the lung are therefore quite different in extent from those in the quantity of tobacco consumed; they are more like (though still different from) those in the quantity of cigarettes consumed. The study of the relation between the national consump tion of tobacco and the national incidence of cancer of the lung has scarcely begun." No responsible agent in tobacco smoke has been detected. The suggestion that arsenic, introduced by insecticides sprayed on the growing crop, might be a factor would seem to be discountenanced by the absence of arsenic from Turkish tobacco and the high proportion of lung carcinoma found at necropsy in Istanbul (Daff, Doll, and Kennaway, 1951; Saglam,. 1944). Benzpyrene has not been detected in cigarette smoke (Waller, 1952). Longcontinued exposure of mice to atmospheres containing tobacco smoke has failed to produce lung cancer (Camp bell, 1936 ; Lorenz, Stewart, Daniel, and Nelson, 1943), but it may be of significance, in view of our findings, that the smoke was brought to the mice through long tubes. In contrast, tumours of the skin of mice can be produced by the application of tars obtained from tobacco burnt at temperatures which occur in normal smoking (Lamb and Sanders, 1932; Flory, 1941). Goulden, Kennaway, and Urquhart (1952) have pointed o ut that carcinogenic agents D e c . 13, 1952 AETIOLOGY OF CARCINOMA OF LUNG Br itish,.... . MrniCAi Journal IPSiS tobacco smoke and in town dust might supplement one a day (or the equivalent in pipe tobacco).* T he corre ndodtihtievre, aefnfdecthsavoef stuwmomcaarrizceindogthene sa. vaAilanblaegednattainabtooubtacthcoe s1p3o.4n%di.ng Ffoigrurweofmorenththe emsealperocpoonrttrioolnspawtieernets 1w1.a1%s onfolyr 1 moke might be by itself only weakly, if at all, carcinogenic, the carcinom a group and 0.9% for the controls. 111 ut might act as a co-carcinogen in the presence of, for xample, the benzpyrene in urban atmospheres. For men these differences are present consistently in each o f the five areas o f inquiry. For women they are I (.1 Summary present in London but not in the 28 patients observed;in the provincial tow ns, where only one w om an with i In an investigation designed to throw light on the the heavier sm oking habits was found. |etiology of carcinom a of the lung nearly 5,000 hospital 'patients have been interviewed by fou r specially appoin ted alm oners. T he interview s took place in the years 948 to 1952 in hospitals in London, Bristol, Cambridge, Leeds, Newcastle-upon-Tyne, and (for a limited purpose) In the rural areas o f D orset and Wiltshire. The questions asked covered a very wide range, including the occupa tional histories o f the patients, where they had lived and the form s of dom estic heating in their homes, their pre vious attacks o f respiratory illnesses, their habits with regard to sm oking, and, for other purposes, som e parti culars o f their dietary habits and use o f purgatives. | Preliminary figures on tobacco-smoking were pub lished in a previous paper (D oll and H ill, 1950). The (present paper gives corresponding data for the whole iof the material collected as well as the analysis o f other questions included in the inquiry. The main compari sons are between 1,465 patients w ith carcinom a o f the lung and an equal number of " matched control " patients with other diseases, each of these being carefully chosen so as to be o f the same age, the same sex, and, so far as possible, in the same hospital at the sam e time as a lung-carcinom a patient. Estimated death rates for Greater London (on assump- . tions stated in the text) indicate that the mortality from carcinoma o f the lung m ay increase in approximately ' simple proportion with the amount smoked. Amongst men of ages 45-64 the death rate in non-smokers is negligible, while in the heavier-smoking categories it is estimated to reach 3 to 5 deaths per annum per 1,000 living. ...... \ - 's Regular users o f petrol lighters were found with equal j frequency 'in patients with lung carcinom a and in the control patients with other diseases (42.9% and 41.3% ); V the proportions w ho said they inhaled were similar (64.6% and 66.6% ); and so were the proportions of cigarette-smokers who had smoked mainly hand-rolled ' cigarettes (20.7% and 19.1 %). On the dther hand, rather fewer of the cigarette-smoking patients with lung carci nom a had ever used a cigarette-holder regularly (5% against 12% o f the controls), and only 3 out o f 504 had sm oked filter-tipped cigarettes, compared w ith 15 out of 467 controls. The observations in these respects are too few for a definite conclusion, but conceivably they may have a bearing on the appreciably lower risks reported here for pipe-smokers compared with cigarette-smokers. There is no appreciable difference between the two Each of these methods o f smoking might partly separate groups in the number o f persons belonging to the out an active agent before it reached the respiratory Registrar-General's five social classes, and no associa tract. tion has been found between any type o f occupation and lung carcinoma which would suggest the presence of an aetiological agent likely to be of general significance. The validity o f these various results is studied, and it is concluded, that the association between smoking and carcinoma o f the lung is real. It is not argued that tobacco sm oke contributes to the developm ent o f all With regard to the possible effects o f fumes in the cases o f the disease-- a m ost unlikely event. It is not atmosphere, both w ithin and w ithout the hom e, there is argued that it is the sole cause o f the increased death no significant difference between the groups. O f the rate o f recent years nor that it can w holly explain the lung-carcinoma patients 23.0%, and o f the control different mortality rates between town and country. patients 21.5%, had lived near a gasworks for a year or more. Their use o f coal, gas, or electric fires or other forms o f heating in the living-rooms o f their homes did not differ appreciably. In conform ity w ith the national death rates rather fewer o f the lung-carcinoma patients lived, or had lived, in the countryside and more, correspondingly, in the towns. ACKNOWLEDGMENTS The following hospitals co-operated in the investigation: , Bristol.--Frenchay (departments of plastic and jaw surgery, .,T neurological surgery, and thoracic surgery). Cambridge.--Addenbrooke's (radiotherapeutic centre) and.,i;: Papworth Hall. Leeds.--General Infirmary (thoracic department). ' . The lung-carcinoma group in comparison with patients with other form s o f cancer more often gave a history of a previous attack of pneumonia and o f chronic bronch itis. Detailed analysis o f the data suggests, however, that this difference may be due merely to the lungcarcinoma patients, with their respiratory symptoms, recalling more readily titan other persons their previous London.--Brompton, Central Middlesex, Fulham, Hackney, Hammersmith, Harefield, Lambeth, Lewisham, Middlesex, Mount: Vernon and the Radium Institute, New End, Royal Cancel, Royal Free, St. Bartholomew's, St. Charles's, St. James's, St. Mary's* '53V . |J f St. Thomas's, University College, and Whittington. , " " Newcastle-upon-Tyne.-^;General Infirmary (department of neurological surgery) ..and Shotley Bridge General (departments of% plastic surgery, radiotherapy, and thoracic surgery). " ' jj; attacks o f respiratory illness. The data are not accurate enough for an aetiological relationship to be postulated. O f the 1,357 men with carcinoma o f the lung 7, or 0.5% , were non-smokers (as defined in the text) ; o f the Wiltshire and Dorset (in the survey of habits of rural patients only)._Blandfdrd, Bridport General, Dorset County, Odstock, Poole General, Portwey (Weymouih), Salisbury General, Swanage, ' % Victoria (Wimborne), Westminster Memorial (Shaftesbury), Wey-< - ? mouth and District, and Yeatman Memoi ial (Sherborne) (, 108 w om en there were 40, or 37.0 %. The corresponding figures for their paired controls were 61 m en (4.5% ) and 59 w om en (54.6%). O f the men with lung carcinoma 25.0% reported that they had been sm oking, before the onset o f their illness, an average o f 25 or more cigarettes We are indebted to the medical staffs of these hospitals for * (, allowing us to interview their patients and to have access to the ' riVjj hospital notes; also to the individual members of the staffs;*,., medical and lay, who notified the cases and collected the case- notes for examination. The work could not have been carried , out without their co-operation. 1286 D ec. 13,1952 AETIOLOGY OF CARCINOMA OF LUNG "We are most grateful for advice on the interpretation of patho logical specimens to Professor W. G. Barnard, Drs. A. B. Bratton ami J. W. Clegg, Professor VY. D. Newcomb, Drs. D. M, Pryce, R. J. Sandry, A. C. Tliackray, K. Weavers, and R. R. Wilson; and for assistance in determining the final diagnoses in the patients interviewed to Dr. A. Adelstein, Mr. G. P. Alexander, Mr. P. R. Allison, Mr. R. C. Bell, Mr. R. Belsey, Dr. J. R. Bignall, Dr. D. M. J. Burns, Mr. R. Dobson, Dr. J. Faulkner, Mr. G. M. FitzGibbon, Mr. G. K. Harrison, Mr. G. A. Hendry, Mr. F. I. Herbert, Mr. E. Hoffman, Mr. G. A. Mason, Professor J. S. Mitchell, Mr. K. Price, Dr. G. O. Richardson, Mr. C. A. Ross, Mr. W. M. Ross, Mr. G. F. Rowbotham, Dr. K. R. Stokes, and Dr, G. Strang. Sir Ernest Kennaway and Dr. Percy Stocks took part in a conference called by the Medical Research Council, at which the investigation was initiated, and we have been fortunate in having their helpful advice throughout its course. Dr. John Knowelden and Dr. Ian Sutherland kindly read our draft report, and we have benefited by their criticisms. We are also grateful to Dr. W. P. D. Logan, chief medical statistician of the General Register Office, to the Government Social Survey, and to the Statistics Division of the Board of Trade for the provision of data, and to the Statistical Department of the Ministry of Labour for preparing the punched cards used in the mechanical analysis of the results. Finally, we wish to thank Miss Marna Buckatzsch, Miss Beryl Iago, Miss Keena Jones, and Miss Rosemary Thomson, who inter viewed the patients and, together with Mrs. Joan Bodington, analysed the results, They worked untiringly, and at all hours, and the main burden of the work fell on their shoulders. RrniKEN'CES Board of Trade (1952). Accounts reiating to Trade and Navigation of the United Kingdom, January, 1951, and January, 1952, H.M.S.O., London. Campbell, J. A. (1936). Brit. 1. exp. Path., 17, 146. Council for the International Organizations of Medical Sciences (1952). To be published. Daff, M. E., Doll, R., and Kennaway, E. L. (1951). Brit. J. Cancer, 5. t. ----- and Kennaway, E. L. (1950). Ibid.. 4, 173. Doll, R. (1952). Brit. J. industr. Med., 9, 180, ----- * and Hill, A. B. (1950). British Medical Journal, 2. 739. Dutigal, N . (1950). Lancet, 2, 24-5. Flory, C. M. (1941), Cancer Res,, 1, 262. Goulden, F., Kennaway, E. L., and Urquhart, M. L. (1952). Brit. ,/. Cancer, 6, 1. Kennaway, E. L., and Kennaway. N. M. (1947). Ibid., 1, 260. Lamb, F. W . M., and Sanders. E. (1932). J. Hyg., Loud., 32, 298. Levin, M. L., Goldstein, H., and Gerhardt, P. R. (195CD. J. Amer. med. Ass., 143, 336. Lorenz, E., Stewart, H. L., Daniel, J. K.. and Nelson. C. V. {1913\ Cancer Res., 3, 123 (abstract). McConnell, R. B., Gordon, K. C. T., and Jones, T. (1952). Lancet. 2. 651. Mills, C.. A., and Porter, M. M. (1950). Cancer Res., 10, 539. Miiller, F. H. (1939). Z. Krebsiorsch., 49, 57. Parnell, R. W . (1951). Lancet. 1. 963. Rtgdon, R. H ., and Kirchoff, H. (1952). Texas Reports on Biology and Medicine, 10, 76. Saglam, T. (1944). Bull. Fac. Med. Istanbul, 7. pt. 2, 3793. Scbairer, E., and Schoniger, E. (1943). Z. Krebsiorsch., 54, 261. Schrek, R., Baker, L. A., Ballard, G. P., and Dolgoff. S. (1950). Cancer Res., 10, 49. Stocks, P. (1952). Brit. J. Cancer, 6, 99. Waller, R. E. (1952). Ibid., 6, S. Wynder, E. L., and Graham, E. A, (1950). J. Amer. med. Ass., 143. 329. The centenary of the birth of Antoine Henri Becquerel. the French physicist, falls on December 15. Like his father and grandfather before him, he occupied the chair of physics at the Mus6e d'Histoire Naturelle, Paris, to which he was appointed in 1892. Three years later he became professor of physics at the Ecole Polytechnique. In 1896 he made the discovery that uranium ores at ordinary temperatures emit an invisible radiation similar in several respects to x rays, and capable of fogging a photographic plate after passing through metal plates, Pierre and Marie Curie investigated this radiation and showed it to be an atomic phenomenon. They named it radioactivity. The discovery of radium by the Curies followed in 1898, and they shared the Nobel Prize for Physics with Becquerel in 1903. Becquerel also made important contributions to the know ledge of magnetism, polarization of light, phosphorescence, and absorption of light in crystals. He died at Croisic, Brittany, on August 25, 1908. PSYCHIATRIC DISORDERS IN LATER* LIFE* R. S. ALLISON, M.D., F.R.C.P., D.P.M. i Physician in Charge of Neurological Department, p ov '/ Victoria Hospital, Belfast ' yal -S fr ; " Grow old along with me ' , The best is yet to be, :4 The last of life, for which the first was made;" A s Robert Browning's words imply, it is probable thail psychogenic factors causing sim ple anxiety and hysteri cal reactions play less part in later life than they do aP* earlier ages. But, as Sir H um phry Rolleston remarked!? quoting another authority: " Whatever may be said m favour o f it, old age is a losing game." There is the ? high incidence o f cerebral vascular disease, and mtra- cranial tumours occur with som e frequency. Affective disturbances are com m on. But 1 should like to refer' chiefly to the two principal organic syndromes seen m later life: the syndrome o f acute clouding of conscious- ness, delirium , or confusion, and that of presenile ' dementia. There is particularly the relation of the ? former to biochem ical and toxic factors and of the latter ' to intracranial disease. 1. Biochemical and Toxic Causes of Psychiatric Symptoms in Later Life Great as are the differences between the age periods of growth and decline, both the very young and the very old show an undue mental instability in response to disturbances in their body chemistry. Thus, an elderly person who ? exhibits at the most only slight mental signs of ageing, may, through the fortuitous introduction of a biochemical or . toxic factor, develop acute mental symptoms indicative of ; clouding of consciousness. In increasing order of severity i. these are; (1) Simple impairment of attention, registration, and recall, often associated with a facile or euphoric affect j. and push of talk. (2) Perseveration showing itself in the spontaneous behaviour, in the speech, and in the response to simple requests. (3) Disorientation in time, place, and for persons, usually at some stage with confabulation. In y many instances, concrete orientation may be correct or the \?v time sense only lost, but further testing will reveal evidence of abstract disorientation. (4) In more severe cases there ,v are grossly defective grasp and delirium with noisy and yy irrational behaviour. Recognized causes of such (usually reversible! effects are toxaemia from fever or urinary infection ; the incautious use of bromides, morphine, barbiturates, or other drugs ; and vitamin deficiency. Other causes which, although known, are less generally appreciated include dehydration, alkalosis, anoxia, and hypoglycaemia. 'I Dehydration y If a sufficient fluid intake is not secured dehydration will ,y arise in any feverish condition or after a surgical operation y; (such as prostatectomy) requiring general anaesthesia. The y statement that the patient has " passed urine " has little y weight in this respect unless it is known that the output is y adequate and the specific gravity satisfactory. The after- | care of old people undergoing operation for cataract also deserves note. Hallucinations and disorientation or acute y; delirium may occur as the result of a combination of unfavourable circumstances such as recumbency, bandaging *Read in opening a discussion in the Section of Psychiatry ai the Joint Annual Meeting of the British Medical Association and Irish Medical Association, Dublin, 1952.