Tobacco Documents Online

Smoking, passive smoking and histological types in lung cancer in Hong Kong Chinese women n o i~ r c~ Shis r.tateriai tr.g'y bo votecw by ccp9Ti6n't T.H. Lam', I.T.M. Kung2, C.M. Wong', W.K. Lam3, J.W.L. Kleevens', D. Saw4, C. Ili9ttilt+e 17 Us. Coeo' S. Scneviratne9, S.Y. Lamz, K.K. Los & W.C. Chan• pepwrments oI'Convnrutity Afedicinr. 3Pathology: sAledicrne, fJnivrrsity of Hong Kong; ' Queen Elcabeth Hospitol m+d' sKowloon Hospital, Hong Kong. i Sumt.ary In a ese control study io Hong Kong, 445 asea of Chinese fetmk lung cancer patients all oonfirmed' pathologiully wae compared with 445 Chinese femala healthy neighbourbood controls matched for age. The predominant histological type was adenocarcinoma (47.2'/.)., The relative risk (RR) in ever /' smokers was 3.91 (P<0.001. 95'kCl-2:g6, 5.08): The RRs were wtiraicaDy ngnificaatiy raised for all taajor aell types with significant trends between RR and amount of tobacco smoked daily. Among never smoking women. RR for p.aivc smoking due to a smoking husband was 1.65 (P<0.01, 95yL'C1 - 1.16, 2.35) with a tignificaat trend betwxo RR and amount smoked daily by the httsbaad. When broken down by mll types: the numbers were wbcuntiall only for adenocardnoma (RR-212. P<0:0I, 95%Cl - 1.32; 3.39) with a significant trsnd between RR and atttount smoked daily by the Lusb.nd:, The restilts taygsst that passive smoking is a risk faetor for lung aneer; partuntl:r)y adeaonrdnoroa in Hong Kong Cbinae wamen who never smoked. In Hong Kong, lung cancer is the major cause of death in both males and females. In 1985; there were 2,223 deaths attributed to malignant neoplasms of the ttachea, bronchus and lung (ICD 9th Revision Code 162) which actounted for 29.5'%, of deaths due to all forms of cancer, 1,457 in males. (31.7'/.)iand 766 (26.0%) in females (Director of Medical 8 Health Services of Hong Kong, 1986). On a world scale, male lung cancer death rates are not particularly high in Hong Kong. However, the femak rates are among the highest in the world with an age-standardized incidence rate of 23:4 per 100,000 in 1974-1977 (Waterhouse et a1., 1982), resulting in an unusually low male to female ratio. The most common cell type in males is squamous cell carcinoma (33.3'/.) and in females, adenocartanoma (49.6•/.). (Kung et al., 1984):! A case control study in 1976-1977 confirmed the relationship between lung cancer and smoking in maks. but in females about half the lung cancer patients were found to be non-smokers, of whom two thirds were suffering from adenocarcinoma (Chan er af., 1979). Further studies on passive smoking and other risk factors have been nrried out in Hong Kong but they failed to throw much light on the causes of lung cancer in never smoking females (Chan & Fung. 1982;; li,am et aL, 1983; Koo et al., 1984; Koo er a1,, 1985):. The present study aimed w answer the following questions: 1. Is smoking a major risk factor for lung cancer in Hong Kong Chinese women and' if so, what is the relationship between smoking and the histologicat types of lung cancer? 2. Is passive smoking due to a smoking husband a risk factor for lung canaer in Hong Kong Chinese women . who have never smoked themselves and' if a, what is the relationship between passive smoking and histological typeT Materials tuti tttKtiods A standardized structured questionnaire was designed for interviewing both cases and controls. The questions on Correspondence: T.H. Lam. Department of Commuaity Medicine. University of Hong Kong, Lf Shu Fan Building, S Sassoon Road; Hong Kong. Remved 17 March 1987; and in rsviaed form, 17 June 19E7. smoking habit were modified from those of the Questionnaire on Respiratory Symptoms of the Medical Research Council (1966). The subject was asked whether she smoked, or had ever smoked as much as one dgarette a day (or one cigar a week or one ounce of tobacco a month), for one year. If the reply was negative, we checked again by asking a further question on whether she had ever smoked' any amount of any type of tobacco at all in her whole life up to the time of the interview. Bxatue of very few positive rrsponses to this additional qttestion, we were tktisfied that under-reporting of the smoking habit was not a major problem. As elsewhere, an ever-smoken was defined as one who had ever smoke& as much as one cigarette a day or equivalent for as long as a year. If a subject had ever smoked, questions on the type of tobacco and amount usually smoked per day, age when smoking started regularly and for ex-smokers only, age when smoking was given up permattently, were asked: A never-smoker was defined as one who bad never smoked as much as one cigarette a day or equivalent for the duration of one year. The smoking history of the subject's husband was ascertained in similat way if the subject was married. The same definitions of ever- and never-amoker were used for the husband. A women was considered exposed to her, husband's tobacco smoke if she had livrd together with her smoking husband in the same hottsehold for at least one year continuously. If the htuband was an ever-smoker, information on the type of tobacco and amount usually smoked per day, by the husband and the duration of exposure was obtained. The questionnaire also trontaitud sectiotu on demographic and other variabks. It was tested; atrtended and fmalised before use in the study. Eight government or govcrnmrnt- assisted hospitals in which most of the lung canar patients were treated in Hong Kong granted us pcrntisaion for interviewing of patients. During the interviewing phase of the study, we intended to include all lung cancer patients of the eight hospitals whose d'tagnosis was baaed on strong chnico-radiological criteria and' with histological and/or cytological confirmation. Patients admitted to these hospitals who were suspected byy the hospital t:linicians to have lung cancer or who had already been given a eonf rmed diagnosis of lung cancer wene interviewed as soon as possible after their adatission, before their physical condition deteriorated. Only patients with their d'ugnosis confirmed by a pathologist's report(s) t:

, 1W 674 T.H. LAM er a!. 1 were considered only as suspected cases and they were followed up after being interviewed. Only those who subsequently had a pathology report confirming the diagnosis of lung cancer were included. Those without such confirmation were not included in the present study. The pathology report was required to state unambiguously that the patient was suffering from lung cancer before it was accepted. Information on cell type if available, was noted. Cases without information on cell type or unclassified because of undifferentiated tumours were grouped' under 'others and' unclassified'. The few patients with rare tumours such as carcinoid were excluded. Because these hospitals were visited frequently by the interviwers so that all eligible patients would' be interviewed other than the few patients who declined; to co-operate or were too ill; we believed that we had misse&only very few eligible patients. For each case, a healthy female control matched for age ( t 5 years) living in the same neighbourhood of the case was interviewed. The procedure of control selection was that when a patient was interviewed and included as a pathologically conf rmed case. the age and address of the case was noted. The interviewer then went to the address of the case and started to visit the nearest neighbourhood addresses until she found a woman who appeared' healthy and was within 5 years of age of the case. A few questions on present state of health were asked to check that the subject was indeed healthy and if so, the same questionnaire was completed. Thus the controls were matched for sex, age and place of residence. Interviewing took place between 1983 and 1986, and involved expenenced female interviewers. The language used was mainly Cantonese. Each interview took about 30min to complete. Cooperation of interviewees was good and non- response was rare ( - I •/.). The present paper presents the findings on the smoking history of the subjects themselves and for the never-smokers, the history of passive smoking due to a smoking husband were included as cases. Patients with a provisional diagnosis Resnlts (amount smoked daily multiplied by duration)L Because of the similar results and space fimitation, only the results on all forms of tobaeco. with single women included and by amount stnoked' daily arts reported in the present paper. risk factor. Fisher's exact test (two-sided) was used to check whether the RR was significantly different from unity. xi test for linear trend was performed to test whether therr was a trend between RR and the levels of exposure (Breslow & Day. 1980). Subjects with missing data were excluded from the analysis. We carried out separate analysis on cigarette only or on all forms of tobacco, by including single (never-married) women or by excluding thsat, by amount smoked daily, by duration of exposure or by total amount of exposure (CI) (Woolf s logit limits) were calculated for each level of Table IV shows the RR for p¢sstve smo tng lue to a Thii-ty four percent of the cases were confirmed primarily by bronchial or lung biopsy. l2'/. by lung resection. 8% byy lymph node biopsy; 9•/, by pkural biopsy. 17% by sputum cytology. 12% by pleural fluid cytology, 6% by bronchial aspirate. brushing, etc., 0.2% by autopsy and 2% by other methods. The distribution of the cases by cell type and by smoking history is shown in Table I. The distribution of cell types differed somewhat according to the basis of diagnosis. Resection and pleural biopsy yielded 70% adenocarcinoma while other methods resulted' in 30-35% adenocarcinoma. Bronchial and lung biopsy resulted in -30%% while other methods resulted in about 10•/% squamous cell carcinoma. A comparison of cases and controls by age and place of residence confirmed' that they were similar in the two matching variables. The mean age of the cases was 65.6 years (s.d. 11.2 years) and that of the controls was 65.3 years (s.d., 10.9 years). Comparison by other demographic variables showed' that the cases and controls were comparable in place of birth, duration of' stay in Hong Kong, level of education, marital status, and husband's occupation. Thus, by matching the controls with the cases by age and residence, a high degree of comparability, was achieved with regard to many other demographic variables. Table II shows the Relative Risks (RR) by history of ever- smoking and cell types. Among the cases for allicell types combined, $4.5% were ever-smaken and 45.5% were never- smokers whereas among the controls, the corresponding percentages were 23 9°/, and 76. t•/,, overall RR fo 't ~trer-smolting wras l gI 36e RRs was'lfi~it~lttt~"!>tiseQi>~+ (~--of °tLe 4` oell ° typi~;~ bdtii"`>bs'gb& ' " xnait~ ap~' ~i~s(1tR~.12A0)ti~~olb.vtd ~RRgg,rdl_araYwma (RR- ~.93) att~. ~~'aoma (RR ~ Ii.87). Mnbk iIl shows the RR by amount of tobaxo tmtokedr"~ Four hundred and forty-frve cases and 445 rnntrols werc ~~eJl ' 3ignifXanYly~~etrr for all~ included. Relative risks (RR) and 95`/% confidence intervals types combined and for each of tbe <aU types T.Me 1 Distribution of oel) type by sawking habit of usa and eomparison with Kuaa er af i(1984) series ~ Syyorwtts tdf cmrinonw Snta/J cell emcwawm .4denocmcinorna L.arsa cell ca.cinoina Otherr and tncl4uified Total t y n: '~L. r•. '. n: •/. w % e 'A n •/. ~, Present Series Never smoker 2E 30.4 9 17.6 131 62.4 9 45.0 25 34.7 202 45.4 t,ICI ~ Ever smoker 63 69.5 42 E2.4 79 37.6 11 55.0 47 65.3 242 54.4 ~. • Missins data (' 1.1 - - - - I 02 . W Total 92 100.0 S t 100.0 210 100.0 20 100:0 72 100:0 443 100.0 ('h of 444 nm)i (20.7) (11.5) (47.2) (4.5Y (16:2) (f00.0) ~ Series of Kung et af. (1984) 77 43 169 34 16 341 (•/% of 341 casex) (22.6) (12.6) (49.6) (10.0) (5.3) (100.0) smoking husband and cell types. Single (never married) women were treated as non-exposed to husband's smoking. The RR was 1.65 for all cell types combined. For individual cell types, the numbers were too small to be statistically significant except for adenocarcinoma, with a RR of 2.12. Table V shows s!!te RR„foc. passive smoking amo 1.. ~ Ilk ! ~ ~ ;. ~ :7 ~.o h significant RR or trend was found for other cell types and the details are not reported here. Because similar results were obtained when single women were excludedL these are also not reported. it should be noted that the proportions of singk (never-married) women in the cises and controls was 6.B'/, and 5.2% respectively. /

-. .. . ...~r.. .. ~,v..v nv...~ •.., •L,~L, .,V...,L. t T.Me 11 Htstory of ever smoking (all focros of tobaocol m 444 cases and 443 controls. by ceH types Srnokin8 kuton• oJ svbjecrs Cell type Case No Yes Contrad' No Yrs Relative n.tk (Q 9S•/. CI) P Sqnamous call'orcinoma 28 63 72 20 8.10 <0.001 (4:16, 15.77) Small ota) carcinoma 9 42 36, 14 12.00 <0.001 (4.65.,30.98) Adenoearcittoma 131 79 1W 51 1.87 < 0.01 (1.23: 2.85) Lrie orll arrinoma 9 11 17 3 6.93 <0.05 (1.53, 31.39), UtDers and unclassified 25 47 54 18 5.64 <0.001 (2.71, 11.60) AI1I oell types 202 242 337 106 3.81 <0.001 (2:86. 5.08) Notrs , For each cell type. the cases were compared wnth their matched controls. One cue and 2'eoottols with missin8 data on smoking wers toac9uded: TsYk !O Amount smoked daily (all forms of tobacco) in cases and'controls by cell types All cell typts Sq<,mnos<r cell'corcinoma Amount smoked daih• br arD}rcu Cau Relative risk Control (Q 95% CL) P Case Relative riik Conrrol (d 9J•/. Cl) P Nil 202 337 1 28 72 11 1-10 101 63 2.67 23 11 5.38 (117: 3.83) <0.001 (2:32: 17,16) <0:001, 11-20 90 28 5.36 28 6 12.00' (3.39. 8.48) <0.001 (4.49. 32.10) <0:001 ! 21+ 39 9 7.23 10 1 25.71 (3.43:15:24) <0.001 (3.14, 210.30) <0:001' Total 432 437 89 90 Test for trend xs_89.5. P<0.001 ri-41.96, P<0.001 Small nll carcbioma Adenocarcbtoma Amrnort srnoktd' Relative risk Iidarirt risk dailr br ssd*crs Casr Control (d 93% Cl): P Cnse Control (d 95% CI) P 1!lil 9 36 1' 13i 158 1 1-10 16 10 6.4 36 29 1.50 (2.18, 18.77) <0.001 (0.87. 2.57) >0.05 11-20 14 4 14.0 27 14 2.33 (3.70, 52.92). <0.001 (1.17, 4.62) <0.05 21 + 11 0 - 9 5 2.17, <0.001 (0.71. 6.64) > 0.05 Total S0 50 203 206 Test for tsend r2 - 32.61.,P <0.001' X2-8.04:, P<0.01 IRe aell omcinorna Others ad rsclnsnfied Amowu sesoked' doil,r b;t• ssdjrcu Csst Relnriw risk Control (! 95% Cl) P Casr Relative rifk Caurol (1 9J•ti Cl) P Tsil 9 17 I 25 54 1 1~1',0 6 3 3.78 20 10 4.32 (0.76, 18.79) >0.05 (1.77, 10.57) <0.01 11-20 4 0 - 17 4 9.18 <0.05 (2:80. 30.11!) <0.001 21+ 1 0 8 3 5.76 >0.05 (1.41. 23.57) <0:05 Total 20 20 70 71, Test for trend r3-8.17. P<0.01 r' - 19:86, P <0.001 Norrs Subjeets wnth missin8 data on amount smoked daily rere exeluded.

A the relau.e nsks for adenocarcinoma found in other Hong Kong studies: 1.59 (Chan rr al:, 1979). 1.80 (l.am et al., 1983). 1.88 (Koo er ali. 1985) and'2.1 (Lam. 1985). The significant trend observed for ad'enocarcinoma provides further evidence that smoking is also a risk factor for this eell I type. The association between histological types and smoking was reviewed recently by an IARC Working Group (1985) which concluded that all the three principal types of lung cancer. ti°i=m squamous cell, small cell and' adenocarcinoma, were probably caused~by smoking, although the relative risk was least extreme for adenocarcinoma. The results of the present study have therefore supported the IARC conclusion It should be noted. however, that the proportion of never- smokers was 62:49,, in adenocarcinoma, as compared with 26.1'/, in squamous and small cell carcinoma; and that some ofl the adenocarcinomas among smokers may well not have been caused by smoking. The causes of the high rates of lung cancer, particularly adenocarcinoma in never smoking women in Hong Kong remained unctrtain, and prompted~ the present study. Furthermore, this probkm had become more urgent since Kung et al. (J984)~ sbowed that there appeared to have been aniincrease in the relative frequency of adenocarcinoma in both sexes in the comparison of their series of lung cancer cases in 197,3-198: with an earlier senes in 1960=1972:, Since the publication of the results on passive smoking by Hirayama (1981) and Trichopoulos er al, (1981), passive smoking was postulated as a risk factor for lung cancer in never smoking women in Hong Kong and elsewhere. In Hong Kong. Chan and Fung (1982) reanalysed the use control study, data of Chan rr al: (1979)~ and found that among non-smoking women there were more passive smokers in controls (66'139)~than cases (34/84). The 84 cases includ'ed~ 34 adenocarcinomas and other cell types. In a case control study by Koo ct o); (1984) on 200 female lung cancer patients and 200 healthy district eontrols, 69 adeno- carcinomas and 19 cases not confirmed pathologically were included. The RR in never smoked wives with smoking husbands was 1.48 (P-0.16): and is close to that in the present study (1.65). The RRs for passive smoking in never smoking females by cell types were: squamous cell 1.75, small' cell 1.10, adenocarcinoma 1.11 and large cell 1.44 (Koo ar- al., 1985). However, in a study by Tsm (1985) on 163 female lung cancsr cases and 185 ortbopaedic controls, the author focussed the analysis for passive smoking on 60 adenocarcinoma cases and 144 controls, both ases and controls being non-smokers. For peripheral tumour, be found an incrcased RR of 2.64 (P<0.05) for passive smoking due to a smoking husband. For central tumours, the RR was 1.61, but was not significant. The RR for adenocarcinoma, central and peripheral tumour combined was 2.01 (95°/. Cl -1.09, 3.72; P<0.05; our calculation). Passive smoking in other ctll types was not reported. In the present study the overall RR for passive smoking due to a smoking husband was 1.65 (P<0.01) in all cell types combined. When broken down by oell types, a statistically significaat RR was found only in adeno- carcinoma but not in the other cell types, although this may have reflected chiefly the smallness of the numbers involved. The value of RR of 2.12 was very close to that of 2.01 reported by Lam (1985). The 95'/. Cl for the pttsent study (1.32, 3.39), was narrower than that in Lam's study (1.09, 3.72), however, because the number of subjects was smaller in the latter study. Analysis by central or peripheral positions of the tumour was not possible in the present study because of lack of' information: It is probable that the true relative risk is nearer to the lower end (1.30) than to the upper end (3.36) of the confidence interval, because it is difficult to believe that passive exposure is more harardous than active exposure, and for adenocarcinomas the relative risk (comparing all smokers with all txver-smokers,. including passively exposed never-smokers) for active smoking was only 1.87. The significant trends observed between RR and amount smoked daily by husband for all cell types combined and for adenocarcinoma provides support the view that the relationship is likelyy to be causal. Recently, Blot and Fraumeni (1986) reviewed the epidemiological and other, evidence on passive smoking and lung cancer, and concluded that the existing evidence is highly suggestive that long-term exposure to environmental tobacco smoke increases the risk of lung eaneer. Summarising the available data, they estimated that the excess risk was -30%. The excess risk rose with increasing exposure, reaching +70% among heavily exposed non- smokers. Wald et al. (1986) also calculated a relative risk of 1.35 for lung cancer among non-smokers living with smokers by pooling the results of 10 case control studies and three prospective studies and concluded that breathing other people's tobacco smoke is a cause of lung cancer. Compared to the 13 studies included by Wald cr al: (1986) the present study included the largest series of never smoking lung cancer ases (i99 cases):, Results of the prcsettt study would add mors evidence on passive smoking as a risk factor and they would contribute towards part of the explanation for the high incidence of lung cancer in never smoking women in Hong Kong. With regard to the possibility of bias through the misclass- i(ieation of current and ex-smokers as lifelong non-smokers, Wald et al: (1986) stated that the extent of miselassification bias was influenced by the proportions of men and' women in the population who had smoked at some time and the greater the proportions (of women in particular), the greater the bias. By choosing the high proportions of 50•/% of smokers in women and 70'/% in men and a low observed rclative risk of 1.35, they concluded that the misclassification bias was unlikely to account for all the association between lung cancer and passive smoking. In Hong Kong, the proportion of smokers in men was 32.8'/% and in women 4.1'/. (H'ong Kong Census and Statistics Department, 1985). These figures. particularly, in women, were muehlower than the figures used by Wald et a!' (1986). Also, the observed RR was higher in the present study. Thus the extent of influence by misclassification bias would be much less and coul&not account for the relatively high RR in the present study: Furthermore, a t:omparison for adenocarcinoma on the RR due to active smoking (1.87) and that due to passive smoking (2.12) seemed to suggest that the risk for passive smoking was quite similar to that for active smoking for this particular cell type. This was not the case for all other cell types in which active smoking posed much higher risks than passive smoking.. The apparcntly prater risk of adeno- carcinoma than of other cell types from passive smoking conflicts with findings in other studies and this may be a feature of small numbers. However, Peto and Doll (1986) in their racent editorial on passive smoking stated that the observed risk need not necessarily be the same in all countries as type of tobaoco, past changes in smoking habits, and the extent of passive exposure both at, home and elsewhere may all i differ substantially between different countries. In places like Hong Kong where people lived in more overtrowded conditions with poor ventilation, passive exposure miy be heavier resulting in a higher RR. Moreover, Wynder and Goodman (1983) noted that the predominant cell type of lung cancrr in non-smokers is adenocarcinoma and postulated that passive inhalation may primarily increase the risk for adenocarcinoma because side- stream smoke, which contains many pesous components. can reach the deeper parts of the lung more readily than can mainstream smoke with more particulates. Together with the findings by Lam (1985) on peripheral adenocarcinoma. our results do offer some support for Wynder and Goodman's postulate that passive smoking may, be a risk factor particularly for adenocarcinoma. At the very least, reviews

W 676 TH LAMrral. TOk IV Passive smoking due to a smoking husband (all forms of tobacco) tn 199never smoking cases and' 335 never smoking controls by cell types Smoking kistorr of hurbmvls Case Control Relative ri.tk' Cell type No Yes No Yes (& 93% Cl) P Squamous call ounttoma 15 12 37 35 0.85 >0.05 (0.35, 106) Small cell carcinoma 2 6 18 Ig 3.00 >0.05~ (0.53. 16.90) Adenocaranoma 53 78 92 64 2.12 <0.01 (f.32, 3.39) Large cell carcinoma 2 7 8 9 3.11 >0.05 (0.50. 19.54) Others and unclassified 12 12 28 26 1s08 >0A5 (0.41. 2.82) All cell types 84 145 183 152 1.65 <0.01 (1,16„ 2.35) Nottr. For each cell type, the aaes were compared'.vith their taatchedi controls on passive smoking for ever smokers and never-smoken. Results on ever-smokers wer¢ nor included here. One ease and 2 controls with missing data on smoking and 3 oses and 2 controls with missing data on husband's smoking wwera excluded. Ta1k V Passive smoking due to a smoking husband (all forms of tob.oco) in never smoking caes (all cell types and adenocarcinotna) and ttever smoking controls by amount of tob.aco smoked daily by husband AU al/ types Adn.ocmcinoww Anwunt smoked daUy by hustimd Case Control Relative .nsk (& 95% Cl) P Relative rtirk Cnte Control' (d 95% CI) P Nil 84 183 1 53 92 1 1-10 22 22 2.18 17 12 2.46 (IL1!„4.15) <0.05 (1.09, 5.54) <0.05 lil-20 56 66 1.85 37 28 2.29 (1.19, 2.87) <0.01 (1.26:4.16) <0:01i 21'+ 201 21, 2.07 15 9 2.89 (1.07, 4.03) <0.05 (1.18, 7.07) <0.05 Total 182292 122 141 Test for trend x° - 10,17; P<0.01 xT- 1t.07. P<O.OOI Noter. Subjects wnth missing data on amount smoked daily by husband were excluded. D64:1111rios The present study was a ax control study on lung ancer in Hong Kong Chinese women with a larger number of subjecu included than in the two previous local nse control studies (Chan er al., 1979; Koo et ol., 1984). All our nxs were pathologically oonfumed, unlike these two previous studies which included cases confirmed only by clittieo- radiological criteria. The psimary advantage of its relatively large-size (the largest such series yet repotud) and the improvetttcnr over previous Hong Kong studies by including only pathologically confirmed eases enabled calculations of histologic-specific risk estimates. The controls used were healthy women from the same neighbourhood matched for age. Comparability between cases and controls with regard to basic demographic variables was good, suggesting that these demographic variables may not have a major confounding effect on the results reported. As shown in Table 1, the distribution of cell type in the cases in the present study was twmparabk to the large pathological study of Kung er al. (1984) which included surgical material such as bronchial biopsy, trans-bronchial biopsy, needle biopsy and resection specimens. Biopsy of lymph nodes alone were not included. Cases without histo- logical examination of the primary tumour of the lungs., or which were diagnosed by cytology alone were excluded. Despite the differrnoe in the basis of diagnosis between the present study and that of Kung et al. (1984), the similarity in the results suggests that the cell type distribution observed in the present study should be close to the true distribution. For smoking by the subject herself, the present study confirmed the inereaeed risk of lung cancer found in previous studies in Hong Kong; but indicatedr a slightly higher telative risk (3.81) than in the study of Chan er al: (1979) (3.48) or of Koo et ol: (I985) (2:77). The significant ttend obeerved suggests that the association is likely to be , causal. With regard to cell types, statistically significant RRs were found for all cell types, including adenocarcinoma. In previous studies in Hong Kong, the RRs for adeno- carcinoma were greater than unity but did not reach a statistically significant level, perhaps due to the smaller number of subjects studied (Chan ct d., 1979; Lam er al:. 1983; Koo et d:, 1985). This led to the hypothesis that smoking was not a risk factor for adenocardnoma in Hong Kong Chinese women. The results of the present study suggat that smoking is significantly associated with adeno• carcinoma, although to a lesser degrre tlian with squamous or small celli earcinoma. The RR of 1.87 compared' well with A

678 T H L.i51 r al of passive smoking and lung cancer can no longer suggest that the results in Hong Kong fail to support the existence of a real relationship. In concltuion. however, we note that 25.2°; (,53l2d0) of our patients with adenocarcinoma were neither smokers themselves nor passive smokers due to smoking husbands.. Although smoking and passive smoking tnay, aacount partly for the high incidence of adenocarcinoma, exposure to other factors should be further examined to elucidate the astiology of lung cancer. particularly the high incidence of adeno- carcinoma in this population. We are most grateful to the International Development Research Centrr and University of Hong Kong for their very generous support in providing the research grants to this proJect and to, Dr D.W. Han for his continuous suppon and advice. We wish to thank the trtedical~ superintendents of Granthant Hospiial:Kow{oon Hospital, Kwong Whh Hospita), Nam Long Hospital. RuttonJee Sanatorium and~ United Christian Hospital for their permisston to interview the patients and' the sta(T involved. particularly the pathologists for their co-operation; to Mrs 1. Cbeang. Mrs J. Wong, Miss S.C:, Wong, Miss Connie Wu. Miss C.W' Yip and Miss Riu Lo for interviewing and other raeart:h assistance: to Miss Agnes Chow and Mrs T. Lam for their secretarial assistattce and to all l the intetviewees for their oo-operation and participation. Finally, we are indebted to Dr MJ. Colbourne for his comments on the techntcal'l report submitted to 1I.D.R.C.. We are particularly gntefW' to Mr Richard Peto and' Sir Richard Doll for reading and commenting on the report and for their encouragement. Refere.ees 6LOT. WJ. • FRAL'MENI ,J.F. (1986). Passive smoking and, lung cancer. J. .Siatl Concer I'nsr.. 77.993: BRESLOW. N.E. t DAY. N.E. (1980) The mra(vsis of case costrol srudirs. lrtternauonal Agency for Research on Canar: Lyon. CHAN, W C.. COLBOURNE. M.1!. FL'NG. S.C. A HO. H.C. (1979). Bronchial cancer in Hong Kong 1976-77. Br. J. Cmtcer„3!, 182. CHAN, W{ & FUNG. S.C. ,(1982): Lung cancer in non-smokers in Hong Kong. In Cwcer cmnpalgn. Vol. 6. Canar epidemiology. Grundmann. E. (ed) p. 199. Fischer Verlag: Stuttgart and New York. DiRECTOR OF MEDICAL AND HEALTH SERVICES OF HONG KONG 11986): 1985-1986 Departnsretd Report. Gnvernmertt Printer Hong Kong. HIRAYAMA. T, (19g1): Non-smoking wives of heavy smokers have a higher risk of lung cancer: A study from Japan: Br. Med J.. 2f2, 183. HONG KONG CE*tSL'S ! STATISTICS DEPARTMENT (f985). Special Topics Report 111, Social Data Colected by the Grnera)! Household Survel. Government Printer~ Hong Kong. IARC WORKING GROUP (1985). 1'ARC Monograpkr an the Evolyorion of the Corcinogrnie Risks of CAemicols to Htnnmu: To6acco Smoking. VoL 38. International Agency for Research on Cancer:Lyon. KOO. L.C.. HO. J.H:C. A SAW:D. 11984). Is passive smoking an added risk factor for lung cancer rn Chinese women? J. Exp. Clin. Cancer Res., 3, 3. KOO. LC.. HO: 1.H.C. [ LEE. N. (1985). An analysis of some tuk factors for lung cancer in Hong Kong. hu: J. Cancer. 35,,149.. K(1NG, I.T.M_ SO. K.F. t LAM. T.H:,(1984). Lung cancer in Hong Kong Chinex: Mortality and hinologial typa„ 1973-1982. Br.. J. Cmtcer, S0, 381. LAM. W.K.,(1985): A clieical and epidnniologicol studY of carcinoma of hng w Hong Kong. M.D. Thesis, University of Hong Kong: Hong Kong. LAM. w.K:. SO. S.Y: & YU; D.Y:C. (1983). Cliaicaj features of bronchogenir carcinoma in Hong Kong: Review of 490 pauents. Cancer. 52, 369. MEDICAL RESEARCH COUNCIL'S COMMITTEE ON RESEARCH INTO CHRONIC lRONCHTTIS ((966) Questionnaire on Respiratory Symptoms. UK. PETO, J. ! DOLL•,R. (1986). Passive smoking. Br. J. Cmtcn. 54.381. (editori6l) TRICHOPOULOS. D.. KALANDIDI• A., SPARROS. L. t MACMAHON:. B. (1981). Lung cancer and passive smoking. Int. J'. Cancer. 27, 1. WALD. NJ.,, NANCHAHAL. K., THOMPSON. S.G. t CUCKLE. H.S.. (1986). Does breathing other people's tob.ceo smoke cause lung cancer? Br, Med.' J.. 293. 1217. WATERHOUSE. 1:, MUIR• C.. SHANMUGARATNAM. K! POWELL. I. (eds) (1982). Cancer Incidrnce a Five Coruirrnts. VoL IV. IARC Scientific Publications No. 42. llntermtiona) Agency for. Research on Cancer: Lyon. WYNDER. E.L. a', GOODMAN• M.T (1983).1 Smoking and lung cancer: Some unresolved issues.,Epidrnaol. Rev., S, 177. 1