Tobacco Documents Online

I IL Ay13Ut,,S Jotar:a or EFroEwrouoea Vol. 125. No. U Ccp~moi Z 1987, by The Johnc Hopkim Unrversitv School of Hypene and'Pubhc Health Pnnud ui G'.S-A w'rKfiu reurved PASSIVE SMOKING AND LflJNG CANCER IN SWEDiSH WOMEN GORAN PERSHAGEN,' ZDENEK IiRUBEC," AND CHRISTER SVENSSON' Pershagen, G. (National Institute of Environmental Medicine, P.O. Box 60206, 5-10401 Stockholm, Sweden), Z. Hrubec, and C. Svensson. Passive smoking and knp cancer in Swedish women. Am J' EpidsrnJof 1987;125:17-24. The relation between passive smoking and lung cancer was examinsd by means of a case-control study in a eohortof 27,409 nonsmokinp Swedish woman idMtified from questionnaires mailed in 1961 and 1963. A total of 77 cases of primary carcinoma of the bronchus or lung were found in a followwpof the cohort throu9h 1980. A new questionnaire In 1984 provided infortRation on smoicirp by study, subjects and their spouses as well as on potential confounding factors. The study, revealed a relative risk of 3.3, constituting a statistically spntficant inerease (p < 0.05) for squamous cell and small ¢eil carcinomas in woman manied to smokers and a positive dose-rasponse relation. No consistent Nfeot could be s.en for othw histologic types, indicating that passive smoking Is related primariiy to those iwma of iuny cancer which show the highest relative risks in smokers. histology; lung neoplasms; smokin9; tobacco smoke po0ution In recent years there has been a growing interest in the health effects of environ- mental'tobacco smoke. Biologic monitoring ibas demonstrated that exposure to tobacco smoke constituents may be appreciable among passive smokers (1-4). Several stud- ies show that children with parents who 6moke have an increased risk of bronchitis and pneumonia„ and' some data also indi- Cate changes of pulmonary function in adults and children erposed to environ- mental tobacco smoke (5). A few epidemiologic studies have been Published on passive smoking and lung can- 8eceived for publication January 22, 1986„and in Gaal form M.y 6, 1986. ' Ckpartment of Epidemiology, National Institute of R.nvironmental Medicine, P.O:,Boz 60208; S-10+101 Stockbolm. Sweden. (Reprint requesu to Dr. Gonn Pershagen.) ' Present address: Radiation EpidemiofoLv Branch; National Cancer Ihstitute, Bethesda, MD. This study was supported by a grant from the ~edish Cancer Society. 'hbe authors are grauful'to Kristina Pannone. Na- tionali Institute of Environmental Medicine„ Iirs Johnsaon,,Nationsl'Centnl Bureau of Sutistics, and B+raDtta Pershagen for help in data collection. cer (6-17). Some of these show increased risks for nonsmokers married to smokers,, but the results are not fully consistent. Most of the studies were not specifically designed to investigate effects of passive smoking, and there are various potential sources of random and systematic errors which make it difficult to interpret the findings. One aim of the present investiga- tion was to try to minimize such errors, especially with regard to the validity of the information on exposure and effects. MATERIALS AND METHODS Study subjects This investigation is designed as a case- control study within a cohort of nonsmok- mg women. There are two sources for the cohort. Most of the subjects are taken from a sample of about 55,000 men and women aged L5-65 years in the 1960 National Cen- sus of' Sweden for whom tobacco smokingg was investigated by a questionnaire mailed in 1963. Detailed descriptions of the sam- piing strategy and the questionnaire are 17 N©T1CE This maters,l rr.3y be Protect'ed tly coDyrigit law Jiile 17 U.S. Code).

18 PERSHAGEN ET AL givem elsewhere (18). The response rate among the women was 95.4 per, cent. A total of 17,679 (66:8 per cent): of the women stated that they had never smoked any form of tobacco„ and these are included in the present study. The secondsource of subjects is the "old" Swedish twin register which contains about 11,000 same-sex twin pairs born between 1886 and 1925 (19). The twins were iden- tified from birth certificates, and a ques- tionnaire was mailed to them in 1961, pri- marily to determine zygosity and tobacco smoking status. The response rate among the eligible female twin pairs was 85.1 per cent. In all, 9,730 women (80.6 per cent) had never smoked, and they make up the rest of the study cohort. Cancer morbidity and mortality of the 27,409 women in the study cohort were determined through 1980 in the Swedish Cgrtcer Register and the National Register on Causes of Death, respectively. The qual- ity of the information in these registers is high for most cancer diagnoses (20). A total of 92 cases of tracheal, bronchial, lung, or pleural cancer were identified (Interna- tional Classification of Diseases (TCD), Sev- enth Revision, codes 162-163) (21). These subjects constitute the case series. Two control groups, each containing two controls per case, were also selected from the study cohort. Control group 1 consisted of subjects who were matched to their re- spective case on year of birth (±1 year). Control group 2 included subjects who were matched on year of birth as well as on vital status at end of follow-up. The subjects in both control groups were selected at ran- dom from subjects who fulfilled the match- ing criteria, with the exception that no woman could be used as a control for her twin sister. The entire study group con- sisted of 460 subjects: 58 cases and 232 controls from the 1963 smoking sample, as well as 34 cases and 136 controls from the twin register. - Exposure information There were two sources of exposure in- formation. First, as described above, data in the 19611 and 1963 questionnaires were used to define the cohort from which the cases and controls originated. The second source was a questionnaire mailed in 1984 to:each study subject or, if she was dead„to the next-of-kin (excluding the husband); in order to validate the data on smoking as well as to assess the exposure to environ- mental tobacco smoke from husbands and parents. If a woman hadbeen married more than once, smoking was investigated only for the man with whom she had' cohabited the longest. Questions omoccupational and residential history were also included. If the questionnaire answers were incomplete, ad- ditional information was obtained by tele- phone interview. The methodology using nest-ofAin to obtain data has been shown to provide exposure information of high quality (22-24). The residential history information from the 1984 questionnaire include& data on addresses (parishes) andtypes of houses in which the study subjects had lived. A pariah was classified as urban if 90 per cent or more of the population lived in built-up areas according to the 1970 Nationali Cen- sus. One-family houses made of material other than wood and with basements were classified as dwellings presenting a greater risk of radon exposure. Indoor radon mea- surements show that the average coneen- trations in such houses are higher than in other common types of dbvellings in Sweden (25). Statistical methods Several methods have been used in the statistical analysis. The matching was re- tained in some analyses, and maximum likelihood estimates of relative risks (ap- proximated with odds ratios) and exact confidence intervals were computed ac- cording to the method of Miettinen (26). In other analyses, the matching was dissolved, and the relative risks and confidence inter- vals were estimated as suggested'by Mantel and Haenszel (27Y and Cornfield (28), re- spectively: The method proposed by Mantel (29) was used to test linear trends in these analyses: Besides the conventional strati- :02351~8-04

PASSIVE SMOKING AND LCNG CANCER'.lt: SWEDEN fied analyses, a conditional logistic regres- sion analysis (30) was carried out in an attempt to control, residual'confounding in the risk estimates and to study interactions. N RESULTS A careful review of' the medical records of, the 92 lung cancer cases showed that in nine cases the primary site was not, the bronchus or lung (there were no primary tracheal or pleural carcinomas), and im six cases the primary site was uncertain. Car- cinoma of the breast, which occurred in five cases, was the most common cause of sec- ondary carcinomas. For 64 of' the 77 pri- mary carcinomas of the bronchus or lung, the diagnoses were based on, histologic evi- dence, and for 12' diagnosis was based on cytology. In, one case, an autopsy was per- formed, but there was no histologic exami- nation. The distribution of histologic types among the primary bronchial and lung car- cinomas is shown in table 1. The classifi- cation is based on the information in the medical records„particularly the pathology reports. Adenocarcinoma is the most, com- mon group, constituting 57.1 per cent of the total: Squamous cell and small ceil''car- cinomas constitute 31.2 per cent. The av- erage ages at diagnosis and at' death for the whole group of carcinomas are 69.0 and 69.6 Yean;, respectively. In the following analy- sis, the squamous cell and small cell carci:- TAaLE 1, Histopatholog>• of primary Dronchial and'lung carcinomas and mean a8ra at diagnosis and at death in a cohort of 27,409 nonsmoking Siuedish women Diagnosis hib: 4r ,\ge (rears I Diasnosis Death ` Squamous eell arcinoma 12 15.6 68.5 70.1 Small cell carci- noma 12 15.6 65.6 65.8 Adenocarcinoma 44 57,1 69.7, 70.2 L+rge cell carci- nama 5 6.5 6-1.9 68.0 O'her priman, carcinomas 4 5.2 54.4 74.8 Total l ' 100.0 690 696 I 19 nomas are grouped together because these types have generally shown the highest rel- ative risks among smokers (31). Table 2'shows the distribution of selected variables among the cases and the control groups. As a result of the matching criteria, the age distribution and vit.al status are similar for the cases and control group 2. In control group 1, there is a shift toward older ages, and more subjects were alive at the end of follow-up than in the two other groups. Questionnaires were returned for 90.2- 96.7 per cent of the study subjects in the different groups. Among the proxy respon- dents, 68.4 per cent were children of the study subjects, 21.3 per cent were brothers or sisters„ and 10.3 per cent were other relatives. There were no differences in the type of proxy respondents between the case and control groups. All of the returned questionnaires con- tained information on smoking by the study subject and, with the exception of one sub- ject in each control group, on whether she had been married and whether her husband had smoked. For the other questionnaire items, e.g., smoking habits of parents, em- ployment, and residential! history, the in- ternal nonresponse rates ranged from 9.6- 32.6 per cent. The percentages in table 2 are based on the number of respondents to each item.. Eight (1.8 per, cent) of the 436 women for whom questionnaire information could be obtained in 1984 had smoked daily during at least two years. Four of these had stopped before answering the 1961 or the 1963 questionnaire, and one had started after that. Two women smoked 1-7 ciga- rettes per day, and one was a pipe smoker. These eight women were ezcltlded in the subsequent analyses. There were no pro- nounced differences between the groups with regard to the percentage of women who were married or the percentage who were married to smokers. For the remainder of the questionnaire items, no consistent differences were seen between the groups, with the possible ex- ception of a tendency toward a iarger per-

20 PERSHAGEti ET AL TABLE 2 Distribution, of selected canables amon{a cases of lung cancer and tuo control groups matched for year of birth,. from a cohort of nonsmoktnQ u•omen Cases Control Control, Cases Control, Control groupl'group 2 group I group 2 Total 92 184: 184 100 100, 100 Localization of primary tumor Bronchus or lung ;; 83.; Other site or uncertain, 15 16.3 Age at death or at end of follow-up (yearsl', 40-69 44 38 93 47.8 20;; 50.5 70-;9 40 90 73 43.5 4&9: 39.' 80-91 8 56 18 8.7 30.4 9.8 Vital stattu at end'of follow-up AJive 5 121, 10: 5:4 65.8' 5.4 Dead 871 63 174 94.61 34:2 94.6 Total questionnaire respondents 83 178 1; S 90:2' 96,; 95.11 Smoked dail.~f 2 3 3! (2A)' 0.;1 (1.") Marriedt ;0 143 151 (84.31 (80.3) (86.8) Married to smoker+ 37 76 77 (;t4.6) 142.91 (44.3)1 Atleast one parent smoker* 12 30 21 (21.1) (21A') 115.9) Employed outside home* 33 73 52 (44.0) (48.3) (34.7) Lived in urban area+ 39 78 82 (60.9) (61!.4); (62:6). Lived in dwelling presenting a greater risk of radon ezposure• 11 13' 9 (17,21 (10.811 (7:0) ' Numbers in parentheses correspond to percentages of total number of questionnaire respondents to each item. f;Minimum duration~of two years. ;',Exposures occurring after the death of their respective case have been excluded for controls alive at the end'of follow-up, centage of cases than of controls who lived duces a relative risk of 3.3 for squamous in dwellings presenting a greater risk of cell and small cell carcinomas (95 per cent radon exposure. A detailed analysis of the confidence interval (CI) 1- 1 I. ~14) asso occupations held by the cases and controls ciated with marriage to a smoker. Within did not reveal any differences between~ the this group, the relative risks were increased groups. The great majority of the occupa- for both histologic types. The relative risks tions were in the service sector and~typical for the other histologic types and for the for women~ of the age group under study, entire group are 0.8 (95 per cent CI : 0.4- e.g., housemaid, cook, seamstress, cleaner, 1.5) and' 1.2 (95 per cent U = 0.7-2.1), and nurse. respectively. In the following analyses, the 15 cases Table 4 gives a dose-response analysis with primary sites other than the bronchus with regard' to smoking by the husband. or lung have been excluded. Table 3 gives, The matching was dissolved in this analysis in a matched analysis, the relative risks for as well: as in table 5: There is a positive primary carcinoma of the bronchus or lung trend in the relative risk for squamousSell in women married to smokers. Never mar- and small cell carcinomas.(x" = 3.9), but ried women and women married to non- not' for the other hlstologic.,types. The rel'- N smokers constitute the reference category. ative risk in the highest ezp~os group, O The results are consistent for both control i.e., wometi" with hus6an"~s 'w'h~oked N groups. Pooling the control groups pro- ^ more than 15 cigarettes per day or one pack cil ~ ~ ~ ~

PASSIVE SMOKING AND LUNG CANCER' IN SWEDEN 21 TABLE 3 Rebtitr risks (RR) ond'95 % confidence intertnis (Cl) for primary careinoma of the bronchus or lung in nonsmoking uomen marrted to smokers with two control groups in a matched analvsu' e t H lo No of Control,`roup 1t~ Control group 2j~ Both control groups 7p isto gic ca/ts RR CI RR CI RR Cl Squamous cell'or small cell carcinoma 20 3:8 1'.1-16,9 3.4 0.8-20.1 3.3 1.1-11.4 Other types 47 0.7 0.3-1.6 0.8 0.4-1.7 0!8 0:4-1.5 Total 67, 1.2 0.6-2.2 1.1 0.6-2.1 1.2 0.7-2.1 `Ne.xr married: women and women married to nonsmokers constitute reference category. Mazimum likelihood estimates of!relatn•e risks and exact confidence intervals (26): + Matched to cases on vear, of birzh, = Matched'to cases on year of birthxs well as on vital status at end of follow,up. TABLE 4 Relatiue risks (RR) and'95 % con/idence interual3 (Cl) for primary carcinoma of the bronchus or lung in nonsmoking u omen i in relation to esttmated exposure to tobacco smoke from the husband" Histologic t.pe Never married or married to a nonsmoken Low exposure toaobacco High exposure to tobacco smoke of husband+ smoke of husbandl No. of RR No. of' RR CI No. of', RR CI cases cases cases Chi- aquarr fortrend5 Squamous cell or small Icell carcinoma 71 1.0 10, 1.8 0.6-5.3 3 6.4 1.1-34.7 3:90 Other types 27 1.0 16 0.8 0.4-1.6 4 2.4 0.6-8:7 0.03 Total 34 1.0 26 1.0 0.6-1~8 7 3.2 L0-9.5 1.45 • Age-standardized relative risk estimates (2 1) and approximate confidence intervals (28). * Husband smoking up to 15 cigarettes per day or one pack (50 g) of pipe tobacco per week or any amount during less than 30 years of marriage. t Husband smoking more than 15 cigarnus per day or one pack of pipe tobacco per week during 30 years of marriage or more. S Test for linear trend (29). of pipe tobacco per week during 30 years of and matehed' controls with information on "' alarriage or more, is 3.2 (95 per cent CI ~' all variables, were consistent with the re- 1.0-9.5) for all histologic types combined sults of the stratified analyses. There was Table 5 shows the influence of parental smoking on the risk of primary carcinoma of the bronchus or lung, controlling for smoking by the husband. There is no con- sistent evidence of an effect, and the 95 per cent confidence intervals for the relative risks in women with at least one smoking Parent encompass 1.0 for both histologic groups. These results must be interpreted with caution in view of the lack of infor- mation on parental smoking habits for 24 per cent of the questionnaire respondents. The results of the conditional lbgistic regression analysis, which included cases no important confounding of the associa- tion between smoking by the husband and squamous cell and small cell carcinomas byy occupation, by living in houses with a greater risk of radon exposure, or by living in urban areas. None of the relative risks associated with these factors deviated sig- nificantly from 1.0 upon statistical testing. For all histologic types taken together, the relative risks and 95 per cent confidence intervals associated with marriage to a smoker and with living in a house present- ing a greater risk of radon exposure were 1.2 (95 per cent Cl = 0.6-2.6) and 1.4 (95

22 PERSHAGEti FT AL TABLE 5 Relatwe risks tRR1 and 95 c con(tdence znteruals /Cll for primacv carcinoma oJ, the b'ronch'us or lung in nonsmoking uomen tn re(trtion to :moktng limb'its of parents` Histolopc Both prenta nonsmokers AtJeut one smoktng parent type ( ase. RR ~y~.f RR' Ct Squamouss cell or small cell carcinoma 0 .0 6 .9 :5-8.2 Other types 28' 1.0 3 0.5 0.1-1.9 'ilota l 38 1.0 9 1.0 0.4-2.3 ' Mantel-hlaenszel estimates of relative risks (27) standardized for age and smoking of husband with approximate confidence intervals (28): per cent CI = 0.4-5.4); respectively. For women who had been married to a smoker and' who had lived in a house presenting a greater risk of radon exposure the relative risk was 2.5 (95 per cent CI = 0.8-8.5), suggesting a positive interaction between the two variables. Discusslonr The results of our study indicate that exposure to environmental tobacco smoke is related to an increased risk of those histologic types of lung cancer which show the highest relative risks in smokers. This is in general agreement with the findings of Trichopoulos et al. (7), Garfinkel et al. (15)„ and Koo et al. (16), although these authors looked at somewhat different car- cinoma types and/or used other definitions of exposure. It would be of interest to see an analysis of the risks for different histo- logic types in the other published; studies on passive smoking and lung cancer, espe- cially those with an appreciable number of cases, as well as in subsequent studies on this topic. Combining the published epidemiologic studies provides a weighted average relative risk of lung cancer of 1.5 associated' with marriage to a smoker (5). The results of the present study are consistent with this esti- mate. A 50 per cent increase in risk does not seem unreasonable in view of exposure estimates among passive smokers (5, 32) and the excess risks of between 100'and 900 per cent for smokers in the lowest exposure category, as a rule 1-9 cigarettes per day, in the major cohorts studied (18, 33~=39) llt should be noted that relative risks for squamous cell and small cell car- cinomas would be expected to be even higher, iie., if the case group is not "dilUted" with adenocarcinomas or other types with weaker association to smoking. Several sources of random and sys- tematic errors have to be considered in the interpretation of the findings. In contrast to eariier studies on passive smoking and lung cancer, the present study has a "double check" on the smoking status of all study subjects. Data were obtained from the 1961 and 1963' questionnaires that were used to define the cohort as well as from the 1984 questionnaire. Our results indicate that misclassification of nonsmokers was a mi- nor problem and that failure to take this problem into account would not severely bias the association between passive smok- ing and lung cancer. This is supported by the findings of other Swedish studies, which show' a high quality of' questionnaire information on smoking, both when the data were obtained from the subjects them- selves and when data were obtained from next-of-kin (22, 23).. Using smoking by the husband as the only measure of exposure to environmental tobacco smoke will result in misclassifica- tions in the exposure assessment. To the extent that such misclassifications are un- related to the disease in question, this would tend to reduce any true association between passive smoking and lung cancer. The similar percentages of exposed'persons among the cases, excluding squamous cell and small cell carcinomas, and the two control groups suggest that errors in the reporting did not affect the cases and con- trols d;ifferently. This lends further support to the association with smoking of'the hus- bands, which was noted for squamous cell and small cell carcinomas only: Obviously, '`,~*IPI 2o23s1~:sQs ,

t PASSIVE SMOKING AND LUNG CANCER IN' SWEDEN 21 it is unlikelv that the next of-kin respon- dents were aware of the histologic subtypes diagnosed for the cases. Our results show that poor quality of the diagnosis may be a problem in studies of lung cancer in female nonsmokers. Second- ary pulmonary carcinomas or carcinomas with unknown primary sites appeared: in about one-sixth of the cases reported in the cancer and/or cause of death registers. This is in close agreement with the findings of Garfinkel (8), which were based' on death certificate diagnoses in the United States. If secondarv tumors are not excluded from the case series, the relative risks associated with~any factor that causes primarily Jung carcinomas are likely to be underestimated.. As noted previously, the analysis may be further strengthened by separating differ- ent histologic types. Besides the quality of the exposure and diagnostic information, the validity of our study is al5o affected by the control of confounding factors. The association be- tween passive smoking and lung cancer of the squamous cell and small cell types was not confounded by occupation, urbaniza- tion, or living in houses with~a greater risk of radon exposure nor were any of these factors associated with a clear increase in risk when passive smoking was controlled. These findings should be interpreted with some caution in view of the internal non- response on the questionnaire for items other than smoking of the study subjects and their spouses. It is, however, improba- ble that uncontrolle& confounding by the factors under study explains relative risks of the magnitude observed, as well as the Positive dose-response relations. Nb infor- mation was obtaine& on intake of food items that may affect the lung cancer risk. Analysis of all the lung cancer cases sug- gested a positive interaction between mar- riage to a smoker and living in dwellings presenting a greater risk of radon exposure, i.e., one-family houses made of material other than wood and witK a basement. In- creased risks of lung cancer associated with living in such houses have been observed previously (40-42), but our study also~pro- vides data on exposure to environmental, tobacco smoke. Our findings are consistent with aninteraction between tobacco smoke and radon daughters similar to the one observed in uranium miners (43) and in smokers living in dwellings with a greater risk of radon exposure (41). It is also of interest to note that the radon daughter concentration has been shown to increase considerably, as a result of attachment to aerosol particles in rooms filled with to- bacco smoke (44). In conclusion, our results indicate that, exposure to environmental tobacco smoke is related primarily wthose forms of lung cancer which show the highest relative risks in smokers. The results are internally consistent and in general agreement with other studies. Our findings are of scientific interest and have public health implica- tions, although it is obvious that lung can- cer, in passive smokers is a rare phenome- non. The accumulating evidence in children and adults shows that serious health effects can probably result from heavy exposure to environmental tobacco smoke. This should encourage further researeh,including both exposure assessments and etiologic studies. Rerml+crs 1. Feyerabend C, Higgenbottam T,, Ruasel MAH. Nicotine concentrations in urine and saliva of smokers and non-smokers. Br Med J 1982: 284:1002-4. 2 Jarvis MJ. Ruaael MAH', Feyenbend C. Absorp- tion of nicotine and carbon monoxide from passive smoking under natural conditions of exposure. Thorax 1983:38:829-33. 3. Greenberg RA. Haley NJ, Etsel R, et al. Measur- ing the exposure of infants to tobacco smoke. N Engl J Med 1984:310:1075-8. 4. Matsukura S, Taminato T, Kitano N, et al. Effectss of environmental tobacco smoke on urinary cotin- ine excretion in non-smokers. N Engli J Med 1984:311:828-32: 5. Pershagen G. Review of epidemiology in relation to passive smoking. Arch To:icol [SupplJ 1986i 9:63 -73. 6. Hirayama T. Non.smoking wives of heavy smok- ers have a higher risk of lung cancen a study from Japan. Br Med J 1981:282:183-5. 7. Trichopoulos D. Kalandidi A. Sparros L. et a11. Lung cancer and passive smoking. I',nr J Cancer 1981:27:1-14. 8. Garfinkel L. Time trends in lung cancer morulity

24 PERSHAGEti ETi AL. among non-smokers and a note on passive smok- ing JNCl11981;66;1061'-6. 9: Chan WC- Fung SC. Lung cancer:in non,smokers in Hong Kong. In: Grundmann E. ed. Cancer campatgn: Vol 6 Stuttgart: Gustav Fischer Verlag, 1982' 199-202. 10. Correa P, Pickle LW:_Fontham E, et al. Passive smoking and lung cancer. Lancet 1983:2:595-7. 1L Trichopoulos D;, Kalandidi' A, Sparros L. Lung cancerand passive smoking: conclusion of Greek study. Lancet 1983:2:677-8: 12. Gillis CR. Hole DJ; Hawthorne WM, et al. The effect of environmental tobacco smoke in two urban communities in the west oflScotland. Eur J Respir Dis 1984;65 Isuppl 1331:12d-6. 13: Kabat GC. Wynder EL. Lung cancer in non- smokers. Cancer 1984;53,•1214-21. 14. Koo LC, Ho JHC, Saw D: Is passive smoking an sdded risk factor for lung cancer in Chinese women''J Exp C1in Cancer Res 1984:3:277-63. 15. Garfinkel! L. Auerbach 0, Joubert L. Involuntary smoking and lung cancer: a case-control study. JNCI 1985;75:463=9; 16. Koo LC, Ho JHC, Lee N:,An analysis oflsome risk factors{br lung cancer in Hong Kong..lnt J Cancer 1985:35:149-55, 17. Wu AH, Henderson BE. Pike MC, at al. Smoking and other risk factors for lung cancer in women. JNCI 1985;74:54 7-51. 18. Cederlof R, Friberg L. Hrubec Z, et al. The rela- tionship of smoking and some social covariables to martality and cancer morbidity. Stockholm: Department of Environmental Hygiene, Karolin- sks Institute- 1975. 19. Cederlof R. The twin method in epidemiological studies on chronic disease. Stockholmi: Depart- ment of Environmental Hygiene, Karolinska In- stitute- and Department of Sociology, University of Stockholtn. 1966. 20. !vfattson B. Cancer registration in Sweden: studies on completeness and validity of incidence and morulity, registers. Stockholm: Department of Oncology and Cancer Epidemiology, Karolinska Institute, 1984. 21. International Classification of Diseases- Seventh Revision. Geneva: World Health Organization, 1957. 22: Pershagen G, Axelson O. A validation of question- naire information on occupational exposure and' smoking. Scand J Work Environ Health 1982; 8:24-8.. 23. Damber. L. Lung cancer in mates: an epidemiolog- ical study in northern Sweden with special regard to smoking and'occupation. Umei. Sweden: De- partment of Oncology, Umei University. 1983: 24. Pershagen G. Lung cancer mortality among men living near an arsenic-emitttng smelter. Am J Epidemiol 1,985;122:684-94. 25. Falk R: Study of a method'to estimate the expo- sure to radon daughters in dwellings. Im Pilot studies and theoretical considerations on the pos- sibility to study the relation between radon daugh- ter exposure in dwellings and lung cancer. (In Swedish). Stockholm: National Institute of Envi- ronmental Medicine. 1982' 26. Miettinen O. Estimation of relative risk from in- dividually matched series. Biometrics 1970;26:,5- 86: 27. Mantel 'v: Hsenazel' H'. Statistical aspects of~ the analysis of data from retrospective studies of dis- ease. JNCl11959:22: 719-38:. 28 Cornfield J. A statistical, problem ansing from retrospective studies In: Proceedings oflthe third Berkelec symposium on mathematical statistics and probabiGty Berkeley, CA: University of Cali ifbrnia Press. 1956. 29. Mantel N. Chi-square tests with, one degree of freedom: extensions of'the Mantel-Haenszel pro- cedure. Am Statistical Assoc J 1963:58-690-700. 30:, Breslow N, Day N. Statistical methods,in cancer research-the analysis of'case-control studies. Lyon: IARC Scientific Publications. 1980. 31. Kre,vberg L. Histological4ung cancer types: a mor- pholoRicallsnd biological correlation, Acts Pathol' Microbiol Scand 1962:Suppl; 157:1-92. 32. Vutuc C. Lung cancer risk and passive smoking: quantitative aspects. Zbl Bakt Hyg I Abt Orig B 19811 77:90-5. 33. Hammond EC. Horn D. Smoking and death rates-report on forty-four months of follow-up of 187,783 men. II. Death rstes by cause. JAMA 1958:166:1294-1308. 34. Hammond EC. Smoking in relation to the death rates of one million men,and women. Natl Cancer Inst Monogr 1966:19:11 7-204. 35. Best EWR. A Canadian study of smoking and health: Ottawa: Department of National Health and Welfare„1966. 36. Kahn HA. The Dorn study of smoking and mor- tality among US veterans: report on eight and one-half years of observation. Natl Cancer Inat Monogr 1,966:19:1-125:. 37. Weir JM, Dunn JE. Smoking and mortality: a prospective study. Cancer 1970:25:105-12. 38. Hirayama T. Ptospective studies on cancer epi- demiology based on census population in Japan. 4Er Ini Cancer epidemiology and environmental fac- tors. Voli 3. Amsterdam: Excerpta Medica, 1975:26-35. 39. Doll R. Peto R. Cigarette smoking and bronchial carcinoma: dose and time relationships among regular smokers and lifelong non-smokers. J Epi- demollCommunity Health 1984:32:303-13. 40. Axelson:0: Edling C. Kling H'. Lung cancer and residency-a case-refenrnt study on the possible impact of exposure to radon and its daughters in dwellings. Scand J Work Environ Health 1979; 5:10-15: 4L. Edling C. Kling,H, Axelson 0. Radon in homes- a possible cause of lung cancer. Scand J Work Environ Health 1984:10:25-34. 42. Pershagen G. Damber L, Falk R. Exposure to radon in dwellings and lung cancer-a pilot study. In: Proceedings of the Third International Con- ference on lndoor Air Quality and Climate. Vol 2. Stockholm: Swedish Council for Building Re- search, 1984:73=8. 43. Archer VE, Wagoner JK. Lundin FE. Uranium mining and cigarette smoking effecta on man. J Occup Med 1973i15:204-11. N. Bergman H, Edling C, Aselson, 0. Indoor ndon daughter concentrations and passive smoking. ln: Prnceedings of the Third International Confer• ence on Indoor Air Quality and Climate. Vol 2. Stockholm: Swedish Council for Building Re- search 1984:79-84. 2023512810