Tobacco Documents Online

.;" I i i-. t\nvemae,0ecemner 1 011 1 Cancer Epidemiolorry, Biomaricers & Prevention Lung Cancer in N!onsmoking Women: A Ivlulticenter Case-Controll Study' Elizabeth T. H. Fontham,' Pelavo Correa, Anna Wu- Williams. Peggy Reynolds, Ravmond S.. Greenberg, Patricia A. Buffler, Vivien W. Chen. Peggy Bovd, Toni Alterman, Donald F. Austin, lonathan Liff, and S. Donald Greenberg Department of Pathologv. Louisiana State Umversttv Medical Center. New Orleans: Loursraru 70112 I E. T H~ Fi . P C. V W C.I. Department ot Flamtivy and Communitv Medicine. Unrversnv ot ~.outhern Calrtornra. Los Angeies. Caluornra 90033 IA w-w I: Calrtornra Deoartment of Health Servi ces. EmerwJie. Cahiornta IP R'. D F 4!. Division or EprdemtotoQV. Emorv Unwersnv School of Public Health Atianta. Geor¢ra 30322 f R S G-. 1! 1;. School or Publrc HPallh. Unrversnv ,,t Texas rteatth Science Center. Houston, iexas :7030 IP .4 8 - A j. Cahtornra Public Health Founaatronj Berkerev: Calitornra 414720 P 8./: and Depanment of Patholbgy:,BavbrColdeRe of Medtune. Houston. Texas 7J030 IS. D. G.1 Abstract The association between exposure to environmental tobacco smoke and'lung cancer in female lifetime nonsmokers was evaiuated' using data collected during the first 3 years of an ongoing case-control studv: This large, multicenter, population-based study was designed to minimize some of the methodological problems which have been of concern in previous studies of environmental tobacco smoke and lung cancer. Both a cancer control group and a population control group were selected in order to evaluate recall bias. A uniform histopathological review of diagnostic material was conducted for case confirmation and detailed classification. Biochemical determination of current exposure to tobacco and'screening or multiple sources oi information to determine lifetime nonuse were utilized'to minimize misclassification of smokers as nonsmokers. A 30% increased risk of lung cancer was associated with exposure to environmental tobacco smoke from a spouse, and a 50% increase was observed for adenocarcinoma of the lung. A statistically significant positive trend in risk was observed as pack-years of exposure from a spouse increased, reaching a relative risk of 1.7 for pulmonary adenocarcinoma with exposures of 80 or more pack- years. The predominant cell type of the reviewed, eligible lung cancer cases was adenocarcinoma (78%). Results were very similiar when cases were compared to each control' group and when separate analyses were c Recerved'5/2/91 ' Thrs research was supponed bv Grant CA40095 trom the Naoonal, Cancer instttute. ' To whom requests tor reprints should'be addressedi,at Department ot Pathology: LSU Medreal Center. 1901 Perdido St.. New Orieans. LA, 70 t12. conducted for surrogate and personal respondents. Other adult-life exposures in household, occupational, and social settings were each associated with a 40- 60% increased risk of adenocarcinoma of the lung. No association was found between risk of any type of lung cancer and childhood exposures from a father, mother, or other household members. Introduction Approxlmatelv one decade has passed since the initial reports of increased risk of lung cancer In nonsmoking women married to smokers (1, 2). The ensuing studies have provided a body of data which suggests a small but significant elevation in risk of I'ung cancer associated with exposure to ETS' ('3-22)1 In reported prospective studies exposure has been assessed by the spouse's smoking history, prtmarilythat of husbands. Inicase-control stud- ies, the prlmary ETS exposure assessed has also been that from a spouse, although exposures from parents,, other household' exposures, and the workplace have been examined in some studies. In general, these studies have included fewer than 1i00 nonsmoking lung cancer cases whose seifrreported smoking status has not been validated by biochemical determination or other means. Reviews of available stud- ies of ETS and lung cancer in nonsmokers by the National Research Council (23), the Internatlonal'.Agency for Can- cer Research 124), and others (25, 26) have conciuded' that although, misclassification ts uniikelv to account for all of the observed increased risk, some misclassification of current or former smokers as nonsmokers is likely (0':5-5.OJo). Because smokers tend to marry smokers, misreporting may introduce some bias in the estimation of the magnitude of the observed effect. This study was undertaken in 1985 in an effort to address a number of unresolved issues related to ETS: (a) Misclassification of Smoking Status_ Multiple sources of information are utilized to ascertain nonsmok- ing status (medical record, physician: and then the study subject or surrogate). Study respondents are questioned twice (at contact to~ set up the interview and at the beginning of the interview). Self-reporte6 current non- smoking status is corroborated by measurement of uri- nary cotinine. (b)' Htstopathological Specificity. Microscopic diag- nosttc slides are reviewed bv one pulmonary pathologist both to confirm eligibility of cases as primarv lung carci- nomas and to provide a detailed review lsubtype, differ- ' The abbreviations used are: ETS, environmental tobaeco:smoke; SE!E'R:. Surveillance. Eptdemtology,. and'End Result: OR. odds rano;'Cl ,conti- dence interval.

c Lun; Cancer in NonsmokinS Women entiationi and classification oi the histopatholoRical cell tvpe.. rc),Recail' Bias. Two control groups, one withicolbn cancer and one trom the general popuiation, are selected ror, ccase-control comparisons. Differential recall between cases and colon cancer controls should'i be minimized since both groups are similarlv motivated to recall earlier exposures. tdl Source oi ETSExposure_ Intormationion childhood'~ exposures irom a tather, mother, and other household members and adult exposures from husband(s), other household members, and occupational and social set- tmgs is obtained! by questionnaire. The risk associated with exposure to ETS from different sources and during different time periods can be evaluated. (e) Coniounders and'Other Risk Factors. Because the magnitude of the main ETS effect, is expected to be small, it is imponannto take into account potential confounding, factors and effect modifying factors in a: study with a sufficientlv large number oi cases and controls. tt is anticipated that upon completion oi this study about 600& cases and', twice that number, oi controls will have parucipated'. This report represents findings irom the ongoing stud'y and includes the largest number of lifetime non- smokers with lung cancers reponed to date. This report was felt to be justified given the public health importance of the issue under, investigation. Methods The study is a population-based case-control study of lung,cancer in women who have never used any tobacco product. This preliininary report includes cases diag- nosed during the first three years (December 1, 1985 through December 31, 1988)' of, a 5-year stud'v. At the time of; diagnosis cases were residents of one of five major metropolitan areas throughout the United' States, including Atlanta (Clayton, Cobb. DeKalb, Fulton. and Gwinnett counaes), Houston (Galveston and Harris counties),, Los Angeles 1Los Angeles County), New Orle- ans tlefferson, Orleans, and St. Bernard parishesl, and the San Francisco Bay Area (Alameda. Contra Costa. Mann, San Francisco, San Mateo, and' Santa Clara coun- ties), representing a population, of approximately 18.5 • million people or 8% of the U. S. population. Case and Control Selection Rapid case ascertainment procedures, which included review of pathology reports from study hospitals, were utilized to identify potentially eligible lung cancer cases. Eligible cases included English-, Spanish-, or Chinese- speaking females, aged 20-79, who had a histopatholog- ically confirmed diagnosis ofl primary carcinoma of the lung (International Classification~of Disease, 9th Revision, code 1621 madepraor to death, had no historv of previous cancer, and who were lifetime nonusers of tobacco. Lifetime nonusers of tobacco are defined' for this study as persons who had smoked fewer than 100 cigarettes an6 had not used any other torm of tobacco tor more than 6 months. Two control groups were selected! The first control group, referred to as the population control group; was selected by random digit dialing and supplemented by random sampling from the files of the Health Care Fi- nancing Administration ior women aged 65 and older. Controls were trequency matched to cases on age (<50„ 50-59; 60-691 70+ vearsi in a 2': T control:case ratio. They met the same eiigibilltv criteria as cases ior age, resi- dence, language; and tobacc&use. Females, aged 20-791 with a diagnosis of primarv carcinoma ofl the col'on (International Classification of Disease. 9th Revision„code 153) who met the language; previous cancer, lifetime nonsmoking, and' residential eligibility criteria of the cases, were identified and fre- quency matched to lung cancer cases by 10-year age groups and race. This second'control group was selected because there is no established increased'risk of colon cancer associated with either active or passive smoking,, and it provided an opportunity to examine the issue of recall bias associated with a recent diagnosis of cancer. A multistep procedure was used to determine life- time smoking status, After identification of a potentially eligible lung cancer case or colon cancer controll the hospital chart was reviewed to obtain demographic data and available information on tobacco use. Patients iden- tified'as current or, former smokers in the medical record were considered ineligible. In studv areas where individ. ual physician notification was requ+red(preferred; the tobacco use history was requested' from the physician for potentially eligible cases and colon controls identified as nonusers of tobacco or with unknown smoking status according to the hospital record'. Women who were identified as current or former, smokers by their physi- cians were considered ineligible. All remaining cases and colon cancer controls believed to be nonsmokers or with unknown smoking status were contacted by telephone to elicit information on tobacco use. Women who re- ported ever smoking 100 or more cigarettes or using any other form of tobacco for more than 6 months were considered ineligible. The identical telephone screening, procedure was used for the population~controllgroup. At the time of the interview, the tobacco use screening questions were repeated to confirm each study subject's reported'nonuse of tobacco. The questionnaire was translated from English into Spanish and! Chinese, and interviewers fluent in those languages conducted the non-English as well as English interviews. Interviews were completed for 431 of 5141 incident cases (84%), 358 of 489 colon cancer controls (7'3%), and 794 of 1105 population controls (72%). Sixty- one (3.8%) of the interviews were conducted in Spanish (n = 14) or Chinese (n = 47): 22 cases ('5°/b); 23 colon cancer controls (6%); and 16 population controls (1.5%d). ~ A next-of-kin interview was solicited for lung cancer cases ~ and colon cancer controls who were too ill or deceased. ~ All population controls were self-respondents because ~ of the sampling method used to identify these controls. A total of 143 lung cancer case interviews and 35 of 352 M colon cancer control interviews were conducted with ~ ~ next-of-kin respondents, representing 34% and 10% a the eligible respondents. An extensive structured questionnaire obtained~ in- Im formation concerning household, occupational, and ~1 other exposures to environmental!tobacco smoke during the study subject's lifetime. Data were also collected on lifetime occupational history; usual adult diet, family and personal medical histories, and other exposures of inter- est, which are not included in this report.

Cancer: EpidemioloRy. Blomarken & Prevention 'abie I urinarv cotrnmelcreatmme rnVmqi bvcase-comroi staws Table 2' Patfiolo¢v rev ew Lung Codon~ Revwwed rouna to be eligible 359 (8S`Yb) cancer ancer, Pooutation . Adenocaranoma 281 cases onrrois ontrois Liarge ceil carcinoma 43 Squamous,cell carcinoma 20. Ssatus Small cell carcinoma 12 Lomoterea J9 '6& nd4 Others ana noo otherwise specitied 3 Eligible 4<100 nq/rngi 237 _53 b70. Inelrqrble i>100 ng/mRr - :.4 - vot rev ewedlmsunicrent material 50 (i12ym) `oi penormeo Seh-responaents. 38 ob 110 Histology by hospital pathologist Next-or-km responoents 134 32 Adenocarcinoma 30. Large cell carcinoma 5 Squamous cell carcinoma 7 Rpsults Small cell carcinoma 2' EI giblp (<100 ng/mgi Others and not otherwise speaned 6 Mean t5D1', 6.95 (12.1,11 5 8_"/11 681 9'68't12'881 Median 2:0 0 5 4 Range 0-71i4 0-88 4 0-95 01 Review pendrng 11 (3°7e) tneLgrble tL- 100ing/mg(i Total cases 420 Range 131-219 143-5J63 103-14 0r4 Reviewed. lound to be mel grble 9 Eligibility Review Procedures Biochemical Determination of Current Tobacco Use. Co- ttntne. a major metabolite ot nicotine, is an indicator of recent exposure to tobacco (27). Urinary cotinine was used to corroborate self-reported current nonsmoking status of study sublects. A urine sample was collected irom all consenting study subjects at the time or inter- view. The specimens were stored at -20`C until shlp- ment, to the American Health Foundation ior analysis. Cotinine was quantitated by radioimmunoassav us- ing the method of Haley et al. (281' with a modification of the antibody of~ Langone et' al: (29). Cotinine levels were adlusted for urine flow based oni creattnlne values by determining the cotinine/creatmtne ratio. Creatinine was determined' by spectrophotometrv using the Kodak Ektachem 40UClinical Chemistry Analvzer.. At this time biochemtcallanalvslsJs complete for 239 of 431 cases (55.5°/,). 260 of' 358 colon cancer, controls. (72.6°/a), and 684' of 794 population controls (86.1 %°) (Table 1). Two of 239 case samples 10_8%l. 7 oQ60!coion cancercontrol samples (2.6%°); and 14 ot 684 population control samples (2:0%l had cotlnlne/creatintne leveisol 100 ng/mg or greater. There is no established cotlrnne/' creatinine ievel'whlch clearly discriminates smokers trom true nonsmokers heavily exposed to ETS. Under rela- tively high levels of exposure to ETS in atrcraft and in exposure chambers, urinary excretion has reached a level of 55 ng/ing creatinine ('30, 31). In this study, women whose cotinine/creatinine level exceeded 100! ng/mg were excluded from the study'to eliminate persons likely to be active smokers, while allbwing for the possibility of very high ETS exposures reflected in urinary levels of 56- ` 99 ng/mg creatinine. Had the lower value of 55 ng/mg been selected' as a cutpoint to avoid~ posstble misclassi- fication of active smokers as nonsmokers. 4 additional cases ('1.6%a), 2 colon cancer controls(0.8%), and' 13 population controls (;1i.9%)' would have' been excluded from the analyses, with negligible effect on the results. Histopathological R'eview: Representative diagnostic mi- croscopic tissue slides for each case were requested from the hospital. These slides were reviewed by one pathol- og(st specializing ln pulmonary pathology. A tota(! of 368 of 429 (86%a) potenbai!cases have undergone review. Ass shown in Table 21.359 (98%liof the reviewed cases were confirmed'as prtmary'bronchogenic carcinoma. The his- topatholbglcal primary cellltvpe of the eligible cases is as follows: adenocarcinoma, 78%; large cell carcinoma, 12%; squamous cell carcinoma, 6%; small cell carcinoma, 3%; others, 1%. The histopathological cellltype distribu- tions were similar in the five study centers. The overall concordance between the review pa- thology diagnosis and the original' hospltal! pathology diagnosis was 81% (Table 3). The concordance varied greatly by histopathological cell type. Ninety-seven % (237 of 244) of the cases originally classified as adeno- carcinomas were confirmed'as this histopathological type upon review. Similarly; 10, of 11 (91%) of small cell carcinomas were so classified'upon review. Concordance rates of' 56% and 67% were seen for large cell and squamous cell'carcinomas, respectively. A relatively large proportion of cases originally classified as large cell or squamous cell carcinomas were classified as adenocar- cinomas by the review pathologist: 18'of 46 (39%) and 6 of 24 (25%)( respectively. Base6on hospital pathology reports. 34 subjects were categorized as "other primary lung carcinomas" which pnmariiv'inclhded diagnoses of pooriv differentiated carcinoma, bronchogenic carci- noma not otherwise specified, or malignant, cells not otherwise specified': Upon review, 94% of~ these cases were classified into more specific histopathological cell types. The nine cases (2%) found not to have primary bronchogenic carcinoma on review were excluded from alllanalyses. Three of these nine cases were determined to betarcinoid tumors, two were lymphomas, three were carcinomas metastatic to the lungs from other primary sites, and'one was a benignineoplasm. The 61 cases that have not undergpne histopathoiogical review are in- cluded in analyses of all lung', cancers combined (n = 420) but are not included in analyses stratified by histo- pathological I type. Statistical Analyses Exposure to ETS was examtned by source. Sources in- clude both adult and childhood exposures as follows: spouse, other household members; occupational ETS exposures:and social or leisure time (nonhousehold, nonoccupational) ETS during adult life; and father, mother, and other household members who lived'in the

Lun; Cancer in Nommohmg Women, Table 3' Distnoution 0t lun¢ cancer mstooatholoqical cell!tvpes ov hospna/ draanosrs and rev ew dia¢nosa. Hosprtal diagnosis Review oiagnos s Adenocarcrnoma Large celli carcinoma Squamous ceil carcinoma: Small cell carcinoma Other lunq carcinoma Tiotal 4denocarcmoma 237 18 6 1 19 281 Large cell carcinoma 6 26 1 0 10: 43 Squamous ceil carcinoma 0! 2 16 01 2 20, Small cell carcinoma 0 0 1 101 1 12 Other pnmarv lung carcinomas i 0 01 0 2' 3 Total 244 46 24 11 34 359 , home 6 months or more during childhood. Childhood was defined as the first 18 years of life: Exposures from parents after that time were classified as other household members during adult life. Dichotomous ETS exposures were first, examined (ever or neveri by type of tobacco: cigarettes: ptpe: clgar; or any of these types oti tobacco. Dose was estlmated: as appropriate, bv' intensity ('e:g., number ot cigarettes/day), duration te.g;, number of years exposed), or a combination te.g;, pack-vearsl. Pack- years of cigarette exposure trom the spouse were calcu- lated by multiplying the number of packs smoked per day'by the number of years the spouse smoked cigarettes while living with, the study subject. Pack-years of expo- sure were summed for all'smoking spouses ot each study' subject. One of'the objectives of this study was to evaluate the association of ETS with specific histopathological cell types of lung, cancer. The skewed distribution of histo- pathological types precluded any meaningful analysis by specific cell type other than adenocarcinoma and' all other cell types combined'. The results are presented for all lung cancers combined (n = 420) and adenocarcino- mas confirmed by'histopathologlcal'revlew (n = 281). Cases were compared to each, control group with regard to the distribution of relevant covariates such as age, education, Income. and race/,ethniatv: The associ- ation of ETS exposure with lung cancer risk was investi- gated first in contingency tables stratified bv, design or sampling variables (age, race, study center) and relevant covariates, Summary adjusted oddk ratios and test statis- tics were calculated'by the method of Mantel and Haen- szel'(32). Unconditional logistic regression analyses were then used to estimate the associations by summary ad- justed odds ratios, confidence limits, and test statistics (33, 34). Results Demographic characteristics of cases and controls are presented in Table 4. Cases and controls were similar with respect to matching variables and most demo- graphic variables. The largest number ot cases ln = 160. 38%1 were residents of Los Angeles, followed by cases from the San Francisco Bay Area (n = 149; 35%), andd then the three smaller study centers in, the southern Uhited States: Atlanta (n = 46, 11%): Houston in = 39, 9%); and' New Orleans in = 26, 6%l. The age distribution of cases and controls is uniform, with 73 to' 74% of each series between the ages of 600 through 79. The proportion of older women in this group of female nonsmokers with lung cancer is higher than that among, all female Iung cancer cases in, the SEER Program 1974-1986, in which only 48% of the cases were aged 65 or older (35). Cases tended to have a somewhat lower household income and less education than the population controls. Approximately 35% of cases and controls spent their childhood' in cities with populations of 50,000 or more, and the maioritv of cases and controls (70%. 68%. 77% for cases, coion cancer controls, and population controls, respectively) resid'ed in cities during most of their adult life. The estimated risks of lung cancer, in nonsmoking women associated with ever, having lived with a spouse who smoked'are presented'in, Table 5. Theadjusted'ORs and the 95% Cl are very similar for all spouse-related exposures regardless of control group: fior all histopath- ological types of lung cancer combined, a 30% increase in risk is observed (OR = 1.28 and 1.29 with coion cancer and population controls). For each of the three types of tobacco smoked, the ORs ranged from 1.14 to 1.26. When the case serfes is restricted to the 281 pultnonary adenocarcinomas confirmed'by histopathological review, the association is more pronounced. Approximately 50% elevations in risk of adenocarcinomas of the lung (P'< 0.05)~are associated'with any use of tobacco by spouse(s); and cigarette smoking,accounts for most of the tobacco use. The estimated relative risk of pulmonary adenocar- cinoma associate6 with cigarette smoking by spouses was 1.36 (1.02-1.84) with the population controls as comparison and 1.31 (0.94-1i.84) with the colon cancer controls as comparison. No association between spouses' tobacco use and lung cancers other than adenocarci- noma (squamous cell, small cell, large cell, and other; n = 78) was observed. Separate analyses were conducted for subjects who personally responded and for whom information was obtained'from surrogate respondents_ The od'ds ratios for involuntary exposure to ETS were very similar for both groups of respondents; therefore, the results are nov presented in the tables separately by type of respondent. One such example is the estimated relative risk of pul- monary ad'enocarcinomas associated with cigarette smoking by the spouse: OR = 1.38 and' 1.30 for surrogate and selfrrespondents, respectively, comparing cases to colon cancer controls. Effects by stud'y center were also examined. The odds ratios by center ranged from a low of 1.17 to a high of 2.64 for risk of pulmonary adenocarcinoma associated' with spouses' cigarette smoking. Because of the limited~ sample sizes, none of the individual study center esti-

'able 4 Ois'Inbution ot IunR cancer casesano controlsaOCORornR to SeFectea oemo¢rapnic Characterosncs. ,unR ~~ibn ldncer ancer 'noulation antrols cases nntrols n = 240, = 35 u 1,0 ,°a1 •,, ~°,i `n ,'°01 ,ludv, center Atlanta -ib 111 Oil 11 412 Sii , •9~,7'1 Houston, 39 19'31 35 10~01 24 ~3'.11 LosAngetes. 160 (381), 123 -3561! 338 145,9) New Orleans 26 I6~?li 1'8~ 15 111 n 14 t5,61 San Francisco 8av'Area 149 /35 51, 129 t36 711 2'8 13's:61 Respondent Studv sublect -7 (66:0), 316 190'11, 780 (100:01 Ne:totkn 143' (340) 01 ~ l.i: (991, Age Ivears/' '0-29 i I1 :] , ~0 31, . n_2r 30-39 10-49 tt 12 61 1 11 0 7 .1 -- !6 31 15.J1 32 . 111 13 :41 13.91 50-59 -3' 11- 31 ii. 056) 121 (I5S1. b0-69 14-1 133,01 103 1_98/ 221 ~?8,31 -J-79 161 138 31 155 .1a 01 35? ii 81 -:ace/ethnic Rrouo, White 266 163 31 240 /68 51 i03 ~64 51' 81ack. 44 /10.51 iR~~ 11681 10.'~ ~13'.71'. Hispanic 32 C 61 14 .301 -02 ' 15 41 . As an 67 116 01 l3~ 110.01 11 3~ ~ 14 ~ 51~, Other 11 -261 r061 I1. i1 '.11 Unknown(retused 0' 10.01 ; .0 21 1 104'1~ to answer Annual income. <S8':000 -2 I1C 11 rid: 117 111 48, 112,61' 58.000-1 12.999 a3 (1i501 i' 048) 115 114 71 S 13:000-19.999 38, /1it.i1 48, 1113 71 1101 I14.11 520.000-34.999 73 117A1 a'a ItF 41 15.3 (1961 135.000-49.999 37 (8.81 44 11401 82 11 0.5) >_s50!000 59 (14 11 3.i 1100) 128 11641 Unknownrrerused 68 (16.21 l611 3 11 94 1112.01 to answer Education Less than.hiRn school i 135 (32.11 84 123 9) 16; 21121 High school 140 (3'31-31 114 06 21 24b ~31151 Some college -1 n16 9 -.1 21 11 181 ~23.21 C olleRe 33 1; oI _8 801 107, .1i3.:1 Graduate _5 16.01 -- 6 31 by 18.91 Unknown 16 t3'.81 Q /..61 12 11.51 Usual childhood residence Farm. 93 (22:1/ 78 (22.21 131 116.81 Rural area 49 101 71 36 n10.31 61 17.81 <20.000 population 92' (2'1.91 81 IQ3'.11 196 125.11 20;t)D0-49.999 population 37 (8.8) 46 (i13 11 98 It2.61 >-50.000 population 146 434.8) 109 (31 11 291 137.3) Unknown. 3 (0.71 1 10.31 3 (0.4) Usual adult residence Farm 23 15 5) ' 15 la 31 10 nl .31 Rural area 10 12:41~. 6 1,1 71 '. 13 10]t <20.000 population 39 19 31 .' 28 (8'0) 45 16.81 2D:000=49:9999 population 53 112.611 b1~. Ifl~7~s1 '. 108 t iI 3191 ?50.000 population 293 (69!8E 240 168'41 601 I'7701 Unknown 2 t0 51I 1 10 3 ) 3 10 41 t mates were statistically signiricant, and thev did not slg- nrficantiv'differ from one another.. Estimates of'relative risk associated wtth~the number ot cigarettes smoked by a spouse were signlficantlv ele- vated only' in the highest exposure categpry, 40 or more Cancer, Epidemiolot,r. -8iomar>tirrs 6 Prevention 39 clgarettes/dav: 2.06 (11.19-3.541 and 1.69 (1.28-2.61) for adenocarcinoma oi the lung comparang, cases to colon cancerand population controls, respectlvelv. Odds ratlos. were slmtlar. although sllghtlv lower, for all twpes oi lung, cancercombined: 1.70 (1i.02'-2.84) and 1.36 (0!90-2>06). Pack-vears were examined as a combined measure of duration and dose oi exposure to the husband`s clga- • rette smoking. The odds ratios tor all cell'types of iung, cancer combined and for adenocarcinoma of the lungg are displayed in Fig,1'.. Separate analyses were'conducted with each control group for comparison. Because the findings were so similar for each group, the results are presented for the two control series combined (n = 1iT31)i An increasing risk of''.lung cancer and adenocarci- noma of the lung associated with an increasing level of exposure to the spouse's cigarette smoking was found. The positive trend in risk by pack-vears of exposure is statistically significant for adenocarcinoma of the lung (P < 0':010: A weaker dose response is observed when all histopathological types of lung cancer are combined (trend, P = 0.07). Exposure to ETS from various sources during adult life was evaluated! The results are summarized in Table 6. For simplicity of presentation, the data in this table also represent the findings using the two control groups combined because the individual results using each con- trol group were entirely consistent. Exposures to cigarette smoking from, spousels); other household members, on the job and in other, activities of adult life ("social")' are each associated with an overall 40-60% significant ele- vation in the risk of' adenocarcinoma of the lung. As noted previously for spouse-related exposures, the risk estimates for all lung cancers without regard'to cell type tend to be slightly lower than the comparable estimates for, adenocarcinoma 'of the lung; Significant positive trends (P < O:05) ' in risk of' adenocarcinoma of the lung were associated with increasing duration (years) of ex- posure to cigarette smoke from: a spouse.other house- hold members, and social occasions. For adult household! exposures from a spouse and others, estimates of risk rose from lowest to highest in the 30 or more years of exposure category; however, trends were not smooth for exposures in occupational and sociallsettings. No association was found between risk of any type of lung cancer and childhood exposure to cigars, pipes, cigarettes, or all types of tobacco combined. Table 7 presents the estimated relative risks of lung cancer and adenocarcinoma of the lung among nonsmoking,women whose father, mother, or other, household member smoked' during childhood! None differed significantly from unity. Years of exposure and amounr.smoked were also examined. No significant elevations in risk were found at any level of smoking by household''members during childhood.. Discussion One of the most striking findings of this. study is the distribution of the histopathological cell types of lung cancer in a population-based series of cases well screened to determine lifetime nonsmoker status. Sev- enty-eight % of 3'59 reviewed'eligible cases in this report were classified as adenocarcinomas. This high proportion of adenocarcinomas and'' the paucity of squamous and small cell carcinomas was consistent across all study

-) Lung Cancer in Nonsmoking Women Tatile 5 Association between smoking status,ot soouse(st and lung cancer nsk" all lung cancer and adenocarcmoma ot the lung Adiusted odds radio' 'pouse ever smoked tooacco tov tvper, Cases Colon cancer controls Populatlon controls Colon cancer controli OR'(95 % CI1 Popuiat on controls OR195'kCll AII lung carctnomas in =420)'. ~n = 351) In- 7801 ?,nv type otnobacco 294 231 492 1.28 10.93-1 751 1129 10.99-1.691 C garettes 264 ?09. 441 1.17 10.87-1 59) 1.20 10.93-1. 5 51 Cigars 64 54. 97 1.14 (0.76-1.71) 1.26 (0.88-1.80) Pipe 63 52 110 1.117 (0.78-1.77), 1.21 (0.85-1.72) Adenocaranoma In = 2811 tn - 35111 (n - 780) ' Any type ot tobacco ?03' 231 492 1.44 11.01-2.051° 1.47 11.08-2.011° C garettes 184 209 441: 1.31 (0:94-T.841' 1.36d1.02-T.84)° Cigars 41 54 97 1.05 (0.67-1.661 1.15 (0.76-1.74) Pfpes 44 52 1,10 1.16 /0.74-1.821 1.20 (0.81-1.79) `Ad(usted tor age tcontmuousl. race twhtte. black. othen: study area (Los Angeles. San Francisco Bav Area. Southern U.S.: Atlanta.,Hbuston, and New. Orleansu annual tamdv, income (<513.000. $13.000-534.999; S35.000+1. and education l<htgh school degree„htgh schoo6 degree, some college or higherf. °P<005. centers., In the study of Kabat and Wvnder (8); a similar proportton 174%) of Kreyberg II type tumors was foundd in their series of 97 nonsmoking females whose self- reported', nonsmoking status was confirmed bv, chart re- view: In the United States adenocarcinoma iss the most common histopathological cellltvpe of primarv Iung can- cer in. women. but the proporhom ot all female lung cancer cases with all subtypes of adenocarcinomas (pap- illary, acinar. bronchioloalveolar, and solid) is 34% (SEER Public User Tape, 1978-1987). Oun studv, in which adenocarcinoma is predominant and is the cell type clearly associated with increased risk X 0 <15 from adult ETS exposures. is in contrast to several of the earlier studies of involuntary exposure to ETS. Tricho- poulos et al. (2) in the initial case-control study of lung cancer and passive smoking among nonsmoking women excluded cases of adenocarcinoma including bronchio- loalveolar; however, that study included no histopatho• logical review: They reported an odds ratio, from, 1.8 to 3.4 associated with the husband's smoking habits. Dala- ger et al. 0 6) reported a 3-fold elevated risk associated with the spouse's smoking only for squamous and small cell carcinomas and no increased risk of other cell types, of which adenocarcinoma and'its subtype, bronchioloal- 15-39 Pack Years 40-79 2 80 hg. I. Adfusted odds ratios for ail lung cancer and tor adenocaranoma oPthe lung assoaated'wuh pack-vears of exposures from spouselsL 0. all'lung cancer. trend P= 0.07; ®, adenocarcmoma. trend'P < 0:01.

Tab/e 6 Association between risk'ot lung,c.ancer ano adudt exposuresto cigarette smoke amonq nonsmokin¢.womeni ~ears ot exposure by source Alllurn[ carcinomas adiusted odds rat o' Ad'enocareinoma ,n the Id,ng, adlusted' odds ratio' 195% Cl1 95% C11'. Household exposure Spouse Everexposed° 1.21.10.96-1 541 - 38/1.04'-1 821' 0 vears 100 10 1-15 1.19 10.88-1 611 1 33'(0 93'-1 -89) 16-30 1.14 /0.82-1 591 1 40 10 96-2.05) >30 1.25 (0.91-1 72) 1 4340.99-2.091 Trend P- 0]4 Trend P=-0 03 Other househotd members Ever exposed°' 0 years 1-5 6+a 1.23 (0.97-1i.56) 1.00 1.20 10.90-1.611 1.23 10.89-1.691 1 39'11.05-1.821` 1001 1 36 10:98 -1 891'. 1 35 (0:93-1 94): Trend P= 0 12 Trend P= 0'04 Occupational exposure Everexposed' 1.34 IT.03-1 ; 3~'. 1 4i t1 06-1 971' 0 vears 100 100 1-1'5 1 23 (0:86-11J.'1'. 1 58 (1.05-2.39i` 16-30 1 45 11.05-2:001°, 1142 10.97-2.071 >30 1' 30 (0:93-T 80F !! 37 10.92-2.021 TrendP- 002 TrendP=010 Social exposure' Ever exposed° 1.58 n Z2-2.041" 1 60 (11119 -2.141' 0 1.00 1 oo 1-15 1.34 10.97-1 84) 1 29 10 89-1 871 16-30 2.01 (1.29-3.15i` 2.40 t 1 47-3'.90r >30 1.65 (0.98-2.80) 1 5010.78-2:771 Trend'P - 0 0006 Trend P= 0 002 `Adtusted tor age, race. study area, annual tncome. and education. ° Re(erent: never exposed. P<0.05. ° Too few subtectss exposed~ 16+ years. ' Social l exposure is dehned as exposure o/ 2 or mnre h/week (rom sources other than occupational and household mem6ers. ncluding spouse. "P < 0.01, veolar carcinoma, comprised 46.1% of the total female nonsmoking cases.In the Swedish study of Pershagen et a/. (35), 57% of 77 female nonsmokers were adenocar- cinomas and 31% squamous and small cell carcinomas. The only statistically significant ETS-associated'i'ncreased risk was for squamous and small cell', carcinomas, the cell types with.the highest relative risks associated with active smoking. At the present time small numbers or squamous cell and smalll cell carcinomas in our data: set preclude an adequate assessment of risk associated with ETS ex- posures for these cell types. The findings of our study lend some support to the mechanism proposed by Wynder and Goodman (36)~ whereby inhalation of sidestream smoke might prtmarily increase risk of adenocarcinoma of the lung. They sug- gested'that inhalation of sidestream smoke through the nasal passages would hinder deposttion, of respirable smoke particulates in the periphery of the lung while gaseous components such as volatile N-nitrosamines, formaldehyde, acetaldehyde, or nitrogen oxides, would c Cancer Epidemiolo/tv, Biomaricen & Prevention be likely to reach the deeper part of the lung„ Both squamous cell and smalll cell' carcinomas tend to be centraliv located, rather than in the perlpherv ofthe lung: Our studvfound statlsticallysign(ficantelevated risks of adenocarcinoma: of the lung among temale non- smokers who had had household ETS exposure or ETS exposure in occupational settings or fromiother, sources, Each of these exposures occurred during adulthood. Exposures during, the first 18' years of life were consist- ently unrelated to the risk of lung cancer. Any exposure (ever/never) from a spouse who smoked' was associated with' at least a. 30% excess risk. Increasing amount per dayand'years smoked significantly increased risk. The pattern, of risk was the same when cases were compared to colon cancer cases or popula- tion controls and was specific for adenocarcinoma of the lung. Findings for all lung cancers combined reflect the association between ETS and adenocarcirtoma of the lung dilute6 by the weak association with other cell types. The internal consistency of findings with. the two controfs groups suggests that recall bias resulting from having a diagnosis of cancer is not a likely explanation of the observed effect. The possibility remains thav non- smoking lung cancer cases and nonsmoking colon cancer, cases are not similarly motivated to remember exposures to the tobacco smoke of others. The longest~d'uration~of'i exposure to ETS is associated with the greatest elevation in risk, 1.43, for exposure of'. 30 or more years to a husbands's cigarette smoking. Although, significant trends were found for other adult exposures, the dose response was not, monotonic; rela- tive risk estimates tended to decline in the longest ex- posure category. One possible explanation is that recall'' of quantitative measures of exposure is less reliable for exposures outside the home and for household members other than the spouse. A recent ten-country study was carried'out by the International Agency tor Research on Cancer designed to validate self-reported recent expo- sure of nonsmoking women to ETS from, any source compared with the urinary concentration of cotinine. Duration,of dailv exposure to ETS from the husband was the strongest predictor of urinary cotinine (37). Studies by Pron et' aC (38) and'' Coultas et al. (39)' suggest that quantitative measures, particularly'for exposures outside the home, are less reliable than categorical measures. The lack of any association between childhood! ETS exposures and lung cancer in our study, as well as the strong,, consistent association with exposures during adulthood, contrasts with two recent reports by Janerich' et aL (22) an6 Wu.Williams et al: (40), Differences in, study design may contribute to the discrepant findings. About 25% (n = 45) of the 191 cases in the New York study were males, whereas our study was restricted to female cases In = 420) (22). The authors report that there were oniysmallldifferences between men and women in the amount of exposure to ETS measured by duration. The mean exposure of women to their husbands' to- bacco smoke was 16.2 ± 16:7 years,while men had a mean exposure of 13.0 ± 17_0'years from smoking wives. Furthermore, there was a higher correlation between exposure from spouses lifetime ETS exposure for women in the study (r = 0.51) than for men (r = 0!37). Intensity (dose): of exposure and temporality' of exposure from male and female smoker sources maydifferconsid'erabiy: Relativeliy small differences in, dose, temporalitv, and 411

-_ Lung Cancer in Nonsmokin/{ Women iable7Assocration between risk'btJunq cancer and chddhood''exposures to tobaccosmoke amonqnonsmokrng women Adlusted odds ratto+ Eversmoked' tobacco Cases Colon cancer controls Populatuon controis Colbn cancer controls OR (95%CII' Population controls OR (95%Cll AII lung carcinomas Father 196 189 420 0.91 (0!67-T.24) 0.82 (0:64-1.07), rolother 44 40 97 0.85 l0!53 -T.381 0.84 (0 j56-1.261'. Other nousehold member 177 1 S2 327 0.83 (0:59-1.181 0.96/0;71-1.29) Adenocarcinoma F.ather 139 189 420 0!9610.69-1t35) 0:89 (0.66-1.19) Mother 30 40 97 0;91 10.54-1.551 0:89'd0.56-1.431 Other household'member 125 152 327 0.81 10.55-1.201 0:91' 10.64-1,29) ' Adiusted nor age. race.:studv area, annual rncome.,ano ed'ucatuon. ° Childhoodl ts denned as ttrst 18 years ot life duration in combinatton may yield more meaningful dif- ferences in exposure than that measured by duration alone. The inclusion of' males in the New York study; with possiblv lower d'oses of ETS exposure irom smoking wives for fewer years and during a more recent time period, mav have reduced'the relative risk estimates that were not gender specific. A stud'v in northeast China, which was comparable in size to our study; actually foundl a decreased risk of lung cancer associated with ETS exposures from spouses and a suggestive increased risk associated'with paternal smoking (40). As suggested by the authors, these women had heavy' exposures to both indoor and outdoor pollutants., which, may have obscured any effect of ETS. The studies which have examined childhood expo- sures are more limited than those which have focused on tobacco use by spouses, and the overall' findings are inconclusive (3, 5, 11-14, 22, 41). Studies of'.the reliability of recall of ETS exposures suggest that recall of a parent's smoking history is less reliable than that for spouses (38; 39), and~ this mav' account in part for inconsistencies between studies. lanerich et al. (22) found a 2-fold in- creased risk associated with 25 or more smoker-years during, childhood and adolescence but no increase for childhood exposures of'less than, 25 smoker-vears (OR. = 1.09). In, most studies which have reported positive associations, the findings have been primarily for mater- pal ETS exposures in smokers rather than in nonsmokers. Correa et al: (5) found a significantly increased risk of lung cancer (OR = 1.36) among smokers whose mother smoked but no increased risk in nonsmokers and no elevated risk associated' with the father's smoking. Wu et aL (14) reported a nonsignificantly elevated~ risk of ade- nocarcinoma of the lung (OR = 1.7) in females, 80% of whom had a history of smoking, Similarly, inia Swedishl study of female lung cancer which: included primarily smokers, a nonsignificantlv elevated risk was associated with maternal (OR = 1' 8)i but not paternal (OR = 0.8) smoking (42). Other studies have failed to find an in- creased risk of lung cancer associated with, childhood exposures (11, 12, 43). None of these studies examined maternal smoking as distincr from, other childhood ex- posures. Childhood ETS exposures albne may be insuf- ficient to:increase lung cancer risk in lifetime nonsmokers but mav increase risk inipersons exposed transpiacentally or during childhood who later smoke themselves (5). The female lifetime nonsmokers with lung cancer in our study are considerably older than the temale lung cancer cases reported in the SEER program, most of whom have activelysmoked. This may represent a cohort effect; that is, older women are less likely to have smoked. The age disparity might also reflect possible differences in response among active and passive smok- ers. The lower dose of' ETS might require a longer dura- tion oi.exposure for pulmonary carcinogenesis. Although this report represents the findings of the first 3 years of a 5-year study, it is nevertheless the largest case-control study reported to date on this topic. The findings provide additional evidence in favor of a causal relationship between exposure to ETS and lung cancer in women who have never used tobacco themselves. A dose response, not likely due to chance, was apparent for exposure to tobacco smoke during adultl life from a variety of exposure sources. The association was specific for both adenocarcinoma of the lung and for all' lung cancers combined compared to colon cancer. Acknowledgments The authors are gratetul,for the cooperation ot all of the parnctpattng, hospnals in the dve stud'wareas and the many physicians who helped make this study possible. The authors also thank Rafael (arpa, Youping Ltn, Gail Smtth. Mahboob 5obhan. and Diego Zavala tor programming and analytic support: Laurel Cederqutst. Annie Fung. (udv Goldstein. Helen GregorY:,and'Anne Morse tor field supervision: Drt Nancv I. Haley and Caren Axeirad for directing the urtnary cottmne analvses: and the dedtcated' medical I record abstractors and interviewers in each study centen References 1. Hirayama:,T. Non-smoking wives of'heavy'smokers have a highernsk, of lung cancer: a study from lapan. Br. Med. I., 282: 183-185. 1981. 2. Tlnchopoulos. D., Kalandidi. L.,, Sparros, L.. and McMahon, B. Lung cancer and passive smoking. Int. l. Cancer, 27:.1-4; 1981. 3. Garfinkel. L. Time trends in lung cancer mortality among nonsmokers and a note on passive smoking. 1. Natl:, Cancen Ihst., 66: 1061-1066. 1981., 4. Chan. W. C.. and Fung, S. C. Lung cancer in non-smokers in Hong Kong. /n E'. Grundman ted,l: Cancer Campargn. Vol. 6. Cancer Epide mrobgy, pp. 199=202: StuttgAn: Gustav Fisher Verlag, 1982. 5. Correa. P., Pickle. L. W.. Rontham, E.. Lin, Y.. and Haenszel; W. Passive smoking and'hing cancen Lancet. 2: 595-597. 1983. 6. Trichopoulos. D.. Kalandidi. A., and Sparros. L. Lung cancer and passive smok ng: concluston of Greek study. Lancet, 2: 677-678, 1983. 7. Koo. L. C., Ho. I. H-C.. and Saw; D. Active and passive smoking among female lung cancer patients and' controls in Hong Kong. /. Exp. Clin. Cancer Res.. 4: 367-375. 1983. 8: Kabat. G. C.. and' Wvnder, E. L. Lung cancer rn nonsrnokers. Cancer /Phila.l. 53: 1214-1221, 1984.

Cancer Epidemrobgy, Blomarken 6r Prevention 43. 9. Gillis. C. R.. Hole. 0 I'.. Hawthorne. V M. and'Bovle. P. The etfects of environmental tobacco~smoke in two uroan communities rn the 1Nestt of Scotland. Eur I. Resper. Dis.. 133 ISuppl.l:,122-126. 1984 10 Buffler. P A.. Pickle, L. W , Mason, T I. and Contant: C. The causes of lung cancer in Texas. in M Mizell and P Correa ieds,I: Lung Cancer: Causes and Prevennon, pp. 83-99. New 4ork: Verlag Chemre /htema- t;onalJ Ihc..,I984!. 11. Garfinkel. L.. Auerback. 0.. and louben. L. Involuntarv smoking and lungcancer: a case-contro6studv. I. Natl:,Cancerlnsti. 75 463-469. 1985. 12. Akiba. S.. Kato. H., and Blot. W. I. Passive smoking and Iung cancer amwsg lapanese women. Cancer Res.. 46 4804-4807. 1985. 13. Sandler. D! P.. Wilcox. A. I.. and Everson. R. 8. Cumulative eftects of lifetime passive smoking on cancer risk. Lancet. 1: 312-314. 1985. 14. Wu, A. H.. Henderson. 8. E.. Pike. M. C.. and Yu. M, C.,Smoking and other risk tactors tor lung cancer in women, I1 Natl. Cancen Inst.. 74: 747-751,1985. 15. Lee, P. N:, Chamberlarn, I.. and Alderson. M: R. Relationship of passive smoking to risk of lung cancer and other smokmgassocaated diseases. Br. I. Cancer. 54: 97-105. 1986. 16. Daliger, N! A.. Pick'le. L. W.. Mason. T. I., Correa. P.. Fontham, E.. Stemhagen. A,.,BuUler; P. A.. Zeigler. R. G.. and Fraumem. I. F., Ir. The relation ot passive smoking to lung cancer. Cancer Res.. 46: 4808-481' 1,. 1986. 17. Hitmble. C. G.. Sarnet. ILM.. and Pathak. D. R. Marnage to a smoker and lung cancer risk. Am. 1. Public Health, 77: 598-602. 1987. 18' Koo. L. C:. Hb, l. H.. Saw. D., and Ho. C. Y. Measurement or passtve smoking and estimates otJung cancer risk among nonsmoking Chinese lemales. Int 11 Cancer. 3'9: 162-169; 1987. 19 1 L'am. T. H.. Kung;,T. M,.,V1+ong, C. M.. Lam. W: K.. Kleevens. I. W. L., Saw. D.. Hsu. C.. Senevrratne. S.. Lam. S. Y., Lo. K. K., and Chan. W. C. Smokrng; passive smoking and histological types in lung cancer in Hong Kong Chinese women, Br: I. Cancer. 56.` 673-678. 1987. 20: Inouve. R.. and Hirayama. T. Passrve smoking and lung cancer in women:,/n: M. Aoki. S. Hisamrchu and S. Tominaga leds.l. Smoking and Health. 1987. pp. 283-285. Amsterdam: Elsevier Science Publishen, 1988. 21. Geng, G. Y.. Liang. Z. H., Zhang; A. Y.. and Wu: G. L. On the relationship between smoking and' female lung cancer. In- M_ Aoki. S. Hisamtchr, and'S.Tominagaleds.I:Smokrng and Health, 1987. pp. 483- 486. Amsterdam: Elsevier Science Publishers. 1988. 22. lanench, D. Ti., Thompson, W. O.. Vareta. L. R.. Greenwald. P., Chorost. S:. T'ucct. C.. Zaman, M. 8.. Melamed. M. R:, Kielv. M.- and McKheallv. M. F. Lung cancer and exposure to tobacco smoke in the household. N. Engl. I. Med.. 323: 632-636. 1990. 23. Environmental Tobacco Smoke: Measuring Exposures and Assessing Health Effects. Committee on Passive Smokrng., Board on Environmental Studies and Toxrcologv: Natronal I Research Council. Washmgton. D.C.: Natronal'Academy Press. 1986 i 24. I. K:,O'Ne+ll, K. D. Brunnemann. B'. Dodet. and D. Honman /eds.l. Environmental Carcinogens. Methods of Analysis and Exposure Measure- ment. Vo1. 9, Passive Smoktng„ IARC Scientdtc Publications no: 81. Lyon. France: Internatronai Agency for Research on Cancer. 1981. 25. Spitzer. W. 0.. Lawrence. V., Dales. R.. Hill. Gl: Archer: M: C.. Clark, P.. Abenhaim. L.. Hardy, I.. Sampalis, 1:, Pinfold. S. P:, and'Morgan. P: P. Links between passive smoking and disease: a best evidence synthesis. It A repon of the working group on passtve smoking. Clin. Invest. Med., 13: 17-42. 1990: 26. Wu-Williams. A. H., and'Samet. I; M. Envrronmentalhobacco smoke: exposure-response relatronshipsin epidemrolbgic studies. Risk Analysis, 10:39-48.1990: 27: Halev: N. I.. Hoffman. Dj, and Wvnder. E. L. Uptake of tobacco smoke components. In: Mechanisms of Tobacco Caranogenests New York: Cold Spring,Harbor Laboratory: 1986. pp, 3-9: 28. Haley; N. I., Axelrad. C. M,. and Tilton. K. A. Validation of self: reported smoktng,behavron biochemical analysis or cohmne and thiocy- anate. Am. I: Publ. Health. 73`. 1204-1207, 1983. 29:Langone, l. I.. Gjika, H. B;,,and Van Vunakis. H. Nicotine and its metabolites. Radioimmunoassays tor nicotine and cotinine. Biochemistry, 1'2'5025-5030,1973. 30; Hoffman. D. W:. Halev; N. li, Adams. l. W.. and Brunnemann, K:D. Tobacco stdestream smoke. Uptake by nonsmokers.,Prev: Med., 13: fi08- 617. 198M, 31. Mattesen, M. E., Boyd, G.. 8yar, D.. Brown. C.. Callahan,J. F., Corie,. D., Cullen, I. W.. Greenblatt! I., Haley, N. I.. Hammond. K., Lewtas. Il, and Reeves. W. Passive smoking on commerctal'atrhne flights. IAlv1A,. 26 i: 867-872;,1'989. 32. Mantel. N!, and'Haenszel, W. M. Statistical aspects of the analysis of data from retrospective studies of disease. /. Natl. Cancer Inst., 22. 719- 748, 1959: 33. Schlesselmann. I. I. Case-Control Studies: Destgn, Condua. Analysis. New York:,Oxford University Press. 1982. 34. Klembaum.A: G.. Kupper. L. L., and Morgenstem. H. Epidemrologlc Research: Principles and Quantitative Methods. Belmont. CA: Lifetime leamtng Publicattons. 1982. 35. Pershagen. G.. Htubec. Z., and'Svensson. C. Passive smoking and lung cancer in Swedish women. Am. ( Epidemrol.. 125: 17-24, 1987. 36. Wynder. E. L., and Goodhsan„M. T. Smoking and lung cancer- some unresolved issues. E'pidemrologtc Rev., 5: 177-207, 1983. 37. Riboli, E.. Preston-Martin. S.. Saracctu R., Haley. N. I., Trichopou/os, D., Becker. H., Burch, 11 D.. Fontham.,E. T. H..,Gao,Y. T., Jindal, S. K., Koo. L. C.. L'e Marchand. L.., Segnan, N., Shimizu. H.. Stanta, G., Wu- Williams. A. H.. and Zatonski, W: Exposure of nonsmoking women to environmental tobacco smoke: a ten-country collaboratrMe study. Cancer. Causes Contrdi 1: 243-252: 1990. 38. Pron, C. E.. Burch,:l. B._Howe. G. R., and'Miller, A. B, The reliability of passive smoking histones reported in a casetontrol study of lung cancer. Am. I. Epidemiol., 127:167-273, 1988. 39: Coultas, D. B.. Peake. G. T., and Samet. I. M. Questionnaire assess- ment of Ilfehme and'recent exposure to envrronment. Am. I. Epidemiol., 130' 338-347: 1989. 40: Wu,VVilliams. A. HL Dai:,X. D., Blot. W., eral. Lung cancer among women in nonheast China. Br. I. Cancer. 62: 982-987:,1990.. 41. Sobue. Ti., Suzuki, T., Nakavama, N:. et a1. Association of indoor air pollution and passive smoking with lung cancer in Osaka. Iapan, Gan No Rinsho, 36: 329-3331 1990. 42. Svensson. C., Pershagan. G.. and Klommek, 1. Smoking and passive smoking in relation to lung cancer in women. Acta Oncol., 28: 623-629, 1989. 43. Gao. Y-T., Blot. W- I..Zheng W., et al. Lung cancer among Chinese women, lht. 1. Cancer„40: 604-609. 1987.