Tobacco Documents Online

(CA:ICER Rr'SEARCH 46. 4808-t811. September 1986), J The Relation of Passive Smoking to Lung Cancer' Nancy A. Dalager,' Linda Williams Pickle, Thomas I Mason, Pelayo Correa, Elizabeth Fontham, Annette Stemhagen, Patricia A. Buffler, Regina G., Ziegler, and Joseph F. Fraumeni, Jr. Epidentioloay ond Biouarinia Pro~ntw, Kotioad Cancer lnttitws, &rMtsdo, Muryland 20891 fN. A: D., L. W P., T. l: M., R. G. T., l. F. F. f,' Department ojpotha%,, Louifiono Siate Utti.ersiry Medical Center, New Orfeaet, Loririana 70111(P. C. E. F.%,Diriaion of £pidnnio/u8Y and Disrnu Control, New lersry Stote Dtpa twkw ojHeoltA, Tnnton. NlwJaxy 0d690/.1. S.J,• and E'oidenfoleVResrw[A Unit, University of Texas ScAool of PrD/ie HraltA, Hototon, Tesat 770Y3 (P. A. 8.J- ABSTRACT To evaluate the role of passive smoking la the development of I." cancer among .oesmoksrs, data were pooled trom tAree large hscident case-control fnter.iew studies. Nieety-nine l.et; cancer cnses and 736 eoatrols never a.ed any torm of tobacco. O.erall the adjusted odds r.tkr for lung cancer among noosmoka» ever living wfri a sawker was 0.8 (95% confidence leter.ak 0.5-13) risisa to 1.2 among tbo.e exposed for 40 or toon years. Persons living with a spow who smoked eiptrsttes were at increased risk (adjusted odds rado,l3: 959i confidence ister.aa, 0.8-2.$): When adjusted for age and 6ender, tBere was a siptifloeCtreed is risk with incressitta amomts smoked per week by the spouse (P : 0.05) and with comulative pack-yean of exposare (P = 0.03). This effect was limited to fetnaks, espedally older women waose huskaeds were heary smokers. The elevated risk associated with spoase smoking was restricted to squamous and small celt carcinomas (odds ntio, 2.9; 959e confidence interval, 0,9-93), which provides additlomJ eridettce linkint passive smoking to lung cancer. 1NTRODUCTION. The respiratory effects of passive smoking among nonsmok- ers are of increasing concern; evidence suggests that such ex- posure may increase the incidence of bronchitis and pneumonia in early life (l) and decrease lung function among nonsmoking adults (2). Recent attention has centered on the possible risk of lung cancer among nonsmokers exposed' to environmental' to- bacco smoke (3-6), although epidemiological studies have been limited by the small number of cases available for analysis. The National Cancer Institute has recently collaborated on three large case-control interview studies of lung cancer which in- cluded questions about passive smoking. One of'these studies, conducted in Louisiana, showed an increased risk of lung cancer among ever-married nonsmokers who had a spouse that smoked (6). To increase our sample size for study, data on nonsmokers from all three case-control studies were pooled and analyzed. MATERIALS AND MTTHODS Persons reporting that they had ttever used any tob.cs+o products (cigarettes, pipe, dkars, souQ or chewing tobaoco) were selected' from three casaconttol interrisw studies of lung caocer conducted in Loui- siaoa, Texas, and New Jersey. The methods used in each of these studies have been reported pnevioasly (6-9). Hecause all three studies were designed in collaboratioa with the National Caooer tnsdtute, they were similar in many respects, as shown in Table i. Medical and pathology ret:ords were abstracted to determine the final diagnosis of each case. All were incident primary lung cancer ases disinosed between 1976 and' 1932. with nearly 100% histologically confu'med. Personal intetviews were oonducted with the taudy subject or, if the Reah.d 4I23/!6c araepwd S/x7ll& The oow of publintiom or tki4 artide weee dekayed im pert by the ptymea of pqs e6aresa This aroek mmt thaefon be hetsblr mrrked ad.erelsnasae ia accordance witit tt U.S.C. Ssetbs 1734 soldy to kWinta thit ttut t Support.d by the Natioaai' Cancer lmtimt.. USPHS Coetrset N CM1023- 00 (Texar). N-(.'P9L1031•00 (New Jersey). and N{Ff-1023-00 (Soat! l.owis4 aoa), Work i. North touietrea w.s supported by a{etit horl the laolaa.r Divisfon ort6e Ametina Caacer Society to Dr. Pelsyo Corn.. Cattat ' To wbos requests for reprints should be addras.ed, ae National lmtilQp, Laadow 6nildisg, Boaa. 3A06, Detlwsd.. MD 2W2 subject was deceased or too ill'to respond, with a sutrogstc respondent Except where noted, the questions asked in each of the study areu vr. very similar. Details regarding the source and level of passive smoking exposures varied according to study area (Table 1). Texas provided the kast specifirdata by ascertaining only if any member of the subject i house, hold smoked while the subject was either a child or an adult and the total number of years of that exposure. New Jersey, on the other hand, inquired about the smoking habits of each household member dtuinl the subject's youth and adulthood. Louisiana requested infortnation regarding the smoking patterns of spouse, mother, and father, but not other members of the household. An estimate of the potentul underre. porting of passive smoking exposure in Louisiana is provided by the New Jersey control' aroup; 6% of the nonsmoking tnales reported passive smoking originating from household members otlier than spouse, mother, or father. The final study population abstaining from tobacco consisted of 99 histologically confirmed lung cancer ases and 736 controls, This represented 1.2% of all male cases and 9.196 of all female cases in the original study populations, as compared to 15.0% of the male eoatro4 and 44.9% of the kmak controls. The rmal data rtle included all variables that could be standardized across the study areas. Several potential confounders were examined, including aettder, rsoe, ase, study area, respondent type (subject or next of kin), any sey. reported chronic lung conditions, employment in suspected high-risk industries, asbestos exposure, carotene and total vitamin A intake, and whether parents had smoked. Due to the limitation imposed by sm.U numbers, age was dichotomized into two age Qroupi (<6J years and 63+ yean). Logistic analyses utilizing three age groups did not sub. stantially alter the adjusted odds ratios reported here. High-tisk indu.. tries were those identified in a recent review of cast-control studies ot lung cancer (10) and included fishing, construction, lumber maoafaN turina, chemical and petroleum manufactutinR, primary metal mana- facturina, and shipbuildina. Nutrient indices were cakrolated from the food frequency questions for each study am, using nutrientcoauat ia a typical portion of each food (11, 12). Low intake was deftned as the lowest quartile of intake for controls from each area. Becatne of the comparability problem resulting from the selective exchesion of petsoas diasnosed with chronic lung conditions in Louisiana, analyses wete duplicated excluding all persons who reported having a chronic la+q condition in all three study areas. Statistical methods included the calculation of crude ORs' for kms caacer risks associated with passive smoking exposures. Becanse of th small numbers in tflisanai"ysis, Oits were calculated using a 03 rd adjustment (13). Dose-response effects were examined using atttatiw analysis and the Mantel-Haensxel test for trend (14). The lapstic modd was used to exclude the etrects of potential contound'ers and to obtai maximum likelihood estimates of the adjusted ORs (15-1n lkcidos concerning parameter deletions for the model were bned oa tDe t statistics for significance of the individual parameter estimstezr o6 changes in the value of the log likelihood, and on the soodoest•ol fx of the model as measured by the comparison of predicted to obtervel ORa, both stratified' and crude. Maximum likelihood 95% Cls tor tle ORs were calculated from the logistic model (1S) or from the attttibl auslysis ( i E). RESULTS The data from all three study areas were merged to ewst0t the effect of any passive smoking exposure in the bome el'~ s'The abCeeriatior med arc OA, odds nti6t Ql, ooande.ea 480g

PASSIVE SMOK1NG AND LCNG CANCER Table I Comparison of the three lung cancrr care-control intrrvirw sradirt front whfch the nonustrs of roDacco were selected Study Louisiana Texu Wew Jersey Geographic arc. 29 Louisiana parishcs Gulf Coast of Texaa 6high•risk areas of New Jersey Casc diagnosis period 1979-1982 1976-1980 1980-19g1 Study design Hospitat based Population bued Population based Raciat groups included Whites/bfacks Wh'ites W hites/blacks Gender groups included Maks/femalb Makes/females Males Coopention rate Cases 91:! 91.5 87.7 Controls 93.5 87:9 73.4 No. of cases 1057 (M) 31S (F) 462(M) 454(F) 896 (M) No. of contro(s 1073 (M) 320 (17 451 (M) 464'(fl 1043 (M) % histologica(!y confirmed 97.8 100 100 passive smoking dau Mother, father, spouse Ever any household member Specific household member Nonuess of toh.oou Na of cases 8(M) 28 (A) ! (M) 42(F) tb (M), No. ofcontnob t77 (M) 1S6 (F7 4d(M) 1%(F) 159 (M) , No. of compkted (nterviews x 100 No. of svbjects contacted for interyiew ronment. There were slight variations in the passive smoking Tab(e2 odds.otwrfortwtscancerauociaredwiaa,poitar smW*S e=pow,r questions among the three study areas. However, since expo- anans e.e.-'"m^ed nonrten of toea«o ia tott;:tam a,td New Irnry Total d the bulk of assiv from p arent and rc p res t p ouse en e p res s s e su smoking experienced by nonsmokers in the home, the decisionn to pool data on whether subjects had ever been exposed in the Cases Controls Crnde OR Adjusted OR' home environment appeared reasonable. There was no apparent increase in the riskk of lung cancer among those who reported ever living with a household member who smoked (crude OR, 1.00; 95% CI, 0.64-1.56): Controlling for the strongest con- founders (gender, age, and study area) reduced the OR to 0.84 (95% Cl, 0,32-1.34). No significant differences were seen in the risks across sex and age strata or according to cell type of lung cancer. A crude summary estimate of the duration of passive smoking exposure per individual was calculated' for the Louisiana and' New Jersey data by taking the maximum number of years that smoking was reported for mother or father and adding the years reported for spouse. In Texas, the reported value for years lived with any household member who smoked was used.ln a strat- ified analysis, adjusted for age and gender, there was a slightly elevated OR of 1.24 (95% CI, 0:62-2.31) for those reporting 40 or more years of living with a smoker compared to 0.86 for <20 years and 0.82 for 20-39 years. These ORs were not significantly different from unity and showed' no significant trend. Since combining all known sources of passive smoking ex- posure might mask the effects of time period and intensity of exposure, the data wero further analyzed with ttigard- to specific sources of exposures. Because the Texas study lacked detailed data on the source and intensity of the passive smoking expo- sntes, the remainder of the andysis was restricted to data on aonusers of tobacco from the Louisiana and New Jersey studies. The most complete information available for this combined population concerned the smoking patterns of the spouse. A total of 48 casts (22 males and 26 females) and 466 controls (318 malmand 148 females) were ever-married nonusers of 48 466 1.97 (1.03-3.42) t.47 (0.76-2.83) Amount spouse smoked/.vk <140 ciyuettes t:36 (0.41-4.2t) 140-279 ci`atettes 1.31 (0:48-3.47) 280+ cipmtes 2:7t (0.Y4-E_s2) Significans treed (P - 0.05) for amount smoked Duration of spoux smoktng acposura 1 20 yr 1.73 (0.s2-5.42) 21-3o yr 1.78 (o.60-s.)o) >30 yr 1.24 (0.42-3.53) Pack-n of eYposurc <20 pack-yr 0.7! (0.17-3.03) 20-35 pack-yr 1'.90 (0.Sb-6.07) >35 pack-yr 2:1'S (0.84-5.40) Signifiaat trend (P w 0.03) for pack-yr of expowre • Overall OR (1.47) adjusted for sender, ase, and study area in the logistic model (1S-17); a!l otber ORa adjusted for aeodee and age (9a). Cl, 030-2.90) ~3g`hilf appears: to reflect the greater frequency ~ ~d amo»at at[~~~flttsbaadt of'nonsmoking wometl 1contpared" t(i'the wives` of noasiaokitig men. In the control series, women were much more likely than men to have a spouse who smoked' (48.0% compared to 18.2%), and the average exposure originating from the smoking spouse was greater for women than for men (mean pack-years, 40.6 for women and 27.1 for men). The'Qreatese,qlsk was seen amo . • oldeuwomen(63+years)wh iRtMfhfl ' , .~: , ... ~ yrAtcdosel~espcn~i;elssitsashipe(l~3~ta~ c;years°of ez A'd?u°w'stmQ for coafoundina by sender,: age, ana snudy area n- dnoed the odds ratio to 1.47. When controlling for age and teadcr, a significant upward trend was seen for increasing 4mount smoked per week by the spouse (P - 0.05) and cumu- htive pack-years of exposure (P - 0.03). Duration of spouse smoking, independent of amount, showed no consistent pattern. Sex differences in ristc were observed, with aQustsd ORs of 1.96 for fetnaks (9i% (3, 0.82-4.70) and 0.93 for males (95% ffxposctre. Despite a suggestion of increased risk for men in this category (OR, 2.48; 95% Cl, 0.52-10.23), the numbers were small, and no dose response was detected. The ORs for lung cancer among nonsmokers were examined for the following histological types: adenotxrcinoma; squamous and small cell txrcinomas; and other cell types which included bronchioalveolar, undilPerentiated, mixed, and not otherwise specified carcinomas, as well as carcinoids. As shown in Table 3, adenocarcinoma accounted for approximately one-third of the lung cancers in both sexes combined. A larger proportion 4809

PASSIVE SMOKING AND LUNG CAtvCEA , J Tabk 3 CdJ rype diartlbution of Jvn; rwcd onwnr e'ef'n+°med dOnuttnt Of fobOC[o b)'SGx Msk Femilt Cell type wo, % No. x Kdencarcieom. 7 31.8 9 34.6 Squamoua uW .enaU au arcinoma 9 40.9 5 19.2 Ochs' 6 27•3 °2 °6.2 BioecNio.tveotrcarcinoma 2 9.1 3 11.5" unaitr«euriatee 2 9.1 3 t s.s stiaed 0 0 1 3.e tvaotba.lae>p.cineaea1e;: 2 9.1 3 11'•s Crcu osd 0 0 2 7.7 Tabk 4 CeU 07rgecjRe Ob fa hy aaa..eocf.ee/ wftA JPMar siwokiRj txpum sawV e.erM.mrie4 so+.ae,eof a"eev in Loririw awf New Jaser Total Cell type Cases Conttob fSrude OR Adjusted UR• Adeaoarciamns 16 466 1.25 (0.44-S.S1) 1.02 (0J3-3.16) SqwmoYS aed sieall B i 14 466 2:6 t(0.93-7.32) 2.la (0.91-9.10) soma wc a Ott>et tti 466 2a1(0.61-3.33) L31 (O.tO-3.37) • l,opstie modd indYded sender. qe, and uudy am n potentiul oonfounders (18-17)M of squamous and small cell carcinomas was seen for men than for women, while other types of lung cancer were more common in women than men. The adjusted ORs (Table 4) associated with exposure to a smoking spouse varied from 1.02 for ade- noCarciIIoma to 2.88 for squamous and small cell carcinomas. The elevated risk for the squamous and small cell category was due mainly to the female cases in Louisiana; all five of the nonsmoking women with these cell types had spouses who smoked. The spouses of four of these five women smoked at a level greater than 25 pack-years. The men whose wives smoked showed a moderate increase in risk for squamous and small cell carcinomas (crude OR,1.48; 95% Cl, 0.34-6.39). Exclusion of all subjects who reported having a chronic lung condition did not alter the risk patterns. Except for gender, age, and study area, no confounding was detected. The increased risks for lung cancer associated with passive smoking were not aooounted for by race, respondent type, any self-reported chronic lung condition, employment in a high-risk iatdustty, asbestos exposure, total vitamin A or carotene intake, or smoking by the parents. When the referent group was restricted to those persons reporting no psssive smoking exposure from either a spouse or parent, the patterns of risk remained consistent with those we have presented. DISCUSSION i i '- i: . 0 . ~ F..1 ~: ra ` women although the excess was nonsignificant and lacked a dose-response relationship. In our study the risk of lung cancer was not increased when passive smoking exposures from child- hood and adulthood' were examined collectively, emphasizing the need to obtain source-specific exposure data When the lung cancers were analyzed" by cell type, the in- creased risk associated with passive smoking appeared rtti stricted' to squamous and small cell carcinomas, the types most closely linked to active smoking (19). This pattern suggests that passive smoking may contribute to the risk of lung cancer through mechanisms siitWar to those of active smoking, al- though sidestream smoke contains higher concentrations of certain compounds, such as nitrosamines, compared to main- stream smoke (20). In a recent case-control study by Garfinlid et el. (2t)6 significant risks for both squamous cell carcinoma and adenocarcinoma were observed among nonsmoking women exposed' to a spouse smoking . at home, with the risks for squamous cell cancer being 3 times greater than for adenopr- cinoma. Among nonsmoking Chinese women in Hong Kong, Koo et al. (22) found that the risk of passive smoking way greater for squamous and small cell caneers than for adtnosar. cinomas. Although our analyses included nonsmokers from three luge series of lung cancer, the small number of cases still pnxluded any definitive answers on the carcinogenic effects of passive smoking. Other limitations concern the difficulty in quantifying exposures from passive smoking derived from interview data and in detecting relatively iow-kvel effects. Since our study na based on questionnaires, it was not possible to evaluate certaia other exposures (e.g., indoor radon daughter products) thatmzy affect the risk of lung cancer among nonsmokers. Our study was also limited by the assessment of pauive smoking exposures experienced only in the home environment and by the use of a relatively crude measure of exposure. We assumed for this analysis that the amount and duration of a spouse's smoking habit approximated the passive smoking ex- posure realized by an individual at home. Fuller chuacteriz.. tion of passive smoking should address the intensity of expo. sure, a function of the amount of time spent in close proximity to a smoker as well as the amount that individual smokes in our study, the sex differences observed' in exposure and risk suggest the desirability of continuing to focus attention on the nonsmoking wives oJ'smokers, while encouraging the collectios of data on workplace and other nonhousehold exposures to ambient tobaooo smoke. Whenever possible, future epidemiological studies should incorporate laboratory measurements of tobacco smoke cooa stituents and by-products such as cotinine, the tnaior metabotite of nicotine detected in body fluids. Among nonsmokers, a dow . le Of un~Mncer response relationship has been observed between the kvds of ° 1, =" uriaary cotiniae and self-reported exposure to passive smoking 1r . iaatal by tbe ataoaut ~icTea ,,p~t~eitl~Small numben lung canaer, mainly of the squamous and small cell typea, and relatively low exposures msde tt diffictitt to asxss the role confirmation will probably require larger study sizes as well a of passive smoking anwng tobacco-abstaining men whose wivea more extensive and innovative assessment of exposure to aavi- smoked. While the overall ORs were not statist3al!!r sigaitkant, ronmental tob.cco smoke. , nonsmokers suggest that ettnioAmen 'iibotc~ai'estl~ Pa:s;ve smoking ittcreasa the risk of ""'° ~b ';"" ~_ _~ ;A ~ (23 24) Thus while the available epidemiological data a th~•C>ndtoE.^at~a~y "doae~ie,penJes `rtsk~of~;ltt~-aoat~,am~opg ~.>~camt~ia ~ con ~' ~ ` ` ~wr(th ottie' obaeivaijons iiir"te AC7i1VOWLEDGMENTS Greece reported a sigtliSant increase in the lung cancer risk of We am bidebted to Dts. Robeet N. Hoover and Wd6am J. BW d nonsmoking womea whose spousea smaked Subeeqttently, the tlx Natiooal Canos Latitute and J.net B. Scboenberg of tlr 1~ prospective survey of the American Cance: SocietY (5) found Jetxr staa Dey.rtmeot of Health for their ediwri.t revie* ad ** ao elevated tdsk for passive smoking atltong nonsmokiltg ydtior aad to Diaasa Jea.eefor tbe manmaiptpcep.raom 4910 litesstute. Nintyams (3) in Japan and TricDopoulas et aL (4) in

PASSIVE SMOKING AND LUNG CANCER ,REFERENCES f. Harltp, S•, and Davies. A. M. Infant admissions to hospital and maternal smokin8. Lancet, 1: 329-532, 1974. 2. White, J. R., and Froeb, H. F. Small airways dysfunction ia nortsmot'en cAronically esposed to tobaxo mwte. N. En81. J. Med., 302.• 720-723. 1980. 3. Hirayama, T. Non-smokine wives of heavy smoken have a h(BAer risk of lung cancer a study in Japas. Br. Med J., 282. 183-I8S, 1'981. 4. Trichopaubs. D., KaEandidi. A., Spartvs. L, aod MaeMahon. B. Lung cancer and panive smokine. IM. J. Cancer,l7c 1-4, 1981. ~. Garfialiel; 14. Time tren& in luaf caae.r mortality amoe8 nosstnolws 1ad a note oe pmive smoiiin{. 7 NatL Caooer Imt., 66: 1061-1066, 1981. 6. Corte.. P., Pidck, L W., Foatham. E., Lia, Y» and Haeaatel, W: Pwiv. smokin{ and luat; cancer. i+aoet: 2 595-597, 1983: 7. Cotte., P., Pickie, 4 W., Fonthaa E., Dalyer, N. A., and Haemad. W. The nwss of IunB e.noer in Louisian. Gc M. MimU and P. Correa (ada.)~ Lung Cancer Cawa and Ptsrentior~ pP. 73-82. Drwfield Beacl, FL• VerlaB Chemie feoaaatioeal, Iaa, 1984. 8. Hullkr, P. A., Pickle, L W:, Masos. T. JL, and Contnot, C. The nuas of luu{ wncer L Tesaa. 1ir. M. Mhdl aod P. Coma (eds.). Lung Cancer Causes and Ptevesdoa, pp. i3-99. Deerfield Heach, FL Vat.g Chemi. Inteeoatibwl; Im.,1984. 9. Ziegler. R. G., M.pn, T. J» Stanhr8ea. A., ef d. Dietary atoeeae aad vitamia A and riak of hog cancer aaaoet white aoea is New Jersey. J. NatL Canar lamt., 7!c 1429-1433,1984. 10. Pickle, L W.. Corrta, P., aad Fonthaoti E. Recent tsae-coetroi studies of lun6 cancer in the United St.ns. 1n: M. Miscil and P. fiortea (eds.), Lung Canar. Cames and Pte.entian, pp. 101-113. Deerfield Be.eh, F1: Verly Chemie Intemational, lee., 1984. 11. United States Departmeot of Apiculture H'aadbook No. 8. Waahingoos4 DC: United States Goveenment Printing Office, 1%3.. 11 United Ststes Deparsmeat of Apieulturt Handbook No. 8-1' to 8-& Wash• in`ton. DC: United States Government Printing Office, 1976-! 982. 13. Fkiss. J. L Statistical Methods for Rates and Proportions. pp. I 1981 S 43-46. New . ons, nc., York: John Wiky & 14. ManteL N. Chi-square tests with one de8ree of freedom; extentions of the Mantel-Hacnsrel procedure. J. Am. Statistics Assoc., Sd• 690-700; 1963. 1S. Pickle, L W» and Martin, E- J, Automated analysis of casecontrol'd.u using the SAS system. Lr: Prooetdinp ofthe Fifth Annual Conference of the SAS UscnGroup InternttioaaL Cary, NC: SASlnstitute.Inc., 1980: 16. Cox, D. R. Analysis of Binary Data. Londoa: Methuen, 1970. 17. Entdmaa, L Stcpwiae logistic a8ressioa fe: W. 1. Dixoo (ed.), BMDP . Statistical SoAwart, pR 290-344. Berkeley. CA. University of California Pres, 1983. I8. Gart, J. The eomparisoa of peopottiotur a review of si{nifrance team eosfidbace latei.als and aqitrtman for stratifieatioa Rev. Iat. Statbtia lau. 148-1'69, 1'971. 19. DnB, R., Hill, A. Il., asd Kteyber& L The si{oitSe.eee of eell type in rdydos to the etiolop of htn's tmKet. Br. J. Caoeer, !): 43-4d, 1957. 20. Btaanemaoa, C. D., Adtma, J. M, Ho. D. P. S„ and Hoffmm, D. The in/luenoe of tobacco stnoke os indoor atmoaptierea: IL Volatile and toEacco. speciBe aiao.aminq h aaaie- and sidestream smoke aed their taetribatioa to isdoot poUutio.. l+c Prooeadinp, Fourth Joint Conference on Semioe a Earitosmeatal Pollutants. pp. 876-880. New Urleaea, Lok, 19T7. WasbiW ton, DC. Aaserican Chemial Seeiety. 1978. 21,. Garffafsl, L. Auerbach, O., aed Jonhen. L Iavoluaeary smokin8 .nd hu8 caser. a eo.-eoauol study. J. Nat4 CatKa Inat., 7J: 463-469, 1985. 22. I(oo, L C, Ho. J. H. C., and Lee, N. An analysu of tovr rbk faeton for lung caeosr ia Hong [otst. lut, J'. Canoer, J!: 149-133, 1985. 23. Wakt, N. J» Borehus, J., Baiky,A., Ritehik C., Haddow, J. E., and Knight, G: Urioary, eotdnieg aa marhx of breathing other people's tobacco smoke. Laocet, 1: 230-231, 1984, 24. Maqukuts, S„ Taminao, T., Kitaee, N:, at aL Etfetv of enriroamental tobacco smoke on urinvy cotinise pttset'soa ita aoasmoiuxs. N. EoBL J. Med, 311. 828-t32, 1'981. 4811 4