Tobacco Documents Online

ANOREWS OFFICE PRODUCTS CAPITOL HEIGHTS, MD IK)

17. Relationship Between Passive Exposure to Cigarette Smoke and Cancer Jo:u:thnn M. ftnmet Asaociote 1'rnfencor, Nepnrtment of M.•dic3ne~ nnd Ohr Nrr Mrrico 'fumor ifegiotry, Cancer Center, University of New Mcxicu, Albuquerque, Nev Mc.ico 67151 INTNODUC'1'lON [buenn] uu::ucintionn between oclive cip.::rettr mm.. kin'• 'ind cancer of the luuM nnd uther sites hnve bcun lone ectnbii:di;d nn the hasis of extensive toxicoloPienl, exprrimental, nnd epidemiologicnl evidence. Only recenUy, however, hat: pnnriveexpoeure to tobacco smoke been considered an a potentinl ri^k factor for luntl cancer in nonsmokers. This putntive role uf passive smoking hns become an emotionnlly chnrYed and hiphly controvursinl subject with pote.ntiully important regulatory and economic implications. Tobacco industry argumenta drfendivr the individunl's rip.l:t to free choice concerning smoking would bo severely damaged if passive smoking were shown to cause cnncrr In nonamokere. Tbe prevalence ot passive smoking in the United Stntes furihr•r emphasizes the potential public henlth consequences of thie; exposure. Yricdmon and co-uorkers (I) quentianed 37,UEil nmmeokiny membera of s health maintenanceorganizntion conce•rninp, pns':ive• amoking at home and elsevhere. Overall, 63 percent reported somr• . expoeure and 34•5 percent received at least 10 hours per wrrk- Unpublished flndings from on ungoing case-control study In Nce Mexico ohow that 29 percent of nonsmokine, male and 56 perront of nonsmoking female controls have lived with a cigarette ::moklnn spou::e. tusocintion between passive smoking and lung cnneer drrivea biological plausibility from the chemical composition of oidestream smoke, the confirmetion of exposure in nonsmokerx with biolnt~icnl markera, and the failure to find a.threerhold for reapir.vtory carcinogenesis in active smokers. Sidcstretrm smoke c•ontnins the 227

i 228 INNOOR AIR ANU HUMAN HEALTH Fnnu• Loxir mld tinnOril'nnic afPntB ns mai0stl•enm Hmr`kel 9ome nrr In-rr.ruL in murh biphor cancentrnLions because of the burnin/runditinns mldrr vhich sidentrenm ::moke Ls f.enernted (2). .luvrsli/'nl.lour• wil.h nmrkrrs of tobnrco umoke expor.lrre Ilnv. 4 nnviuain/'ly d.•mnnfvtrxtod thnt.pnnsive smokinp reeults in iuhnlal.iun +md nb.^.nrplinn of uidonN'rmm smoke components ("5). F'or ' ~ IJ, k:.l.l /1. u1. (4) ror,•ntly r:•purlyd inurrnnr d uri mn .9 ro~ti muiur Irvrl:: In rxposed nonamokera and a done-responsr• rrlnt.i¢nubip Prl.wrrn urinnry concentratlon nnd the duration of rcporl.r.l nxpm'•un, lu Jalnnl, halsuklun nnd collenYues (5) found Lhnl Lhl I're::rm'r ol' :mlokrra In the homennd in the rorkpluce, and wbcu ro::idnnt'rr wrro nr.sociated with increased urinrlry co0ininrIrv,ln. Yin:llly, rl.udies of nct.ive smoking have uniformly indi:'ntrd rxcr:::~ lunf onnce•r risks nt lower levels o1' cigurettr ::u•ukinr rmd nonr huvr implied the Prencnre of a threshold (2). Tllin fnprr willrnview the epidrmiolop.icnl evidence relevnnll lo tbr hylolhrni:: Lhnlt pnssive smokinp causes lunp cancer. First, m.ILodnlry'io:rl olmcidrrutions relevant to studyinR thin nnsocinl.ion ut ilI br nddrrnr,J. ,.rcnnd, the nvnilnhle epidemiolopicnl evidencr wlll 1',' 1'.'Vl/'w"1. 1•llinlly. ttle exlntinQ. data will be IlYeeanCd :•rniucL convrnliomml criteria for determininp the caus:llity of .:.mn,intimi - 1.6•::nmr rriteria, in fnct, that were used in thr 1',t4 :'urp~ml Urnrrnl'e. keport for evalustinp fhe nssocintinn LvLwi luup ~an~•~r nnd nctive nmokinp (6). NVYrh011n1.uG I CAI. 'I'hr .i.+::nrinlinn Imtueen pnssive amnkinp nnd lunp. cnncer im:'. Iron nlprnuehrd wit.h ronventionsl hypothesis-testiny desipns: the o.e•-control and cohnrt studies (Tables 1. and 2). Kneh hns well rllnrncLrris~rd :IdvnntuPrs and dinudvnn Uq,cs (7). The res10 ts of' hofL mny be rll'frrted by misclassificntion of exposure and confnundinr by nl.hrr risk factors, whereas other types of bian oni4u,•ly Inl'lnrnra ennh design. The potential for information Linn, iclroduco d ly the interviewer or the subject, in of I•nrtirulnr impwrtance in cuse-control studies of this hypothesis. Ri:irlu::aification of exposure refers to the incorrect entryoYiv.'Itiml of +mtmllly exposed subjects as nonexposed and of nonrxpo::rd ns exlor.ed (R). When misclassificntion occurs rnndolnly In rrl:itiuu::bip to the selection of e study's subjects, it reduces mrn::ur'es ol' eiYert towards unity{ if nonrendom, it may increase or drrn ne:e rfl'ect neasures. The rluectinnnnire measures that have been employed in Inve:aipntions ruodurtr•d to date may have introduced random mir.c.lnssification on exposure to cigarette smoke. While gas phase crmpunonts mny also br important for cnrcinogenesis, the following dincusr.ion vill primnrily consider cigerette smoke psrticulate. In TAULk: I PASSIVE CIGARE T TE SMOKE hlvesliFaLiona ShowinP, Significant Kffecta of IrIo:live Smoking on Lunp Cnncor Nle.k :audy Prospective cohort study in Jnpnn ul' 91,440 nonsmokinpp women, 19(i6-1y'19 (19). Cnse-control study in Creece with 40 cases. 149 coutrols, 1Y]R- 1900 (22). Cnsc-control atudy in Lhr U.S.A. with 22 female nnd 8 malos cosos, 155 femnlo nnd 180 male controls (25). Fludlnrn Age-occupntion adjusted SOINs, by hunbsnds' smoking: Non-nmokern - 1.00 8r, or 1-1y/dny - 1.61 ;, 20/dny - 2.L1/5 , Odds rntiosby husbnndn' smoking: 11on-smokers - I.00 Pox-amokers - 1 .11 ~ 20/day - 2.4 >~ 21/dny - 3.4 Odds rotios by spoase smok I nf: Non-emokcrs - 1.00 1-40 pnck-yenrs - 1 .48 b 41 pnck-yenrs - 3.11 229 fnmml_nL Trend aUltinticnlly nipnlfiront.. All histolnpir:'.. Trcad vuiti::tirnfly ciC:li ficnrlt. Iii ^to- Inpi/•s otllrr' I.II:III ndrnocnrciuom:l :.ndtmmichiolnul v~~•Inr rercinonin, ^ipnificnut in- rrrnse for )~ 41 pnck years. Itronrbiolonlv:i•Inr cnrciumme errludb'd. I

230 INDOOR AIR AND HUMAN HEALTH TARLF 2 I_nvr;tip:itions Not Showing BiCnificant Rffectn flf Pn^cive Smoking On LunR Cancer Hir.k .^.LudY F'indinl'n Conunrnt I'ro::Prctivc cnhort Ap.e-udjusLed °Mlls, by All 6istolu/'.ics. stndy in thr U:'.A or husbnnds' smoking: 1'fb• 1 `q nryr::mnlinp I•ivJ•n, I•rr.li-I'il/ Nun-::muVrr:: - I.IN/ < 20/day >, 70/dny - 1.10 I'ror.pertlvr rnhnrtt study in Tcotlnnd of ttlRl a.nlt's nud frmnlr::, 1Y(y-1'!fV (1b) Y.nr.n-rnnt ru I ::I.udy in Ilnnt. Nony nl' i••I femnlr• rnsrrz nnd 1`i'1 rnntrnlr., 1971.-1o'r7 (2H,:'rl). 1'nr.e-rnntrui r•t.udy thr IL:.A. 25 mnir rmd h~ I'rm::lc rn,on w11Ir nl::trhrd coutrnJr., 1!t/1-14ft0 (74). Cnssn-rontrol sludy in Ilon/' kuuf wilh I'll nonmml frmrrlr e•r::rr:, 19tn-190, ('y0). A/'e-ndjusted SMIIs, Preliminary, smn71 fnr damestic exposure numbers Fbrlvs - 3•25 Pemiles - 1.00 Crude odds ratio nsso- All histologiea. cioLed wiUl smoking . two reports nre spouse of 0.75- inconsistent on ' the exposure variable. Odds rntios for current All histologies. exiwsure at home were Pindings neCativc 1.26 in males and 0.92 for apousu smokiuc in femnles. variable e:s well. Odds ratio for combined All histolopies. home and vorkpJnce exposure of 1.24 (P>0.40). ' PASSIVE CIGARETTE SMOKE 231 the Uniled titrrlrs, cigarette smoklnp is m ma•lor :murr,e of lndnar respirable purticulntos nnd thus a mnjor drterminont of vurintiorl umanp individunls in exposure to this pollutnnt (9-11). Within n renm, concontrnLions will be determined not unly by the slnmpth nf r.uurcrn, ::ucb nn ripnrattc nmoFinO, but by lruiidinp rlnirnrtrrir.l.ien und vanlilu(ian :rtr• (!)). Timr-nctivity Pel.lrrn:e I'nrtbrr mudily Lho prul'ilr nfroxpnr.urr (II). 9'hur., with rrpn-rd to drrm-air oxpo::urc, simpln doscriptinns of spnu::r smokinp brhnvior cnnnoll sutir•1'nctori]y define /trndients of rxposure. They ean, hnwr•vor, doc,ument thnt expooure to tahnrco smakn has necmrrd. I'imilnr limitationn apply to yuc:tionrvrire dorivrd indicrvrr of workflncr ixpn.rn•. Wllh r•-r.r.nI I.o Lnlul 1':rt.civ,• xPusiu'r lu lahn ,:.moIr vnrinblr:: Lhnt du not include llme mrt::idr of Lhe homr will lrnd to minclun:;ifir.ntion. In the populotion e:tudird by Friedmwr rl ni. (I), bi/'h prnportion:: of non.^,mokin/' mrOrtn'arld I'emnles••SrPnrled expo::ure outnide of the homo. Workplnee expu::urn vnn r:sncinlrA uith highrr urinary cotinine levr•ln in the recrnt r+MPort from .Inpnn by Mntr.ukurn et al. (5). Thus, nrndom misclnn;:il'irnl.inn of exposure in Iikeiy with questionnaire indicen. Studicn th:rt hnve used such men^ures mny be conservnlive since rnndom minclnssificutim: reduces effect mensures towrrd unity. 126V1kW OF TIIY. HVIDHNC6 Mvidence concerninp Passive smokinf and Lmg cancer hrrn bren soupbt indirertly in descriptiro data nnd directly with cnse-eontrul and cohort studies. Time-trends of lunp crmrrr mortnlity in nonsmokers have been exumined with the nrtionale thnl . increnaing passive smoking ^hould be mirrored by incfeor•inp mortnlity rutes. E.n::trom (12) celculute•d luny cancer mortnlily rutea from vnriou:: naLionwide sourcen for the period 1914-19GU und concluded thnt u rrol increase had occurred nmong. wiles after 1'/55. In contrnst, Gnrfinkel (13) did not identify time trunds in non::mokers in the Ilorn Study of Vntrmm., 10'i4 to 1969, or in lhe American Cnncer tluciety study, 1900 to 1972. In n lsrpr nuGpr.y serics, Auer'bach and colleagues (14) did not find incrrn::.d abnormnlitien in the bronchini rpithclium of mnle non^.mokrrn decrmsed in 1970-1977 in compnrison with those decensrd in 1955-19L0. While this review emphnsi.ees lung cnncer, asrocintionn of possive smoking with cancers of othrr sites or with other dirrnr.e:: would strenp•then Lbe• evidence cnncrrninp passive smokinp nnd funr. cnncer. An inve::tigntion of all cnncrr denths in femnlen rer.idinp in Wcste•rn Penn::ylvnnie has been frcqucnlly cited as showinp nn adverse effect of pnssive smokin0 (15). Ni11er intervirved surviving relatives of 537 deceeeed nonsmokinC women coneerninp the smoking hotits of their husbands. A siPnificnntly incrrnsed relative risk of crmcer denth wns found in the women who wore not employed.outside of their homes- Tho Jnrpc number of putrntioi

232 INDOOR AIP AND HUMAN HEALTH subjrcta lhul were not interviewed and the possibility of Inibrmut.inn binn detnwt from this report. Oillin ot e1. (IL) 1'ollowrd 16,171 herlltby Seott.i..h indiViduels, age9 45 to 64 yCArl', over nt )uest a 6 yrnr fcrtod. In u preliminary report conc.erninY. B,i2H subjects, nll-euuse mortality was comparable in non.^•moking mnleo vith rnld without domentle tobacco smoke exposure, but won increased by nenrly SU percent in exposed nonmmoking wnmen. A enve-eontrol Htudy of 4j8 cancer cnsee involving multiple siten nnd 471) comLLruln r.huwcd increased relative risks from oxposune durin" nhildl.und and durin/l ndulthood (17). In a 25-yesr cohorrt ntudy in Ameterdnm, nll r:mse mortulity in females was not affected by the bu:bnndn` smokinp stntus (18). . Mnre rnlrvnnt is. the direct hypothesis- testing evidenrn, provid,rd by rnn._cmltrolnnd cohort e;tudies. In IK31, two pnpuro: w,rc publi::hod vliich nyrorte:d nilptificnntly increasod ri:;kra of luup cnnr,r in uonr•mnkin/'.womeu whose husbnnds smoked cigorettee (Tnble1). Ilirviyulnu (I!/) couductod a pros4ective cohort study ol'')I,S40 notunnekin/- women in Jrqmn. Stmldnrdieed mortality ratios 1'or lung crnncer innrencod sipnificantly with the amount amoked by the hu::L:uldu. 1'he findinps were unchanged with control of potentially confounding vrriables nnd with extension of follow-up from 14 years to 16 yo:ms (20). Cverall, the relative risk from passive exposurr was t.b whercnc thot from aeUve smoking was 3.11. lliroyama hnn nlno rrlertcd : eifnlficnntly elevntcd relative risk (2.94) In uonsmokllu'p men vith smaking wives (21). Mollnvinh its publication, thin article received inten::ive ncrutiny nud rorrespnndcnce in the British Medicul Journal offend ooncernn nbnut stntiutie:al methodology, about population sclection, nFout uncontrolled confoundinf, by factors euch as cooking fuel oxposut'o nnd r•nrior•cotwndc status, and about the seemingly high relntivu riek. In his responses, Ilirnyama eatisfnctorily rebuffcd mn::t of these rritiricros; in particulxr, confounding did not apponr to ,•xpl:nin tbe findingn though active smoking by reportedly nnnsmnkiul'r wom.n cun not be excluded. In this repnrd, 1liruytmn (70) bn:•, rclmrted thot the findings after 16 years of follow-up are cunsist,nt with el'i'ects of.passive smoking on mortality from oniphyr.rmn +md rhronic bronchitis, nasal sinus cenacr, and iscbemic lu•nrt tli:lrnr.c. leiningicnlly, thene effects secro somewhat le:a pl:mt,ibl(, .th:,tl Innp esu)ccr and these new nssociations rniae conccrn ahuut roufoundiul• by unreported uctive amoking• Ilirnyuma hus .,xplnined the level of relative risk by the low percentages of . wnmrn working oul.::ide the home in Jnpmt, low divorcc rates, smnll iuum nis:rA, nud luck of inhibition about smoking in the pnaence of non::mokcrs (71). No datn concerning r'espirable pul'ticulate lovris in ILe sub,jects' bomen have been provided, however. Alr:o reported in 1981 were the results of a case-control study iu At.hrn;:, Crrvre (22) (Table 1). Female lung cancer cases with a din/•no::i:: other than ndenocnrcinoma or bronchioloalveolor iarcinomn I PASSIVE CIGARETTE SMOKE 233 were identified nt tlu•ee large hospitals and controls were selectrd at s Imnpital for orthnpedic disorders. All eubjects wrre interviewed by the nnme physician and their amoking atntue nnd that of their hnnbnnda Mna abtnined. S1nPle: women wet'e considered nn married to nonamokere and chnnf.es in maritnl atatun were considereel. The final nefies included 40 nnnamokinFg cases mld 149 nonsmoking controls. A significant trend of Increaoing risk with presumed extent o( ' paseive exposure wna present when edthrr tLe husband::' cun'ent or lifetime nmokiu(', hnbitv were uned fur ntrutificntion. Tbe I'indinhs were unchmV':d whesl the serdco wnn expanded to '17 cnsee; and 225 controls (23)• Less criticinm has been published conccrning the Creek study than concerning Iliraynma's inventip•ntion in Japan. Andiscunsed byKnbnt and Wynder (24), the ottempt to restriet }he case serirn to hintolo/ties nthe•r tlum ndenoanrcinomn nppenrs premnture nt prcnent., Further, tbc diagno::in of lung cnrmcr was mndc witllout histolo/•i'ol or cytological confirmntion in 35 percent of the rnne.^•. Noncomparnbility of the cuse and control series must also be considered when they are nscertnined at different institution:;l in this context, Trichopnulos et nl. did demonstrate compurability of the cane nnd controll series for key demoprnphie variables. The possibility of information bins must be raised because cane nnd controls were interviewed by s sin¢lc physician who may have been nware of the sLudy's hypotheses. Finully,. the investigntorn assessed the statistical significance of their findinfs with a chi-square for trend in proportions. The assumption that a former smoking husband provided an exposure intermediate between thnt of n nonsmoker and u current smoker wns not juatified by the outhnrc. Ilowever, the odds ratio is siF.nificantly elevated for the strntum with the bighcat level of current smoking. . The results of another case-control study, publiahed in 198'S, also demonstrated a significant association beteeen pansive r.mokinp, and lunp, cnncer rink (25) (Table I). Correa et nl. obtuinrd information nbout the smoking habits of the paxents nnd spouses of eight mnle nnd 22 female twnsmoking lung cancer cases nnd of 115 controls. Lung cancer risk increased with the spouses' lifetime cigarette consumptinn. Maternsi smoking was associated with a significantly inereased odds ratio in active smokers but not in nonnmokcrs. lln stratification by nex, the incren::e was atatistically nirnificunt only in males. The relatively smnll numbers of subjects in this inve::tipntion mendnte caution in Interpreting its resertts. Ilowever, the overnll findings Were unchnnYed, as reported in n recent abstract, whon these data were combined with comparable information from two othercane-control studies (26). Thc overall desiKn wa a appropriate bnt information bins may affect the results of case-control studics that rely on interview for exposure information. The. exposure variable, cumulntive cigarette consumption, differa frnm the

234 INDOOR AIR AND HUMAN (IEALiH mrvis,nr:g used by Ilirnynmre (19) and by Trichopoulos et al. (7: ). 11, vnuld be useful to rermnlyze these dntu with compureble exponuro vuriables. TI, e ranultr, ol' two other invest]gotiona have also bcI n intcrprotrd os r.hoaiuf, an increnard lung cancer risk ussocintiA rith pessive smukinf. In Cermany, Knoth et a1. (27) nccumuluted n nrries of 792 lunr, ennoer cases of which 59 were in 1'mmalen, Thirty-nine of t.lu.^.r va:ncu find nat nnoked but 24 of Lho nonnnmkerr. had livod in hou:,rbnld:: with smokera. Because tbu invustiCUlor:; did unt intorview n cmetrol series, they relied on crncv:: ntutintico to er.limate tLe nn0iciputed propor0ion of anmkinp, I:pollRea in Lhe frnrral populntion. In the nge group 50 to 69 ycur:: correspmulinf, to the husbands of most patients, the census nbowed only 22.4 perernt rurrently smoking. In another recent report, Cillis et al. (I/,) described tlre itsults of a cohort study of Ib, 171 muleo and females in Western Scotland (Table 2). Exposure to tobacco smoke iu the environment was chn nmterized by four strvite: nonsmoker and not dumestically exposed, nonomoker and domesticnll,y e.posed, smok(:r nnd not domestically cxposed, and both a smoker uod domastically r:xfosed. F]ortnlity rates for lung cancer nnd for nll olhrr crmcers were culculuted sepevutely for mules nnd 1'emolerI vithin -ae b ctrnlum. AmonP mnlc::, :;iz luug canccr drolLa wcr. observed in nunsmokvro; in the control stratum, the nnnuui mortality rate was 4 per 100,000 whereas in the domestically exposed nonsmokrrs the rate was 1} per 100,000. For males tho rates were similar in the two actively smoking groups. In females, with a totnl of' eiflit deaths from.lung cancer in nonsmokers, fho vn-riation of mortality rates did not sugP.est an adverse effect of domestic tobacco amoke erposure. . The metlmdological limitations of these two studies ru'e evident; neither formally teets for nssocintion between lung cnncrr risk and ras::ive smoke exposure. The Cerman report did not involvo n comparison serion rmd the appropriateness of substituting censu:: data was not nddFesr.ed (27). The authors did not formally test for nssocintiem botwien passive smoking and lung cancer; in fact, tlrey used their :;pur::e data as a platform for discussing social and pulilicnl u:q,ects of passive smoking. Interpretation of th, ::cottish investigation is constrained by the small number'of denthr; in title rrgard, slntistical significance testing was not performed (16). The lack of effect of domestic tobacco smok', exposure in femnles is not consistent with earlier reports ('1'abL• 1) but the number of denths is quite small at present. The renulta of four other investientions suggest lesser or nn. 'effeml.a of passive tobacco smoke exposure (Table 2). Chnn et nl. (20,2y) ferfarmed a cnse-control study in Hong Kong that included 114 ucnnmokinP fimale cases with 139 controls. Appnrently a sirytlo question was asked concerning pnssive smoking exposure. In a 1979 r'rpmt, H'r invostig:rtors stnted tlwt 40 percent of cases and 41 ! PASSIVE CIGARETTE SMOKE 235 percent of controls (rstimated odds ratio of 0.75) replird o/7iematively to a que:;tion concerning expo:;ure at imme or at wark (28). In n 1982 publication; similar find]ngs were reported but the expor,ure vurinble wns described ss related to spouse smokinr (29). The conflicting description of this lnvestigntion'n expomue vnriable requires r.lnrification. Noncompnrability of the cnr,e und cuntrol series with regard to place of residence nnd lark of bisLolop,irol or cytolopicnl confirmation in IR percent of uu:e:: furthur limit this Investigation. A more- recent en::e-control study from Ileny Kon/g also did nnt show definite effects of passive smoking. Koo ot nl. (30) interviewed 200 cases eacertnined through Hong Kong hen]th facilities and 200 controls, selected from the yeneral population to mnlch the nPC, socioeconomic, and gcoPrnphie'4tstribution"of the cases (7bble 2). Mith eomeu, not exposed to smoke at home or ot wnrk nrs the' refercnce r,ntegory, odds rution for expnsure nt homr and at work were not sil.nificantly incrcused. Iloesmoking cnsen Imd fewer hours ol' total estimated exposure than controls. In contemst to the case-conlrol study in Louisiana (2'A, nn effect of mnternul smoking was not found. The mont important of the four publicntione:, construed by mnny sa neCntive, is bnced on the Americnn Camcer ttoclety's prnspertivr cohort study (13) (Table 2). Betveen 1959 and 1960, 375,000 femnlrnonsmokern were enrolled and follow-up of mortality insted throuyh 1972. prnm this cohort, Carfinkel identified 17G.739.nonamokrrs wliose husbands had never smoked or were current smokers, ]rresumnbly on enrollment. The standardized mortality ratios for the women with smoking husbands were frenter than unity but not sihnificnntly.. In the smoking-exposed group, there wnsnoevid(urv for e dose-renponse relntionship. AA separate matched analysis, performed to more completely control confoundinP., provided simiLv results. 1'he Americem Cancer Society study should nnt be rharacterizeA as contradictory to tbe findings of Ilirnymms (ly), Trichopouloo et nl. (22), and Correa et al. (25). Mirat, the standm'dizad mortnlity rnlios nre above unity for tbe exposed rroups. :.ecand, confidence intervals for the mortality ratios in the Amrricnn Cancer. Society study overlap those reported by Ilirayamb (1'))• Third, while each of these investigations employed spouse smokinr as the exposurn variable, the comparability of dose amonf the four in uncertain. kepoce (31) has suggested that the mortelity ratio in the American Cancer Society study bns been reduced by misclassification introduced by workplace exposures. ilis argnmruts ~ land to an adjusted mortality ratio of 1.7 for the Americxn Conrrr I Society cohort. Finally, theune of death certificotes to establish diagnosis in the Americnn Cancer Society study probably introduced nisclnssification of disease status.

27U INOOOR AIR AND LIUMAN HEALTH IL,rrut nud pre•Ilmirmry rer.ulto from a nntianwide cnne-cnnlrol ::tudy nL•m did not drmonntrnta iucreased lung r';umer' ri::k fram donu•al.ic oxpu:me to tobacco smoke (24) (Table 2). Knhat nnd Wynder uxr,mined the effects of currently omoking family members :Lnd ul'~ u'rrnt r•xFn:urre nt. work in 25 nonnmukinP. malc ond 5'S nonumokinp femnlr rn::es with eipunl numbers of controls. For men, the odds rntio for worEplnce exposure of 2.6 wos significantly incrensrd. Curreul 4omr•stie vxposure was not sipnSficant for ILL or fem:ain, Ia n r.nuiller ::uhset of cuses, adverse effects of spouse smokinp wrrr nntt idcnlil'Ird. The nuthors clearly stated thnt Uwir runult.n weru preliminnry and that more dnta are needed. While the numbrrn un• cumll, Lh, y nri• rquivaleot to those in tbe scries rreported by Correu et nl. (2i). . t1UNCLUSIUNS In summrn'y, at present, only nine published investigations provide datu dirrctly relevant to the hypothesis that passive smokiru- in n riak factor for lung cancer. Several others offer indirect rvidcneo. Thin paucity of data contrasts sharply with the literuture cited in the 1964 L'urgeon Cenernl's Report whiuh chnrnct.•riwed nctive cignrette smokinf as a cause of lunt• enncor (4). That nport reviewed 29 cnse-c.ontrol and seven cohort atudirs. Their results uniformly and unequivocally demonstrated the nnaociation between active smoking and lung cancer. Applicntima ol' carefully considered criteria for causality to the cvidcnce led to the designation of cigarette smoking as causally relritod to luna cancer in men. The association wus ,judged on its conoistrncy, ntrength, specificity, temporal relationahip ond ~nhr:nnc,:. rapoit did not explicitly define . Lut iudivuled thnt lhe tern is generally applied to "..: a significnnt effrctual relationship between an agent and an associated disordrr ar disruse in the hoe.t". It also acknowledged the multifuctorinl e tiolopy of luu/- carmer and did riot require a unique reluLionuhip betwcen nnwkinf und mulif.nuncy. Apid icntion of these same criteria to the data for passive snw0ing hif.hlil•hts their xeaknesses. with regard to consintency, the roN Ilcts amonY, the published investieationo are immedintrly rvident (Tnblcn I and 2). Ilowever, becsuse of potenti;il differrnres in dose amonPp the investigationsl it is not certa in thnt ruch hn:: tested for a common magnitude of effect. Furthermore, p.iven the small numbers of cases in most of the p:q•er:;, the point estimates of effect are unstable and confidence limits r.enerally overlap from one study to another. In the pnnitive sluditc, tbe relative risk estimates have indicated rrlntivcly mod,st rffect levels, ranging from about two to threc. 'I'becv vulues ure murh lower than those associated with activc i+mukina• rnrd could mare readily be the consequence of bias. In tbe face of omn11 nnd cunfliatine ntudies, unidentified sources of birns PASSIVE CIGARETTE SMOKE 237 nhould not be rendily dismissed ao nn explnmstion for eiPnificnnt tvrt mudcat rlevationn of risk. Specificity of asnn.:in0ion, tlmt in a unique relntionship between the factor and the disenee, i:l nn Irrerlevunt and unimportnnt criterion for posnive emokinP. kith regard to the temporal nnnocietion of puneave smoklnP• and Leny cancor, thc directianulity is unquentiomrbly npproprintel .rxpururrc precedea Lhe development of the dineuee. The remrrining eritrrion ia ttha cohelronce of the ase.ociation. The biolog/rn] pluunibilil.y of tbn aunocintion between paseive smokinr uud Inng cnncer bns b.•rn previously reviewed and thia criterion appein-s to be met. In cmmlusion, tire association between posnive amokinf und ]nne cnncer dorrs riot yet mreet criteriu npplied to nctive amokinp In Lho 1964 Surccon Cr•nnrnl's Iteport. While conf'irmation of pooe:ive smakinP. ns a risk factor for lung cancer would.df[er new onmunition ngainnt tobncco, the nvuilabie evidence does oot permit definitive ,;udgmenta. ln the fnce of difficult methodological problems, prn-t.iculnrlytlmtof nncurataly quantifyinR dose, unimponchnblc data vill be diff/cult to obtain. New approaches for studying pnssive smking and lunP rnnr'er are clcony ncrvled. The problemn of dosn estimation seem more difficult for IunP, cancer Omn for other putntive health efl'rctn of pnnsive amokin/l. Tbe relrvant exposures may kePin ut birtb nnd occur under a wide variety of circumstnnces. Ilistoricnl reconstruction of exposures by questionnnire mny be the only uvnilnble approach for epidemiologicni studies. Ilowever, further validation of the quentionnaire appronch is needed vith compnrironn against bioingicul mnrkers and measured concentrations of tobnero smoke component.^.. The reliability of questlonnnire ussessmrnt of passive smoke exposure has not been entnblished nor buve sonrcra: of . binn be.en evriluatnd. Intorviewe vith nrxt-o.f-kim mis y be particularly prone to information biar., almost certainly in the directionof ovrrreporting. In fact, as the public becomea ' increasingly uwnre of and sensitised to potential effects of pannivu smokinC, the results of r.use-coutr'ol ntudieo will brcomo ine•,reosingly difficult to interpret. Unfortunately, the cnae-controi design is the most efficient approach for investiCntinR the relatively small number of lunQ cancer cose^ in nonsmokers. Cnhort studies, which might offer better expasm'e dnln, must involvo lnrge numbers of subjects und lenCthy follow-up. Investigntive opproaches which examine outcomes other tbnn Iunr cancer might provide moreimmediate answers concerning pe:nive amoking and respiratory tract cnrcinoBenasia. For example, apaluni cytoloCy miP.ht lie evaluated in nonsmokers in relation to pnuniva tobacco smoke exposure. . While additionnl investiPStions will certainly be performrd, the available data mny already be satisfactory for both reCulntion and prevention. For regulatory purposes, the established cnrcino- genicity of tobacco smoke and the hif.h prevulencr of rxpo^ure.

238 INOOOR AIR AND HUMAN HEALTH nhm,ld b- xuf/'ielent to prompt action. For prevention, the datn on .ulive smnkin/, shouJd Le• sufficient; smoking prevr•ution :,nd reu:¢,tion nannin t.he bent strntegies for minimizing pas::lve ,.rpu::urr. ACK IIUk' L11UU FAI I':N'I':t tlupported in part by a grant from the Nationnl COnrur histi lutr- CA P71b7. Dr. tiumet is recipient of s Neaenrch Cur, rr Urvelopmrnt Awnrd 5 KO 4 IIIA0951. The.author thanks Dr. A. Jud;.on W,IIn f'nr his he1pIL1 comments and Lee Fc•rnnndo for preprnrLt;, tim m;nmr.rript. . ' . IiIfP4'ht:IIC1:1 1. Fl'io-dman, C. U., Petitti, U. P., and Bevol, R. D. "Ih'evalence and Corrrlutes of Peasive Smoking," Am. J. Public Ilealth 73:401-40!, (1963). 2. U.S..Publie Health Service. "The Health Conaequences of SnioY.inf• Canecr. A Report of tbe Surgeon Cmeeral," (Nockvillr, lihryland: U.S. Department of Health and Iluman Servicrs; I'ublic Health Service, 1982). 3. Jorvis, M. J., and Russell, M. A. 11. "Measurement and Estimwtion of f.moke Doeare to Non-smokers from Environmental Tobncco Smoke." Lur. J. Respir. Dis. 65 (Supplement 133):68-75 (19k4). 4. Wnld, 11. J., Borehum, J., Bailey, A., Ritchie, C., Ilnddor, J. F., and Knight, C. "Urinary Cotinine as Marker or Breothing Other 1'eople's Tobacco Smoke (letter)," Lancet . 1:2:0-231 (1984). . ,. Mnt:mkurn, S.. Tominato, T., and Kitenoq N., et al. "F.ffectn of F:nvironmvnlnl Tobacco Smoke in Urinary Cotinine F:xeretion in flonsmoke•rs. Evidence for Passive Smoking," N. EnKl. J.Mrd., J11:82U-tf32 (1984). - ti. U.S. Public Ilealf.h Service. "Smoking and health. Report of' the Advisory Committee to the Surgeon 6enernl of the Public IIca11.h Service," (Washington, DC: U.S. Depsrtment of Ilealth, Educ,tion, and k'elfare, Public Health Service, Center for Disease Control, PUS Publication No. 1103, 1964). 7. hlacMehon, B., nnd Pugh, T. F. "Epidemiology• Principles and Eethods," (Boston: Little, Brown, and Company, 1970). 8. 9. 10. 11. 12. 13. 14. 15. 16. Wy.~ 17. N 0 ~ F+ 1U. IId N ~z Iw+ 19. PASSIVE CIGARETTE SMOKE 239 Kleinbnum, I). C., gupper, L. L., and Morgenstern, 11. "Epidc•mioloRic Besourch. 1'rinciples und puemtitntive Metbods," (Delmont, Cnlifornia: Lil'etime I•enrninp I'ublirntionn, 19M7). National Nenenrch Council. "Cnmmittee or, Indoor Pollutnnt::. lnduor I'ollutemta," (WashinCton. DC: Nntimml AondemY 1981). SpenCler, J. D., 1lockery, p. W., Turner, W. A., Wolfson, .I. M., and Ferris, 13. C., Jr. "l.ong-term Mennuremrvnls al' Nesplrnble Sulfntrs and Particles Innide nnd outslde Ilnmer.," Atmou. NnvL'on. 15:23-30 (1901). rpenfler, J. P., nnd 3oczek, M. L. "tlvi4,bce for 1mpFnved AmLient Air' thwlity and the Need fur IR•rsmml Erpoaure lieneurch," Environ. Sci. Technol. IL3:268A-280A (1984). Enstrom, J. P.. "RisinP. Lung Cancer Mortnlity Among Nonsmokers," JNCI 62:755-60 (1979). Cnrfinkel, L. "Time Trends in Lung Cancer Mortality AmonP Nonsmokrra and n Note on Passive Smoking," JNCI 66:1061-In1.6 (19U1). Auerbocb, 0., Iinmmond, F.. C., and Cnrfinke.l, L. "Chnn/,en in Bronchial Epithelium in Relation to Cigarette Smoking, 1955-196o vs. 1970-1977." N. Enpl. J. Med. 300:3t11-3H6 (1979). Miller, C. II. "Cancer, Passive Smoking and Nonemployed and kmployud Wiven," West J. Med. 140:632-635 (1984). Cillis, C. It., liole, D. J., Hawthorne, V. M., and Boyle, P. "The Effect of Environmental Tobacco Smoke in Tro Urban Communities in the West of Scotland," F.ur. J. gesp..Uis. 65 (Supplement No. 133):121-126 (1984). Sandler, D., Wilcox. A., and Everson, N. "Cumulative I'asvive Exposure to Cigarette Smoke and Cancer itisk (abstrect)," Am. J. Epidemiol. 120:482 (1984). . Vandenbroucke, J. P., Verheeaen, J. I1. H., DeUruin, A., Mauritn, B. J., van der Ileide-Wessel, C. and van der Ile.idee, N. M. Active and Passive Smoking in Murried Couples: Ho::ults of 25 Year Followup," Dr. Med. J. 288:1801-802 (1984). Hireynme, T. "Non-smoking Wives of Heavy Smokers Ilnve a Higher Risk of Lung Cancer: A Study From Jnpan," Br. Med. J- 202:183-185 (1981). .

240 INDOOR AIR AND HUMAN HEALTH Nlrnvmml, 9'. "1'nusive Cmokinp and LunC Cancer," Presented at the Fifth korld Cnnpreen on SmokinP, and Ileulth, Vinnepcr, Cnn::dn. Jnly, 1483. 21. Ilirnymm~i, T. "Non-nmokin¢ Nives of Heavy Smokers Have a lil,h, r Lirk nl' Luup Cm,cer (lrttcr)," Rr. Med. J. Trfhopoulnn, p., gnlnndidi, A., Cpurroa, L., nnd MncMnnnn, 0. "Lunf C+,nr1r runl lb,nsive Smokinr," Int. J. Canr•er 27:1-4 (14P.1). .. Tricbnpnnlr:^, IA., Knl+mdidi, A., nnd Sparron, L. "l.ung Cm,rrr and I'rr.:sive Snaking: Conclunion of Greek Study (lotl.er)," lnncct 1:6'I7-67N (19H3), 24. Knbnt, C. C•, :md Wynder, E. L. "Lung Cancer in Nonsmok,•rs," Cnnc,•r 53:1214-1221 (1914). 2}. Corrcn, 1'., 1'ickle, L. Y., Fonlhnm, E., Lin, Y., end Nnenazel, k. "Prn:;;ivr :lmaking and LunK Cancer," Lancet 2:i9S-i97 26. Dnlnpr•r, N., 1'ickle, L., Mason, T., and Ziegler, R. "l'nnaivc SemkinC and LuuC Cancer (ebstract)," An. J. Rpidemiol. 120:482 (IITA).. 2'/• hlwth, A., I'ot:n, 11., and Schmidt, F. "Passivraucteen als Lun/l,nP.rebr.ursuche bei Nichtraucherinnen," Med. Klin. '/8:66-69 (19R3)• 2b. Chnn, Y, C., Colbourne, M. J., Fung, S. C., and Ilo, II. C. "14'onchinl Cancer in Hong KonY 1976-1977," Br. J. Cancer 39:1112-192 (1979). 29. Chnn, W. C., and Fune, S. C. "Lung Cancer in Non-smokers in Ilnnt; Kung," In Grundmann E, ed, Cancer Campaign, Vol 6, Cancer k:pidemiolory• Stuttgart: Gustav Fischer Verlag, 199-202 (1982). 30. Koo, L. C., 110, J. H-C, and Saw, D. "le Passive Smoking an Added Pisk Fnrtor for Luna Cnncer in Chinese Women?," In Press. J. Hxp. Ciin. Cancer Res. ~I. Repnrc, J. L. "Consisteney of Research Data on Paasive SmokinC and Lung Cancer (letter)," Lancet 1:506 (1984). 18. Critical Review of the Relationship Between Passive Exposure to Cigarette Smoke and Cardiopulmonary Disease tlillicent lllggins, M.N, ' llniversity of Micirll:un Schnnl ef Public Hc•alkb P T619 review of the relatloncldp between passive cepnsurc to cigarutte smnke and cardlupulmonary disease will be I/mlted to non-mnllgnant conditions and:restr/cted to evidence fr,im epldeminiop•Ical and clinical studies. Several comprehenslve rerplrab~rv rcvlcws of the literature on passive smoking and die:uases have bcen published recently (I-7) and reports Irom sevurnl workshops are also available (4-5). 1 wilt descrlbv a few studies tu Illustrate the kind and quality of evidence thnt is avaliable, :u,d glvc an overview and evaluation of the resnlts of published studtcs. The cardiopulmunary conditions mrd diseases vh/ch bave hcen associated with passive smoking In some studies are shown In Table l. Table I. Cardiopulmonary Diseases and Conditions Assorlated witl, 1'aesive Smoking 1'ntbnphyaiologic condlttons carhoxyhemoglobinincreased . heart rate increased blood pressure increased platelet function decreased exercise capacity reduced pulmonnry tunetlon: acute changes persistent redur.tton Horbldity . respiratory symptoms respiratory infections and Illnesses acute and cbronic bronchitfs pneu,nonia antbm.i T,lginal Pate Mortality chronic bronchitis and emplrysema . Ischemic.beart dlsease lung cancur 241

INDOOR AIR AND ' HUMAN HEALTH Edited by RICHARD B. GAMMAGE STEPHEN V. KAYE Health and Safety Research Division Oak Ridge National Laboratory Oak Ridge, Tennessee Technical Editor VIVIAN A. JACOBS Oak Ridge National Laboratory Oak Ridge, Tennessee rI FS /' LEWIS PUBLISHERS, INC. 121 S. MAIN STREET, PA DRAWER 519. CHELSEA. MI 48118