Tobacco Documents Online

Marriage to a Smoker and Lung Cancer Risk 11 MD. \1-S, .kl~-1) D<3hOTHY R. 11.aTH.i1<, }'HD, MS ~Ds;raE.: :~s ., ~l ~ ;~~ uisr.~~r,•h:acd :~sc-~untrui stud~ uf ;ar.c<r ::-. Nc~µ wc `avc collcctcd Jjtj on s(ou.cs' :u^uc.e sre~,,-~ .`:a"rts a.^d on•t.".c•;ob czpvsurc to ashcs:os. T".e - ;;c±cat anaJ~.cs jnciu,_1e tx`y .ascs and "51 cOntrnls wit!s l:noun ;acs:sc and ;c:Sonal smoi,;.^k status, 3f whom :J w.crc I,Ccfong ~^s^ri;crs ..a`. :uns :aaccr clc 10 <'cct 0 ! aPousc c,%:arc,tc ~i:^.~was:cur.u ~^4, rcntorfo^nc-s.mokr,s.oc~c: smuticrs mamcd !u sm_ okcrs had about a t>.v-fuW incrcasck! ns1k uf lung cancer. Lung canccr nsk in never smokers also incrcascd wrth ~!uration of exposure to a smoking spouse, but not with increasing numt?cr oEcikarcttcs smoked pcr day by the spouse. Our fintlings arc consistent with `rc.,vus reFuns of elevateJ risk for lung cancer among never smokers living wrth a spouse Who smokes cigarettes. (.drrt J PuoliF Neairh 1987: 17:5911--t>n2.) T~e c1usJ1 asuf acti>•E cip~_,rcttc smokint; with 'un= cancsr has been acccrtcd for man~- scars.'' !~cccnt _ . . - -,tc evide:tcc :r,dicatcs that in.oluntary exposure c^ids^niclc•c of nonsmokers to :obacco smoke is also asso>iatcd with lung car.;cr.~-" NonsmoV;crs, as wcil as activc ;is;arcttc smokcrs. .^:^alc cn%ir-onmc:ital cobacco smoi:c, which consists of a combtr.atton ot' sidcstrcam. smoke and cxhalcd mainstream smoi;c. The putative _association of cnvtronrrtcntal tobacco s:;:oke wtt`a :Unr, car•cc: dcnves hiological plausibiiity from t`:e :ac: of a dcmonstratcd ;:rEthoid for lung cancer in active ~:;;0 , Kc: s, ;-rom the quaitta:i% c sir.:iiaritics of rnainstrcam and 5 ~cs::c2ni s,~.oi::. and fror„ (!;E F.rr.,cn,:c of mutas;cns in thc ^:-,c ot , zsstvs sr:vi~:rs. o T':c a_sseciation of in,-clunta.rti cxr.osurc to tobacco :ur.g canc<r ~as no~v becn :ramincd in studies :n Japan. ~ccc . 1-:cn~ !:~ir,g. Scotland. Gc na- n~ . ::ac.' :.`.F l.'n:;cd S cs`. ' ~c;c slud:cs g c n c r a ilv ,ndicl;e an in s-ascC ;n nonsmoil' c:s. Studies from JaFan• C>recce, and •hc i'n::r,: Sta:cs ha.c shown elevated nsk Est;:natcs zssixtatctf -ith the c.-tpLsurc of nonsmokers to ;':c:: sFouses' smoktne..', Ir,creased nsks have not been ;ound in all in~es:,gattons, al:`•,ot:t;h estimates of ctTcct from :hose :cpor;s with neca;ivc hnc!;ngs are genct,ailv consistent with t~`:ose from rerorzs showinc <lEvatcd risks.'t'`° In ;9S0 %vc bcgan collectiny +rata in a population-based :asc•c.n:;oi stua\ .'esi~ncd to cxplain difTering lung cancer .1;,u.;cncc in Hts,.antc and non-Hispanic Whites in Ncsv Mc.Ftca.`• The onkinal study .;ucstionnairc inc.ludcd qucs- - ~`rom saouse smoking and Cancer lnstitutc.'a An initial case series was stlccted from patients with cancer incidcnt between January t, 1980 and December 3 1. 1982, For this initial series all cases less than 10 years of age and all Hispanics were included: non- Hispanics age SO or older were sampled randomly to select 40 per cent of the males and 50 per cent of the females. To increase the size of the female non-Hispanic subgroup and Hispanics of both sexes, we selected additional cases: all patients in these groups with cancer incident between De- cember 1, 1983 and November 30, 1984. Of the 724 eligible cases selccted for the study, interviews were completed with 6-11, or 53.5 per ccnt. Of the interviews with cases, 305 werc completed with the cases themselves and 336 were with surrocatcs, generally either the surviving spouse or a child. For the cases in nonsrnokers. the histopathological type of lung cancer was classified by panel review of histopatho-l logical material (N = 17) or by information in the New Mexico Tumor Registry case abstract (N ='_8). The panFl, Which included two pathologists, determined the histopath- ological - ological type on the basis of conventional light microscopy and used a modification of the World Health Organization classification.19.10 Control 5e1ectfon Potential controls were ascertained by two methods. RcsidEnces, identified from lists of randomly generated telephone numbers, were called and a household census was taken from the person who answered. Telephone sampling identified 2,018 potentially eligible households. of which '_$ i ( 14.2 per cent) refused to cooperate with the census. As this technique was not efficient for selecting older controls. an ncrsons were chosen from a list of randoml,v ~o~,EZZ9zo~

smckc. i'hc :~u:at:~c asioc:ac:~\n e1 cn~•tr~~n:ttcn_ , 1~~. . ~ sn,okc µ ti ^~ cancer dc^\•cs taiologicai plau Ility from c( a.+.cc:a;lst:ztc~ t`:eshoi.~ ;ei lun. E3nlCr in acti\ < t;hF yUalttl:t\c str^i:aritics of mainstr-am af~~ ~, cam. an~'ron; t`:e f-rcWcr: of mutagcns in'!;c ^nc ot=pass;\ c T';c ass.•clatton cf sn~ciunt:_,n E.trosurE to tobacco <-;ckc :unC canCc.' has now bcen ;:(arnincd in studics ;,r,Juc:cd :n Ja,^.an• v cr.,. Hong Kong. Scotland. Ge ma- n~ aad 'hc ~".,.••c:. ~ S: a-s.' Thcsc st::,?:es kcncrailv indicate _ _ y an :nar-,.cd "sk :n nons„tokers. f;,:'I:es from Japan. i ;cccc. a. -. c stztcs t;a\e n rlcvatcd nsh ~ . ,,. cs .aa:cs 3!~:,,c ilcd •~ •;hE ExF.c.s,:rs of nonsr.to}.crs :o :':ctr sF~.~u-scs` ;-tok:n€ Increased nsks hai•c not hECn ; aa ad ;n ai? ,a• cc 1s,;;ons. il:' ou> h s^;a:cs of etrect fr0rn t;;osc .`cYor:s ••4-1,ti ^c:a-:S•c ^ndtnCs arc.cc.^.crailti• clJnsist6nt w.t:^. :!;osc ;;o^t :cpor:s shoµing elc:a;cd rsi~s.`l :e In !9S(? •.~xe `'ckan colle1tinc data in a Popul2tion-bascd 4e,ier•.c:d to cxplain .1irCnn,; lunC ~arlcCr - ,__ _ _.. . _ cw k.u~cn~c ;~ i;tsr tnt. and non liispanl; 14'hitc, :n i'l ~t.xlca. The ortkinal sct d> y res::onnairc included qucs• ::ons on :ol•ac.'o smoke e.Crv`~.:rc Crom spouse smoking and ;-1n int;ircct c.Yposu.'- e-e to asbestos Ihrt?uch a spouse's;ob. This -cr.or•-, de<cr:hts the nsks assoc:a;cu ~%it;t these csposures in ., ^,sn:okcrs ,,. Ve\~ \lcs:co. Cue Sctc.:/~n ^.F cascs µcre Hi:pnnlc and nen•'riis,unic residcnts of ~ z-•r MC_ ~2. ;ess .han S5 \cars of age .'.( a,:;c1C1v_ sis of t^rntarv -t: C:a^ccr. L:ascs %scrc dscc,'lllned by the \cw \lc.ltt;c? of :!te Sur>•c:llancc. EplJra:iol• a.^ En.7 ~'itit:ftt 1 rotram of the Vat:on;t? :,)c t'c_ Cirtmenlr,~: \!c,h.-~r~c.>nJ Jr:J InJ :.`.c !_(vcrJcp.,_n_ mcn:jl of \c.•\tc}ICO \tcol:jl lcnlir. -1 .,.:c:c r":uc on.l:^.an \f. S.Imc(. 11t1. Nc,. `(v.' \tCJ1,Jl 1 :nIC.' `w11 ,'. :n,) .tc \\f 57;.:1 ~h- . .. . ,.1> r`aCc^. ~a,^Jnll:cJ ;u . .. ~1: :Q., l^•: "':[^lcJ I .r \ 1'C.'Pt'l'r ~ ccrnoc: 1• 1: .-.•_ - cases sc!ecccd for the study t'rttcrvic%vs were complctcu wttrn ba l, or 88.5 per ccnt. Of thc ltcrvicws with cases. 305 wcrc completed ~~,ith the cases themselves and 336 were mvith surrogates. generally either the surviving spouse or a c.hiid. For the cases in nonsmokers, the histopathological type of lung cancer was classified by panel review of histopatho- logical rnatcrtal (Tr = 17) or by information in the 4cw Me.rico Tumor Registry case abstract (~' ='_S). The panel. which included two patholo£ists, determined the histopath- 0iocical type on the basis of conventional light microscopy and used a mo.iitication of the World Health Organization c!assiftcation.19.'0 Cantrot Sel"tinn Potential controls wcrc asccrtained bv two methods. Residences. idcntiFed from lists of randomly generated telephone numbers, %,.•cre called and a household census was taken from the person who answered. Telephone sumplins identified 2,018 potentially eligible huuscholds, of which '_87 (1s.2 per cent) refused to cooperate with the census. As this technique was not efficient for selecting older controls, ar. additional ?52 persons were chosen from a list ot randomly selected New Mexico residents. 65 years and older. \,+ho \ver-e on the Health Care Financing Administration's roster of Nicdicarc- partici, ant-s. The control group •vas frequency- matched to the cases for sex. ethnicitv, and Il)-ycar atc category at a ratlo of approximately 1.2 controls per case. Ui the 9-t-1 controls sclectcd for this study. ~Su tS3.l per cent) werc intcr vieN-ed. tnteryicw U3ti Collection intcrvie~• The intcrvicves were conducted by bilingual ers. Rest?ondcnts were askea to describe the smoking habits of all spou;rs ofthr index subjcct. For each smoking spOusc. duration of usc and avcrage amount smoked daily w,<;:: rcconicd for c:s:arcttes• clcars. and pipes. Respondents not asked to describe exposures to tobacco smoke at work ;; in uther atu:,Uon, uutade of the home. All jobs held by :: spouse tor on< year or more also %aerc rccorded. as %~•crr: repctrt,s of :oouses on-the-job exposures to arsenic. asbes- tu;• Ic•.7. pcstlcu!c,. and radiatwn. We hypothesized a pnltri that :,ibe~ttt, et,^.o.urc mwht increase lung cancer risk and a,;PH May :?87. Vot. 77• No 5

added the othcr agents to reduce thc emphasis on asbestos and to test for informatiori bias. A dctailed history of personal cigarette use was collectcd from subjects who had smoked for six months or more. Cilculation o! Passivr Etposure Indices Measures of passive exposurc to tobacco smoke and to asbestos were created by summarizing the information pro- vided for each spouse. For tobacco smoke, categorical and continuous measures of exposure were calculated. We des- ignated as "exposed" subjects cvcr married to a spouse who smoked cigarettcs, rcgardless of the spouse's use of pipes or cigars. To examine the ctTccts of cigarette smokc alone, subjects whose spouses had smoked other tobacco products were excluded from some analyses. We created two indicator variables for these exposuress one for all forms of tobacco smoke, and the other for cigarette srnokc alone. We also calculated the duration of exposure to a cigarcttc-smoking spouse and the average number of cigarettes smoked daily bv the spouse(s). If complete data w,ere unavailable for all marTiagc paFtners, these variables were set to unknown. Two categorical variables were created to describe potential indirect exposure to asbestos through a spouse's job. Spouse's job historics were reviewed against a list ofjobs judged a priori as possibly involving exposure to asbestos: asbestos mining, textile manufacturing, auto brake repair, cement or construction work, pipe fitting or covering, insu- lation work, and shipyard work. If one or more jobs held by the spouse appeared on the list, the index subject was claasified as exposed. Similarly, if a spouse was described as exposed at work to asbestos the index subject was considered to be exposed Data Aculrsis For these analyses, cigarette_ smokers were those indi- viduals who had smoked att least six months. Current smok- - - ers were those still smoking at interview or who had stopped within the previous 18 months; ex-smokers had ceased smoking at least 18 months before interview. The status of PASSIVE SMOKING AND LUNG CANCER RISK -- - TABLE 1-Sex, Etnnlclty, and Aqe Olstrlbutton of Sub/ects by Persona( Clgarette Smoking Status In a Case-Control Study In New Mexico, 1980-44 Eigarene Smoking Status Current Former Nsver SuCjecu Age (years) Case Control Case Control Case - Control Male Hispanic \Vhile <65 34 22 10 18 10 x65 47 30. 27 29 21 NcXt-l-lispanic Wtute <65 77 57 19 59 1 38 t65 82 60 62 103 6 63 Female Hispanic Wnite <65 11 8 3 7 2 27 265 27 6 5 5 7 .34 Non-Hispanic 1Mvte <65 74 34 8 17 3 47 x65 64 15 31 19 8 54 potential confounding by personal cigarette use was controlled by entering the average daily cigarette consumption, the dura- tion of smoking, years since stopping for ex-smokers,'and an interaction term calculated as the product of smoking duration and an indicator variable for age less than 65 years or 65 years and older. This model was selected on the basis of analyses described in more detail elsewhere.v The all-subjects models included Eontrpl for sex. The two categorical indicators of passive exposure were tested individually in each model. Trends in risk with number of cigarettes of exposure daily and with duracion were examined by fitting models with indicator variables to. define categories of unexposed, exposed at or below the median, and above the median. Risk estimation for the effect of indirect exposure to asbestos was limited to females as no males were indirectly exposed. Logistic regression models were employed that r^- -;- cmnkin¢ ac described above, for current ~o~,~~zszoz

to be exposed. Data An~vsis For these analyses, cibarcttc smokers werc those indi- viduals who had smoked at least six months. Current smok- -- crs were those still smoking at interview or who had stopped within the previous iS months: cx-smokers had ceased smoking at iEast ls months before interview. The status of cases classit;cd by questionnaire as never smokcrs was verified against hospital chart summarics on file at the New \letico Tumor Registry. Of the 2S reported nonsmokcrs, the summarics showed that three cases had smoked cigarettes and that one case had smoked pipes and cigars regularly. Analyscs of thc data for never smokers were performed with and without these four subjec;s. Because the study included only eight males who had never smoked cigarettes, all - analyses wcrc performed for females alone and for all - - subjects combined. We used the :vtantel-Hacnszel technique to control for ct5nicity and age in estimating odds ratios for passive exposure to cicarctte smoi;c, \vithin_ strata of personal ciga- rzttc smoking.`t In these analyscs, age was categorized as below 65 ycars or 63 years and greater. Among never smokers, t!:e exposurc•responsc relation of lung cancer risk with aver age cigarettes smoked daily by the spouse and with durrtion of passive cigarette exposure was tested using >fantc! extension methods for stratified data.'`'- For thcse- - var•iablcs. strataofc;tposurc were detincd by the median level 1:#-:ong all exposed nE%sr imukcrs. Those never cxposcd ucrc the :cfcrcacc group for all analyses. To c~ mine fur,!;cr the etTec:s of the passive e<pusures, iosistic rcc,rcssiun mo1c!s wcrc t::tcd for smokers and never smoker-s. AIl models included adjustment for cthnicitu and four ,:stegores of ace, variables "or ~hich the controls had been f,ccucncy m:Itc!tcd to the ca~!cs. In the modcl for smol:crt, A_JPt-1 !.tay ' 9B?. vel ". \o. 5 4o4ezzgz0z Trends in risk with number of cigarettes of exposure datly and - - - with duration were exarninc 5y fitting models with indicator variables to define categorics of unexposed, exposed at or- below the median, and above the median. Risk estimation for the effect of indirect exposure to asbestos was limited to females as no males were indirectly exposed. Logistic regression models were employed that controlled for active smoking as described above, for current and cx-smokers,, and for marriage to a smoker for never smokers. Because surrogate interviews were necessary for 52 per cent of the cases, we assessed the effect of information source by performing the analyses separately for self-rcported and surrogatc-reportcd cases, using self-reportcd controls. We cxcludcd from these analyses the 13 controls for whom surrogate interviews had been necessary. All cross tabulations and logistic models were performed with standard programs of the Statistical Analysis System.=' Odds ratios (OR) and 90 per cent two-sided Cornfield confi-e dcncc intervals (CI) were calculated using program 23 from the Rothman and Boice text for programmable calcula- tors.`3,,6 Results The analyses were restricted to those 1,390 subjects with known passive and personal smoking status (Table l). The 35 - excluded subjects were older than those included (mean aC.^ 6S.4 vs 65.6 yezrs, respectively). More cases were ercludc: than controls (5.0 per cent vs 0.4 per cent, respectively), due in part to the greater proportion of surrogate intcrviews for cases than for centrols. The percentage of subjects excluded did not differ by ethnicity or sex. Based on data in the New Mexico Tumor Registry filcs, the cases described by interview data as "never smokers" 599

HU!A6LE. _:::-T AL TABLE 3-Ocds Ratfo' Estimates to( Passive Clpar.ne Exposur• In a - - Cas.-Conlrol Study of Lung Canev In N... M.xlco, 1980-81 ` Peryotat Srt+ok-n4 AJi Sut)ects Females Only FaS3rve Exacaure -- -- S.atLs CR ?0 . Ct° OR 90 : Ct° C-qare"es 6+" Currer+t 1 2 0.9. 1.6 0.9 0.4. 2.2 Former 1 1 0.8. 1.5 0..' 0..'. 2.2 ye,e. : 9 1.3. 6.7 1.8 0.6, 5.4 Cssraces ar.:.cv Current 1.2 0.9. 1.6 0.9 0.5, 1.8 P" E+r ci§ar Fcrmer 1.1 0.8. 1.5 0.6 0.2. 1.7 Never 3.2 1.5, 7.2 2.3 0.9, 6.6 br .av„aty ea n,at ut.r rswrts. 'f wm cv..v c+oaa tsa.ra:c~s: saustm.ra tor .p0 or "F..o1.s.O.u A3 Gw e..K G6n.'rr:.ncr rK..vaa. who were cvcr mar-icd to a smoking spouse included eight adenocarcinomas, two cpidcrmoid carcinomas, two small cell carcinomas, and four largc ccll carcinomas. Thc eight nonezposed cases rcportcd to be never smokers comprised six adcnocarcinomas and two epidermoid carcinomas. A specific histological type had not been assigned to four of the cascs. Of the four cases in rcported.never smokers but who were identified by Tumor Registry information as smokers, one was small cell carcinoma, two were adenocarcinoma, and -- - - - - one was not classified. Because material was only retrieved for 17 cascs for panel review, we did not compare the exposed and noncxposed based on the pathologists' classification. Of the 17 cascs, the cell typc based on the panel's review concurred with that in the Registry for only eight cases. In the never smoking controls, marriage to a smoker of any type of tobacco was reported for 28 per cent of males and for 56 per cent of femaiEs. The corresponding percentages for marriage to a smoker of cigarettes alone were similar, 28 per cent for males and S i p cr cent for females. - Using stratified and unstratified approaches, no ellect of marriage to a smoker was found among current or former cigarette smokers (Table 2). By contrast, among never smokers, cigarette smoking by a spouse, regardless of pipe or cigar use, was associated with a three-fold increased risk of lung cancer. Adjustment for ethnicity (OR = 3.2, 90 per cent CI (Confidence Interval] = 1.5, 7.2) or for age (OR = 3.2, 90 per cent C1 = 1.5, 7.3) did not change the estimated risks. A - - similar close agreement of crude (Table 2) and adjusted TABLE 3-Odds Ratio Estlmates from Multiple Loqistlc• Anatyse_ s of Passive Cloaretle Exposure and Lung Caneer Rlsk, In a Casa-6ontrol Study In Nev+ Mexico, 1980-b4 AI( Subjects Femate Only Passive Exposure Personai Smoking 5tatus OR 90% Ci OR 90% CI Cigarette only Ever' 1.0 0.8, 1.4 1.0 0.5. 1.9 Never 2.2 1.0, 4.9 1.7 0.6.4.3 Cigarenes anG'oc Ever° 1.0 0.8, 1.3 0.9 0.5. 1.5 ppe or cigar Never 2.6 1.2.5.6 2.2 0.9.5.5 •Aa rtxpdelt ndvded variebles b corrtrd lor lne IriiQuoncy rnalWnq on aqe an0 .Bv,inry. and s.x, wn.n approonate. - - - - - - - - '1MoC.ka tor art.ok.rs contrOa.d tor p.rsccW piqar.Ra us. as deacnD.d ur.C.r Marlods. TABL€ 4-Odds Ratio' Estimates by Duration of Spouse Ciqarette Smoking and by Average Cigarettes Smoked Daily by the e Spousa(s) among Nevec Smokers In a Case-Control Study In N.w Mexico, 1980-84 Duration s26 Y.am >26 Ysars SubjeCY CN /or Group OR 90y. Ct OR 90% Ct irenC , Att Sut~ects 2.2 0.8. 5.9 ?.7 1.0.7.1 2.01 Females only 1.6 0.5, 5.8 Z1 0.7, 6.9 1.23 Mean Cigarettes per Oay Szo OR 90*/: Ci OR a0 : Ct Att Subjects 2.8 1.2, 6.6 2.2 0.6, 7.3 1.82 Femaiaa onty 1.8 0.6.5.6 1.2 0.3.5.2 0.46 `Odds ranos nol aQusc.d br age or .cvtic4y. Aawsanerrt kx eith.. d tnse factora G+d not cnar" lne r.suhs The rt/.rsirK ntegory was tt» Mv.r kposed1 by the spouse was limited to never smokers. For the all-subjects and females-only cross tabular analysss, a pat- tern of increased risk with greater duration of cigarette exposure was found (Table 4). In contrast, the logistic models .. . . . , , . ~ _. . _ _.. _ . _ .,. _ '9o4cazsz0a

one was not ctasstnea. ticcausc matcnal was uttty rcuicvcu for 17 cases for panel review, wc did not compare tk ^xposcd and nonexposed based on the pathologists' classih.ation_ Of thc 17 cases, the cc(l type based on the panel's review c_oncurred with that in the Registry for only eight cases. In the never smoking controls, marriage to a smoker of any type of tobacco was fepor:e-d for 28 per cent of males and for 56 per cent of femalcs. The corresponding pcrccntagcs for marriage to a smoker of cigarettes alone were similar,'_g per cent for malcs and 57 per cent for females. Using stratified and unstratified approaches, no effect of - marriagc to a smoker was found among cur:ent or former cirs-arette smokers (Table 2). By contrast, among never smokers. cigarette smoking by a spouse, regardless of pipe or cigar use, was associated w•ith a thrce-fold increased risk of lung cancer. Adjustment for cthnicity (OR = 3.2, 90 per cent CI (Confidence Interval) = 1.5, 7.2 ) or for age (OR =.i.2, 90 per cent CI = 1.5. 7.3) did not change the estimated risks. A similar close agreement of crLde (Table 2) and adjusted estimates was observed for exposure to cigarettes only: ethnicity-adjusted OR = 3.0 (CI = 1.3. 6.8) and agc-adjusted OR ='_.9 (CI = 1.3. 6.7). There were insufficient subjects to adjust simultaneously for cthnicity and age. Although the odds ratios wer c reduced, restriction of the sample to females did not change the pattern of effect from that found in the analyses with all subjects. When the analyses were per- formed separately for self- and surrogatc-rcported cases, the odds ratios %vere comparably elevated for both groups (data not shown). Because the control series did not include sut'icicnt numbers of controls with surrogate interviews, the controls could not be similarly str atified by ta<pc of interview. Odds ratios from the logistic models (Table 3) tended to be '.ow•cr than from the unstratified and stratified analyses (TaSle 2). Risk estimates for the currcnt and former smokers from thE logistic models also showed no effect of passive ciearctte exposure beyond that of active smoking. However, among the never smokers all point estimates were above - unit~~. .-%ssessment of csposure-response relation for the dura-n tion of exposure and for the average cigarettes smoked daily 600 + Alt Subiects r-emales onty 2_.2 1.6 0.8. 5.9 O.S. 511 ' 27 2.1 1.0.7.1 0.7.6.9 2.01 1.23 Mean C+garer,es per Day s29 >20 oR 9oy Cl OR a0- . CI All Subiecu 2.8 1.2, 6.6 ?.2 0.6. 7.3 1.82 Females onty 1.8 0.8, 5.6 1.2 0.3, 5.2 0.46 'Ocas raoos noc sckvsc.a ta a~e or .;tvucrty. L4usun.nt tor woi.r of ms4 tac:ora C+a not c?sarne ne nsaurs. Tho r.Iefx4 catnoqy wu Cfe n.va ..pOs.d. by the spouse was limited to never smokers. For the all-subjects and females-only cross tabular analyses, a pat- tern of increased risk with greater duration of cigarette exposure was found (Table 4). In contrast, the logistic models did not show an increase with duration of exposure in either group: (for all subjects, short duration OR = 1.9, CI = 0.7, 3.7; long duration OR = LS, CI = 0.7, 4.5). The exposure- response pattern for cigarettes smoked daily showed higher odds ratioss for subjects whose spouses smoked a pack or less per day than for those whose spouses smoked greater amounts (Table 4). Control of stratification factors by mul- tiple logistic modeling did not change the pattern of higher relative risk cstimates for nonsmokers exposed to 20 or fewer cigarettes ptr day (OR = 2.0. CI = 0.9, 4.6) compared with those exposed at higher levels (OR = 1.6, CI = 0.5, 4.9). The respective logistic estimates for females were lower: OR for daily exposure of 20 cigarettes or less was 1.6 (CI = 0.6. 4.3) while for exposure to more than 20 cigarettes the OR was 1.2 _ _- (CI =_ 0.3, 4.4). Potential indirect exposure to asbestos was only report- ed for females. In the controls,. 14.5 per cent of women were - - designated as exposed based on their husband's work history - and 8.2 per cent were considered as exposed based o report of their husband's occupational exposure to asbestos. The cffects of the asbestos exposure variables were assessed AJPH May 1987. Vol. 77, No. 5 s°~, ~~z9zoz

PASSIVE 5&40K1~ %NO LUNG CANCER RISK TABLE 5--Estlmates o/ Lunq 6ancer Risk from Spause'a Occuaatlonal Exposure to Asb+stos, by Reportln8 Sourc., tor Female. In a Cise-Controt Study in N.w MexiCO, 15~0-84 Erroieynen: rn Ilsoestas-RelateG Jo0 Persc•~a~ Srr,ctu,9 5w,t:s :,11 Su'7cCS jztt• repor,ed furse+C,.ate' reported =ver' CR 0.5 e.7 1.1 4C-. CI 04. 1 6 0.3. 1.5 0.5, 2.8 N_a er CR 2.5 1.2 3.3 _ 61 1.0. 6.4 0.2. 8.2 1.1, 9.5 ReporleC as E::osed at )York Exsr' GR 9J!r. CI Neve_r OR 9t:7: CI .UI SvaleC-s Se11- reoarted $urroc,ate' re~oneo 1.4 ;.3 2.0 0.6, 3.2 0.5. 3.4 0.7, 5.5 2.2 2.8 2.0 0.5, 4.: 0.4, 20.7 0.3. 13.9 cy*.•x ana bm,.f smo+..n .Cw.a. ';:_s3-!K+urs.C m+cai w.rl e1e oorrvans9n Qiwa tnr p.. sv.o9a:r-noort.d 6asa- with multiple logistic models and found to vary with cigarette smolting habits (Table -9. The odds r-atios wcrc higher for the ncver smoking femalss; and in these ncver smokers the two exposure variables gave comparable risk estimates. Discussion ln the context of a population-based case-control study in :vcw \te.tico, wc have Fxamined the_ risk of lung cancer associated with marriagc to a cigarette smoker. The results indicatcd increased nsk from this exposure in never smokers, but not in active smokers. Methodologic limitations of the case-control approach for studying the relation between involuntary exposure to tobacco smoke and lung cancer must be considered. Misclas- sihcation of both active and passive exposure to cigarette smoke is of particular concern. With regard to active smok- ing, we assigned exposure on the basis of a comprehensive interview with either thc_ index case or a surrogate respon- rier,t. For four of the 228 cases among never smokers, although spouses aware of the putative association of passive smoking with lung cancer may have minimized their own smoking. Spouse surrogates may have supplied more accu- rate information conccrning their o,,.'n smoking o_ king than would have been available from the indcx_ subject. The much higher proportion of surrosatc interviews for cases than for controls could have introduccd differential misclassitication and bi- _ - _ ascd effect measures upwards. Tilr results of f the present case-control study comple- ment those from other casE-control studies' _-10 and from cohort studics,3.'1 which showed increased lung cancer risks in never smokers married to smokers. The magnitude of the etTcct of marriagc to a smoker in the present study, about a two-fold increase in risk (Tables 2 and 3), is comparable to findings by Hirayama3 and by Akiba. et al,9 in JaF an, by Trichopoulos, er al,` in Greece, and by Correa, et p1, and by Dalager, et al•10 in the United States. A weak exposure- response relation was present with duration of passive exposure, but not with average number ber of cigarcttes smoked daily by the spouse (Table 4). In contrast, in a larger case-control study, Garfinkle, et al,a found a trend of increas- ing risk for nonsmoking women with the number of cigarettes - smoked daily at home by their husbands. In active smokers, we found that residence with a smoker did not elevate lung cancer risk (Table 2). The lack.of association in activc smokers is consistent with the quanti- tativc differences in the exposures of active and passive smoking.b Furthermore, active smokers must receive more passive exposure to tobacco smoke from their own smoking, than from the smoking of others. The odds ratios for passive smoking in active smokers, all at or near unity, provide evidence against consistent under- or overrEporting of expo- sure (Tables 2 and 3). We also assessed the cffccts of marriage to a spouse employed in jobs possibly involving contact with asbestos. We hypothesized that asbestos brought into the home by the spouse might increase lung cancer riskk in smokers and nonsmokers. Domestic exposure has been previously asso- ciated with mesothelioma, pleural abnormalities, and changes in the lung parenchyma.=9 We used both a lifetime occupational history for the spouse of the index case and reported contact with asbestos to assess possible indirect exposure of the cases to asbestos. with both approaches for determining exposure, we ----_ rrr.,s.t. S1 4,t4ezzgZ0z

with multiple logistic models and found to vary wttn cabarcttc smoking habits (Table 5). The odds ratios,.vcrc higl, for the never smoking `cmalcs; and in these never smohErs thc two exposure variables gave comparable nsk estimates. Discussion In the contc.st of a t^op ulation-bascd casc-control study - in ~cw Niexico, .16-c haVc examined the_ nsk of lung canccr associated vL•ith ma.rriagc to a cigarettc smoker. 1'he results -- indicatcd increascd risk from this exposure in never smokers, but nat in active smokers. Nicthodologic limitations of the case-control approach for studying the relation bctwcen involuntary exposure to tot,acco smokc and lung cancer must be considered. `tisclas- sincation of both active and passive exposure to cigarette smoke is of particular concern. with regard to active smok- - - - inS, we assigned exposure on the basis of a comprehensive intcn•icw with either the index case or a surrogatc respon- dent. For four of the '-S cases among never smoGers, infor:nation in the hospital record cocllictcd with the intcr- vicw. Because a similar, additional source of data was not available for controls, we did not exclude the four cases from this report. The findings were unchanged, howcvcr, when thcv wcr c removed from the analyses. Wc assessed passive exposure to tobacco smoke only frotn marriage to a smoking spousc; exposures from other sa;oi;cr s at home and in the workplace wcrc not assessed. Thus. subjects may have been misclassified on total passive smoke exposure. Wald and Ritchie=' have shown that non- smoking men married to smoking women rcport greatEr Gx70surE to the smoke cf others outside of the home than nonsmoking men married to nonsmoking women. 1Vald and Richic suggest that information on smoking by the spouse convc\ s-somc information on other sources of exposure. Su,:ogatc inten•icws were necessary for 19 of the 28 nc%cr smokers. While the validitv of surrocate information has bcEn qucsttoncd for somc crhoNurFs,ts thc surrogatc respondents -Acrc prtmarilv surviving spouscs, who provided in!`orrnatton on their o\vn smoking habits and those of prcvtou5 .pousc>. if an~ Extensive misclassttication intro- ,'uccd by the iurTocatc tntcrvtews thus appcars unlikely, a.1PH May ' S6; . Vol ''. No. 5 association in active smokers is consistent with the quanu tativc differences in the expc res of active and passive smoking.6 Furthermore, active smokers must receive more passive exposure to tobacco smoke from their own smoking, - than from the smoking of others. The odds ratios for passive smoking in active smokers, all at or near unity, provide evidence against consistent und6r- or overreporting of expo- sure (Tables 2 and 3). Wc also assessed the cdccts of marriage to a spouse - employed in jobs possibly involving contact with asbestos. We hypothesized that asbestos brought into the horne by the - spouse might increase lung cancer risk in smokers and nonsmokers. Domestic exposure has been previously a_ sso- ciated with mesothelioma, pleural abnormalities, and changes in the lung pa,renchyma." Wc used both a lifetime occupational history for the spouse of the index case and reported contact with asbestos to assess possible indirect exposure of the cases to asbestos. \Vith both approaches for determining exposurc, we found associated elevations of risk for lung cancer (Table 5). The effect was more evident in never smokers, although comparable relative risks would be anticipated if cigarette smoking and asbestos exposure interact rttultiplicativcly in this sctting.`"-30 The magnitude of cfrcct was surprisingly large in vicw of the range of c:cccss risk found in asbestos- exposed _ exposcd workers and of the results of risk estimation.`9.30 ACKNOWL€DGMENTS Suppor•,cd by a snnt from the National Cancer Institute. CA :7187, and by a contract from_ thc Biomctry Branch. National Cancer Institute NO1-CN- S51_6. Dr. Samct is rccipicnt of a Rescarch Career Development Award. 5 K04 HLOOJ5l, from the Division of Lung Discascs. National 1-tcart, Lung, and Blood Institute. REFERENCES I. US Depanment uf Hcalth. Education, and I1'ctfurr: SmokinE and Ftcalth: Retwrt of thE Ad,vsory Committee to the Sur;eun General of the Public Hcalth scrvicc. I'HS Pub. No. 1103. wa.hin;ti+n. DC: Govt Printing Office. 1964. :. US Dcpanmcnt , f Health and Human Servizes. Puhlik Hcallh Scrvt<e: The He:tlth lanacquenccs of 5motjngt t..tmcr: a rcM>rt .d the Surgeun General. Ru<kvillc, yD' Ottice uf Jmitktn} and Hc.tlth, 1`RtZ. 601 I Tie-(.'zz!)z0~r

HUMBLE. GT AL. 3. H nsama T'. Nonsmoking %tvcs of hca.y smokcrz havc a highcr nsk of lung canccr. a study from Japan. Br ',tcd 1 1981: ?8 :133-185. 4. Tncho[`oulos D. Kclandidt A. Spanos L: Lung cancer and passive --- smok ng. Int 1 Cancer 1981; _7:1--t. 3. Weiss ST: Passive smoking and lung c_nccr. Am Rev Rcspir Dis 19S6; 1 tj:l-3. 6. w'ortd Health Crg2nizatton. International Agency for Research on Cancer: IARC Monographs on the E~aluat.on of thc Carcinotcnic Risk of Chcm- icals to Humvis: Tobacco Sm_oking, Vol. 38. Lyon. France: IARC. 1986. 7. Coata P, Pickle Lµ', Forham €. Lin Y. Haenszcl W: Passive smoking and lung cancer. hncet 1933:'_:_`9JL-597, 8. Garhnk:c L. Auer`ach O. loubert L: involuntary smoking and lung cancer. a casc-cor.trol study. JNCI I 1985: 7,5:i61-t69. 9. Akiba S. Kato H. Blot WJ: Pusivc smoking and lung cancer among Japa:xse women. Cancer Res 1986: 46:a801--t307. 10. thlagcr NA. Piekle LW. htasoa TJ. cS al. The relation of pusive smoking to lung canccr. Cancer Res 1986: i6:4SOS--ts1 l, 11. Gar$nk_ le L. i tmc trends in lung cancer morsa.lity among nonsmekcrs and a note on passive smoking. JNCI 1981: 66:1061-1066. 1:. Gillis CR. tiale DJ, Hawthorne vh(, Boyle P: The cfcct ef cnvironmcncal tobacco smoke in two urban communities in the West of Scotland. Eur J Rcspir Dis 1934: 6S(Suppl 133):1? 1-126. 13. Katiat GC. Wyndcr EL: Lung cancer in nonsmoken. Cancer 1934; 53:1.14-I22I. 14. Chan WC. Coa`ournc StJ. Fung SC. Ho HC: Bronchial cancer in Hong Kong 1976-1977. Br I Cancer 1979; 39:18:-19_. IS. Koo LC. Ho J k{-C, Lcc ti: An analysis of somc risk factors for lung cancer in Hong Kong. Int I Cancer 1953: 33:1.9-1SS. 16. Wu AH. Henderson BE. Pike \1C, Yu MC: Smoking and othernsk factors for lung cancer in Women. J1CI 19S3: 74:747-751. 17. Sunet 1T`(. Key CR, kutviri DM. Wiggins CL: Rcspiratory disease morulity in New \ie.cico's.!tmerican Indians and Hispanics. Am1 Public HczJtlt 1930: :0:49_--t97. 1989 Revisions of the US Standard Certificates and Re ports I;S. Key CR: Cancer inci.lcncc and mortality in New Mexico. 1973-77. !n: US Dc; artmcnt of Health and Human Services: Surveillance, epidemiology. and tnd results: incidence and monali(y data. 1973-77 (monograph 57). NIH Pvb. No. 8I-2330. Bethesda. MD: National Cancer Institute. 1981. 19. Butler C. Samct J Sf, Humblc CG, SWccney-ESs The histo0athology of lung cancer in New Mccico. 1970-197? and 1980=1981. 1NCt. :0. World Health Organization: The World Health Organization Histological Typing of Lung Tumors. 2nd Ed. Am I Clin Pathol 1982: 77:1:1-136. 11. hlantcl N. Hacnszel W: Statistical aspects of the analysis of data from retrospective studies of disease. JNCI 1939; 33:719-748. ?:. Mantci `t: l::hi-square tests with_ one-dcgree of freedom: extensions of the Mantcl-Hacns.cl procedure. I Am Stat Assoc 1963: 58:690•-700. 23. Pathak DR, Samct 1M, Humble CG. Skipper BJ: Dcterminants of lung cancer risk in cigarette smokers in New Mexico. JNCJ 1996; 76:397-.603. :4. S1. S Institute SAS User's Guide: Statistics. 1982 Ed. Cary. NC: SAS Institute. 1982. -- :3. Cornfield Js A statistical problem arising from retrospective studies. In: Neyman 1(ed): Proceedings of the 3rd Berkeley Symposium. Berkcley: - University of California Press. 1956: t:133-148. '_6. Rothman KJ, Boice JD Jr: Epidemiologie Analysis with a Programmable Calculator. Boston: Epidemiology Resources. 1982. .7. Wald N. Ritchic C: Validation of studies of lung cancer in non-smokcrs married to smokers (Ictter). Lancet 1984; 1:1067. 28. Gordis L: Should dead cases be matched to deadeontrels' Am 1€pidemiol 1982; 1IS:1-S. :9. National Resear<h Council. Committce on Nonoeeupational Health Risks of Asbestiform Fibers: Asbestiform Fibers: Nonoccupational Health Risks. Washington. DC: National Academy Press. 1984, 30. US Dcpartmcnt of Health and Human Services. Public Health Service: The Health Conscquences of Smoking: Cancer and Chronic Lung Disease in the Workplace: a report of the Surgeon General. Rockvilte, 1`fD: O95cc on Smoking and Health. 1985. The National Center for Hcalth Statistics (NCHS) has recently distributed to the 50 states the 1989 revisions of the US Standard Certificates and Reports of Live Birth. Death, Fetal Death, Induced Termination of Pregnancy, MarTiage, and Divorce. These documents serve as models for the various states to use in developing their own forms. NCHS recommends that revised certificates and reports - - incorporating the 1989 changes be implemented in all states by January 1, 1989. The US Standard Certificates and Reports were developed jointly by the NCHS and state vital registration and stabstics executives. Advice was obtained from persons and organizations throughout tp„tp,t tt«n „f vitnl statistics data and those who complete the documents. Zt4CZz9Zoz

1-tealtA 19J0: - 1989 Revisions of the US Standard Certificates and Re ports The National Center for F-tealth Statistics (i 1CHS) has recently distributed to the 50 states the 1989 revisions of the US Standard Certificates and Reports of Live Birth, Death,. Fetal Death, Induced - Termination of Pregnancy, Marriage, and Divorce. These documents serve as models for the various states to use in developing their own forms. NCHS recommends that revised certificates and reports incorporating the 1959 changes be implemented in all states by January 1. 1989. The US Standard Certificates and Reports were developed jointly by the NCHS and state vital registration and statistics ezecutivcs. Advice was obtained from persons and organizations throughout the United States who representcd users of vital statistics data and those who complete the documents. The content reflects a consensus of what needs to be collected about each vital event to serve both the legal and statistical uses of these records in the 1900s. Among the more significant modifications made in these new revisions are: • the addition of an Hispanic identifier to the live birth and death ccrtificates and the fetal death and induced termination of pregnancy rcports; • changes in the birth certificate and fecal death report to obtain more detailed information about - the pregnancy and its outeomc; and • some of the factors that may have improved quality and completeness of the cause of death. Information about the revision process and copies of the standard certificates and reports can be - obtained by writing or calling; George A. Gay Chicf, Rcgistration Methods Branch Division of Vital Statistics, NCHS 3700 East-Wcst Highway, Room 1--t4 Hyattsville. Maryland 20782 Tel: (301) 436-8815 602 AJPH May 1987. Vol. 77. No. 5 1:1W-.EZzsZoz