Tobacco Documents Online

or J. GrWr. 11!9t61. 54. i7-105 Relationship of passive smoking to risk of lung cancer and other smoking-associated diseases P.N. L.ee,• J. Chatrtbcrlaii'1 & M.R. Aldersont; lrtsnruer of Cancer Rararch. Clijton Road. &Lnont. Srrrrty. d1K. S.uiwary In the laner trn of a large hospaut osecontrol atudy of the telatanship of tM+c of qps,ntc smoked'to nsk of varw.u anwking-astociatod ditcascs, pattents answerad questwn. on thc smolinr hahar uf their first t,lwWc and on ttx cuent of passuve smoke ecposure at horee. at work, dunng travel and dunng kssure. In an atenuon of this study an arternpt,wu made to obtain smoking liabt'dau directly from thc spou.sea of all lifelong non-smoking lung mnsr easa and of two lifelong nonsmokmp matched ctintrols for och case The aucmpt was made regardless orwhether the pauenu had answered paws sy+wkrnt questans in hospital or not. Amongst fifelong non-smokers.,passive smoking was aot associated with any srgnifiont ir+crease in nnk of kont onou. chronic bronchuu, itchacmK hnn disease or stroke tn any analysts. Limitations of past studies on passive smoking art disntsaed and the need for funhe+ nsan:h dndcrlincd. From all the available eyidcntz. it appears that any elTect of passive smoke on nsk of any of the major duoses that have boen associated with aaive amok'ing ii at ntost tmall. and may not eaiu at all. Stud v of Gospi ral in-patunrr In 1977 a largc hospital txise<ontrol was initiated to study the relauonship of the type of eigarette smoked to risk of lung cancsr, chronic bronchitis, ischacmic heart disease and strokc. This study was orried out in 10 hospuali regions in England; interviewing ended' in lanuary 1982. The original questionnaiis did not include questions on passive smoking as it was not considered an important issue in 1977., However, in 1979' it was decided to extend the questionnaire to cover passive smoking for mamed patients for the last fiour regtons to begin interviewing. Subsequently, in 1981. following publication of the papers by Hirayama (1981) and by Trichopoulos rr ot. (1981) claiming, that non- smoking wives of smokers had a significantly greater risk of lung cancer than non-smoking wives of non-smokers, it was decided to increase the number of interviews of married lung cancer casa and controls. The extcnded qucstionna,re was then administered to these patients in all hospitals where intervie,nng was still continuing. foJlow-rp srrdr of spouou of won-srna[ing Aaspital w-patirnts In 1982: after interviewing of hospital in-paticnts had been eompktod. it was decided to carry out a follow-up siudy. In ibis studv. an attcmpt was CwrapondenQ: ~. V . t~e. •treuru addnsc 25 (:edar Itoad Swtoa, Surrsy. St.t2 SDG: tPresertt addrnr' OITia of rupylauoa Censuses nd Surveys. St. Cathenee'a House. 10 Kiapwar., London wcts ur. made to interview the spouses of all or the marricd hospital ip-patients with lunL eanecr who rrported never having smokcd, as well as of two marricd non.smoking controls for each of these index lung cancer casa. The follow-up study was intendcd partly to c:ompare in(ormauon on spousei smoking habits obtained first-hand with that obtained sxond-hand during the in-paticnt intervicws- and partly to obtain sflmc d:rta on spouscs' smokin_; habits for those patients who had not answered passive smoking questions in hospital.. This paper concentrates solely on the issue or passive smoking in lifelong non-smokcrs. Results naatin8 to type of cigarette smoked rc described elsewhere (AIdcrson tt al:, 1985): whilc a dcuilcd {epon, available on request from PNL. considers the overall f ndings from this etscrontrol study. 11getliods and respowse Sttdy of hospital in-patirnts For each of the 4 index diagnoses (Uung ctncur chronic broneltitis. iseluemic bean dieease ar,~: sarole); the intention was to interview 200 cases and 200 ntatched' controls in each of the eight aex/a8e alls (i.e. mak or femak, and aged 3S-4-t. 45-54, SS-64 or 65-74). This gave a target or 1Z800 p.tients. though for some categories (e.g. young femak chronx bronchitics) this would be unattainable. Patients were selected from medical ('induding d+est ttKditane), thoracic wrgery, and radiotlierapy wardt. Controls were patients without one of the four index diagnoses. adindtully tnattdied to aaus on sex. agc, hospital' region ar.C, 0 The Mac~wiRYn rtact l.W.. IRri6

9/t LN' LFF, ri.l when pasubk, hospital ward and time of interview. Subscqucntly, whcn final discharge diagnoses became availahk, thcy were used to reallocate cases and' controls as nooc~sary. Patients without a final diagnosis kept their provisional diagnosis. Wbere changes in wsetontrol status occurred; patients were regrouped into new case-control pairs as appropriate. With the assistance of Sir Richard Doll and Mr Rkh.,rd Pctu, non-indcx diagnoses were classified as follows: class IA 'dcfinitcly not smoking associated' class 1:8 'probably not smoking associated' class 2A 'probably smoking associated' class 26 'dcfinuely smoking associated' Controls with no final diagnosis were eonsiderod' class I'B Overall, there were 12,693 interviews carried out which resulted in 4,950 pairs with elass I controls and 7?+1 paits with class 2 controls.. There were 3,83: intervicws of married cases and controls where the passive smoking questionnaire was completed In order to avoid substantial loss of data, due uo onc member of a pair not being married or not completing the passive smoking questionnaitc, it was decided to ignore matching when analysing the passive smoking data and to compare each inde>< group with the combined eontrols. Numbcrs by sez and easetontrol stat'us art given in,Tablc 1. Tsbtr I Numhcn of married hospital in-patients eomt+trlinR passi.e anoking questionnaires Mdr Frwdr Tad Lung canocr 547 243 792 CtlronK brOnChttls 1f2 64 266 IscMaemic hean diseame 216 221 507 Slroke 161 137 298 Controla C1as. IA and 1!' 839 713 1,352 Clam 2A and 2B' 2" 149 417 Total 22113 1,3+19 3.932 'Ohcr dnrases arc elau+fied by Irgrer of amoking asaociatam - ciau tA: dcfinitcly not, class I!. pob.Wy aot, ctass 2A prolaabiy. ci.w 25 ddinitcly. In the passivc smo.rng p.rt of slse questionnairc, patients were asked when the marriage staned; if and when it had ended; the number of manufactured cigarettes per day smoked by the spouse both during the last 12 months of marriage and also at the period of maximum smoking during the marriagc: and whether the spouse ever regularly smoked hand-rolled cigarettes, cigars or a pipe during the tnarrisde. For seeond or subsequent anarrisges„ queuions related to the first tnarriagc to give the longest btcnt intervali betweerr exposure and discasc onset The paticnu were also asked to quantify, according to a four-point scale (a lot, average, a little, not at all), the extent to which they were regularly exposed to tobacco smoke from other pcoplc prior lo coming into hospital in 44 situations: at home; at work; during daily travel;,. during leisure time. In thc main questionnaire, detailcd qucstions were aakcd on smoking habits and on a whole range of possible confounding variables. folloM-rf+ study ojspousrs ol nnn.smoking Aosritot in-patitnrs From, the hospital study there were 56 lung wnoer cases who reported being lifclong non-smokers, who were married at the time of interview and who were not known, to have been married previously In a follow-up to the main study, an attempt was made to interview the spouses of these 56 cases and also thr spouses of two lifc-long non-smoking controls for each ease, individually matched for sex, marital status and 10=year age group and, as far as possiblis, hospital. Where multiple potential controls in the same hospital werc available, those interviewed nearest in timc to the case were selected Where suitabla controls in the same hospital were not availyblc, those in the nearest hospital were choscn. Because namcs and addresses of the patients were not recorded in the hospital study- it was necrssary) to go back to the hospital both to obtain this information and also to get permission to interview their spouses Following some refusals both by the hospital~ and by the spouscs, suoccssfui interviews were obtained from spouses of 34 cases (t0 wives and 24 husbands) and' 80 controls (26 wives and 54 husbands) whose condition was definitely or probably not related to smoking... Intcrncwing was carried out between July 1982 and August 1983. The spouses were asked about, their eonsumption of manufactured eigarcttcs, cigars and pipes (a) nowadays. (b) during the year of admission of the patient or (c) masimum dunng she whole of the marriage. The spouses were not asked' about the smoking habits of the index patient. The sfwuscs were also asked questions on agc, ocrupatton, social class and a range of other potential confounding factors. Srotisrirol iwrtAodt The statistinl' methods are based on classical prooedures for analysis of grouped data dcrived from case-control studies (Breslow dc Day. 1980). In Eencral, the materiali has been examirsed as a 2 x K x S tabk, with A' representing the levels of the

,~,,,_ „~ ... ,,,..., _,.,. ~ ,...., .. , ~....,.. . . ._ . .,.... .,.~ ,~,.,. PASSIVE SMOKING AND SMOKiNC..RELATED DISF.ASfS p . I , . risk factor of intercst and S the number or strata used to take aecountof potcntiaLeonfounden Results prrscnted are for the combined strata and show the relative nsk (Mantcl-lHacnszrl estimate) together with the significance of its differcnee from a base kvell(nsk I.0): and/or the denc-relatcd trend. In analyses iof the data collected in hospital, eomparisons arc made between aws with a particular index discasc and aW the controls with diseases dcGnitei'y or (nrobablyy not related to ttmoking. Six simple indices of passive smoke exposure were considered in these katter analyses. (i)-(tv)' exposure at homc, at rark, during tnvel, during kisure. (v) spouse smoking manufactured cigarettes in the last 12 months, and (vi) spause smoking manufactured cigarcttes in the whole of the marriage. Bases for (it) are reduecd as not, all patients worked: In addition, a combined index of .passive smoke exposure was akulated' by the unweighted sum of the four individual exposure indicrs (i}{iv), counting 'not at all' as 0. 'httk' as 1, 'average' as 2 and 'a lot' as 3. Retalts l.rna cancer The follow-up study concerned 56 lung cancer ases and 112 matched' controls who reponed never having smoked in their hospital int'erview. C3f thesc. thert: were 47 cases (IS male and 32 fcmale) and 96 controls (30 rnak and 66 femak) for whom some information on smoking habits of their spnusc.c was availabk. Of these 143 pstients, infotmation on siwuse smoking was available both from the slxwce and from the pa tknt for 59 (11 •/.), from the spouse only for 55 (3R6:) and from thc patient only for 29 (20%). Table ll shaws the estimated agc-adjustcd relative risk of lung cancer in relation to stscwsc srnoking during the whole of the marnaFc, h) sc><, source of data, and prnod of smoking. None of the 9 relative risks shown in the table arc staustically siEnifiant.. When data for both sexes and both sources arc rxnsidcred, the ecrimatcd rrUtivc risks in relation to spoust smoking are el'osc to I(I.l I): For individual se><a or sources, where numbers of nses and controls artr smaller, relative risks vary more from unity, but no consistent pattern is evident. Similar conclusions were reachcd, when analyses were based on smoking during the year or hospital interview:. Herr. the overall relative risk was again close to 1(0.9a with limits 0.41-2.09). Table III summariscs concordance between spouse's manufactured taFarcttc smoking habits as reported directly and indirectly for the 59 pattents with data from both. sourees. Discrepancies were seen for 9 spouses (IS•/.) in respcct of smoking at some time during marriage and in the case of 2 Tabk 11 Relatsonship betaen spouse's manufactured cigarette pnoLtinF dunng the whole of the marruEo and' nsk of lung eanrxr amont hklonF non,smokers /sundsrdisrd tor age) Slrraur d,d .nr sMnlr Spousr anolid Stv of Rrldurr nJ prirnr Caxs Carurdf Cavs Caurd! t9J:, fimts/ dasrd on tnurrYr+-s af thr slwrsr m fhlGr.ry lrndy ( IN Rotrnrrt) ~ Male 5 13 5 13 1.0190 .23J 411' Femae S 16 19 39 1.601044-3.7K), ComtsMned 10 29 24 31 1.33/0.50-34tt)', Aasrd on uurrcxa o/tAr iwdr: rninr w iespiral (R8 Fairwrsl Male 7 IS 3 7 1.53(0.37.6.34)'. Female 9 17 i 20 075(02a•240v CornbMned 16 32 13 27 IA010t1-24t) Iasrd an IwwA aewcn .f u{anwaiiw (f11 pue+wsl• Male 7 16 8 14 1.30(0.33-4.39)' Fesnak 10 21 22 43 1A010.37-2.7U) Covnbined 17 37 - 30 S9 1.1110.31-2.39Y "Uwy ¢ontras .rcluded in foUor-rp audy eorssidered: M. Mi" aaatrsas the npouse was counted as a smokn if seported to be so e+tlre+ dir.eely, by the spouse during follow-up intervir., or, iiidir.ctll, by the patient i. bospitaL Note that the 39 pauengs tor .horn information oa spouse rnokinj was avaiLbk from both muras arc'ucluded in all 3 analysa

IOO: P N l.rr rr rrl Tablr lill Cnnenrdancs belwecn sirouse i tnanu/ictured ciEsrette unoling hahits as repnrted W drrectl} and tndtreclly Sex of pttr+U/lGSr cowlroJ llWru Alali• Fr.w/t Cases Cowrroir Cavs Coetrotr Tatd $r+ousr a smoler sometnnc in rnarrutr .¢nrdurF rn Subyect and spouse 2 6 5 13 26 Only suh)ect I 0 0 3 4 Only spousc t 1 3 0 5 Neither 3 11! I 9 24 % Nlbxct!lpoufe a[rC[nlent. 71% 9r/. 6r,; tta•„ isssj Spouse a smoker durtnE ytar of hospital intcrvrc* accordong to. Suhlcct and spouse 1 6 2 4 13 Only sub1ect. 0 0 0 1 1 Only spnusc 1 0 0 0 1 Neither 5 12 7 20 4t % suh)M lspouse aEreement a6;•; 96% 10D, 100'; 9G ; 9T,: spouses (3%) in respect of smoking dunng the year of hosl.ital intcrvrc.._ Thcrc was no consistent pattern in thc direction of dt.crcpancy,. Tabk IV summanscs the results of analyses arried out rclrting 7 indices of passive smoke exposure recorded rn the hospital interviews to risk of lung canrr, amnng lifelong non.smokers. Here the controls used for eomparison are all never smoking paticnts with diseases classified as dcfinitcly or prob+4>ty not, associated' with smoking who completed thc passive smoking questionnaire. Overall the results showed' no evidence of ann effect or passive smoking on lung cancer incidence among lifelong non-smokers. In mak patients, rebtive risks were increased for sorrx of the indices but numbers of ascc were small and none of the d'ifkrencrs apt+roachcd statistical significance. (n females, where numbers of cases were Wrger, such trends as existed tended to be negative and indecd were marginally significantly ncgative (P<0.05)~ for passive smoking during travel and during kisure. For the combined sexes no dilTcrenecs'or trends were statistically significpnt at the 95'/% eonfidcna 11evel; such trenels as earsted' tending to be slightly DlegaliK. The relative risk in relation to the spouse smoking during the whok or the marriage was estimated to be 0.80 for the seaa combined, with 9S•/% confidenee 6inits of 0.43 lo 1.50. Standardisation for working in a dusty job, thc variable apart from smoking found to have the strongest, association with lung cancer ruk ih the analyses described in Alderson rr al. (1985), did not affect the conclusion that passive smoking was not associated with risk of lung cancer among neverr smokers in our study E'kronic broncAiris, iscliarrnic Arorr dcuasr and srroke Analysa similar to that shown in Tabk IV for lung cancer were also carried out for chronic broneliitis, ischaemic heart disease and stroke. Illustrative results for two of the indices artt presented in Table V. No significant relationship or any index of pusive smoking to risk of the 3 diseases was sren. For the sexes eombinedi the relative risk in relation to the spouse smoking during the whole of the marriage was 0.83 for chronic bronchitis (9Y/% confidence limits 0.31-2.20); 1.03 for ischscmic heart disease (limits 0.65-1.62) and 0.90 for stroke (limits 0.33-1',.52). For stroke there was, in both aeaes, an approximate 2-fold increase im risk for patients with a combined passive smoke indci that was high (score of S to 12) compared with those «rhere it was low (scorc of 0 or 1). However, aumbers of cases with a bigh,seorc were low (14 males and 7 fernaks) and even for the sexes oombined, the relative risk estimate of 2.18 was not tnatistically ttitnifsant (Fimits 0.16-5.48). In inteTpreting this finding, it should be noted that active smoking was not found to be clearly related to stroke in thc main study (Alderson er nl.. 1985), rendering a two-fold ineTCasc in relation to passive smoking a priori unlikely.

I PASSIVE SMOKING AND SMUKI'NG-REIUtTF.O UISGASES lol Tal+k 1)" Relstu+nshir hrt.ern various mdrors of passive smnlc c>vlr+surc and nct of' lung cancsr among (lkbnR ann• .moIcrr tstandardiscJ kx,aac andi lot spouse smolinL whctAcr the marnaRr .aa onpxnE ot endedl /assiir nwnLr Molr prrrus Fewso6t roranus Sran rnwMirl cilrrsvrr rnlrsJ/r.r/ Cav. Cnwrrnls R Casn Cmurrs[rR Cavs Cewtrols R At home Not at all 9 101' 1' 21 192 I 30 293 b Lntk 2 21 122 6 65 092 Il 86 09M AMeraEua lot. 1 I I 1.11 5 61 0.81 6 72 0 R6' At isork Notatafl 3 40 I 12 113 1 IS 15:1 I Lrttlr 6 29 3.24. 3 2h 6.1>( 9 55 1.12 Avcra6c's lot 1' 29 046 0 19 00 1 4A 0.19 During travel Nbt at all, t 101 1 28 239 1 36, 339 1 LrttIc 3 16 2.06 2 51 033 5 67 064 AveraEc/a lot 0 13 0.00 0 13 000 0 26 000 Treed (negative) / <0.05 DurinB leisure Not, at all 3 45 1 15 11'6 4 IE 161 1 Lmk 4 49 1.12 14 107 1.05 I8 155 1.06 AveraEe'a lot S 39 3.16 2 95 018 7 1)4 0.59 . Trend (negative) P<0055 Combined mdrx• Score 0-1 1 27 1 to . 75 f 11 102 1 Scorr 2-4 7 55 4.34 5 61 063 12 116 1.08 Score 5-12 2 15 3.20 0 21 0.00 2 36 0.50 Spouse smoked man aFs in last 12 months No 10 IOS I 20 193 I 3[1 298 1 Yes 2 29 096 11~ 122 0.76 13 151 079 Spouse smoked msr op in whok of tnarriaEe No 7 93 I' 13 99 1 - 20 112 1 Yes 5 40 247 19 229 0 55 24 269 0.80 'Basesd on sum of 0= nor at all. I - httk. 2- averaEt, 3 - a Iw tor at Aorne, at .ork- durmE travtl; dunnE knurc. Disc.ssion. Over the past 4 years thcre has been eonsiderablc research interest in the relationship between passive smoking and risk o[ lun f cancer in nonsmokers. N'h;k somc studies have claimed a positive effect (Htrayama- 1981; Triehopcsulos er al.. 1981;; Corrca n aL, 198z; Garfinkel tr al:, 1985„ Gillis rrr al:,, 19R4. Knoth er al:, 198z). others (Buffkr er al.. 1984. Chan. 1982: Garfinkel. 1981; K.bat and' Wynder. 1911A1, Koo tr d'.. 1,994) have fewnd no signiG: n: ..._..,r. ~:;.. Pc!_t.i.r risks of lung nncer for non-smoking women nnarried to snwkers compared to nonsmokinE women married to non- smokers ran8c from somewhat over 2 in the Trichopoulos and CorTea studies to around 0.75 in the Bufller and Chan studtes The weighted relitive risk from thesc studies ha been estimacd~ by us ass appro><imately 1.3. Whik there is, thcreforc, a lendency for a small positive assrrriotion between passive smoktnF and lung eancer„ roccnl reviews or these d:rta CLcc. 1994. t.ehncrt er al.. 1994) have concluded thaa overalli thcrr is no reliabk scicntific evidcnce of a causal relationship between passive smoking and IunB cancer. ln thesc reviews a number of general points have been made. First. dosimctric studies liave shown that, in alarettecquivaknt terms, passive smoking onlytesults in a relytivcly sma1l eaposurc 1o the non- smoker. Hulod v d. (1978), for esampk, showed IAat' even under quite estrcmc eonditions the time taken for a non-smoker to inhale the equivalent or 1

I 182 1.N CfF r,.1 Tskli V Relafu.nahip between two indiaes of rauivt smoke esposurc and nck of chronic bronchitis, ischacrnic kan ducasc and stroke among lik{ons non•unokcrs (standardised for aRr and: lor spouse smoking, whether the tenamaae was onFrnnB or rndedl Iecwrr a+wAr - Malr prirw fnwdr prrrnr: Sran towAewrf rs/wrwrr ir+dra/krrtl Cav} Cewrrd. at Csui ConrroH R Caur Corwrolr R CArnnir ArrwnrArnii Combined Indca' Score 0-1 1 27 1 7 73 1 8 102 1 Score 2-4 2 SS 0.83 A 61 1.05 6 116 L00 Score 5-12 I I S 1.90 1 21 1.03 2 36 1.30 Spouse smoked man cip in whole of marria8c No 8 93 11 4 89 1 12 192 11 Yes I 40 0.34 1) 229 1.22 11 269 0.93 larAor1wir Aron di.raw Combined inSt.' Score 0-1 13 27 1 23 75 1 38' 102 1 Scorc 2J 112 SS 043 9 61 0:59 21 116 0.52 Score 5- 12 3 IS 043 4 21 0.81 7 36 o.61 Spouse smoked man cip in whole of marriage - Nu 26 93 1 22 89 1' 49 182 1 Yes 15 40 1.24 55 229 093 70 269 1.03 Srrolr Combined imSc.' Scort 6- 1 5 27 1 19 75 1 24 Ii02 1 Score 2-4 10 SS 1'1 10 61 016 20 116 097 Score 5-12 4 IS 1.77 7 21 244 11 36 2.18 Spouse smoked man cip in.rhok of marriage No I! 93 11 119 89 1 37 112 1 Yes 6 40 0:84 49 229 0.92 SS 269 090 •8are+d on iwm of 0. aot at alt. I - littk, 2- average. 3 -a Iw for at home, at work, dunnE ttavel, dunn8 kkisurc. orx eiprcttc would' be I1 hours as reprd's paniculatc matter and 50 hours as re=ard's nicotine. Similarly. Jarvis rt al. (1985) have shown that the incrcasc in sahvary cotininc in relation to passive smoke exposure is Icss than 1'.6 of that in relation to active smokc esposurc. Extrapolating linearly from the 10•fold relat'rve risk of lung cancer in relation to activc smoking would thcrefort predict a relative risk in relation to I+assive smokinj less than I.II, while a quadratic e:trspolation, as wRgested by Doll and Pelo (11978) would predict a lower risk qill. The conflict brtwccn the dose and the elaiirsed raponsc is particularly clear for the results of Hiraysma (1991) who found a sirnilar diect on lung cancer for passive smoking as for active smoking or S cigarettes a day. Second, all the studies suffer from weak exposure data, most pudies only obtaining information on the spouse"s smoking habits and' sionc obtaining objectire data by rncasurernent of ambient kt•+re1s of smoke eonstituents in the air of the home or workplace and/or of concentrations of constituents in body fluids. Tfiird, no studies adequately take into acoountt thc possibility that misclassificalion of active smokers as non-smokers may have consistently biased' rclativc risk estimates upward. Active smokers have a high relative risk or lunj cancer and spouses' smoking habits art: positively oorrelated. Bccause of this, it can be sliown that if a trclativrly small proportion of smokcrs deny tmwking, this results in an apparent eicration in risk of lun8 eanoer in 'non-smokers' married to smokers eomparedd to 'oon-smokers' married to aon-smokers, even when sw Rrt+r dfect of patsive smoking e:ists. A demonstration that this source or sirs is or reat iinportsna can be found in the study of Garfinkcl rr a1: (1985). Based on unvalidated smokina data uken from hospital notes, a relative risk of lung cancer in relation to ausband's smoking at liorne of 1.66 was takulsted„ with relative risks of at kast 1.3 sesn in relation to qch

a,+~,..r.~...+c-.......o..o~~- . PASSIVE SMOKING AND SMOKING•Rf.LATED DISUSCS 18.4 I kvel or husband's cigarette smoking and in rslation to husband's cigar and! pipe smoking When additional sources of information on smoking habits were uscd, the overall retative risk was reduced to a marginally significant 1.31 with an ekvated risk only, rcally discernible in relation to heavy eigarette smoking by the husband. Even here. it is notabk that the elevation in risk wss not evident when smoking data were obtained from the subject or her spouse directly, but was only evident when the data were obtained from the daughter or son or anothcr, informant, ii.e. from those people who were ksc likclv to have known the full smoking history. The lower relative risk may still have srisen wboll) or psrtly as a bias resulting from miscfYssiGntion of smoking habits. Fourth, many of the studres are open to specific criticisms For example, the conclusion of Gillis tr a!: (1984) that male lung cancer deaths in non- smokers rose from 4 per 10.000 in those nott exposed to passive smoke to 13 per 10,000 in those who were exposed was based on a total of only, 6(!) deaths and was not statistically signifrcaM! Also the claim by Knoth n a!. (1983) of a relationship between passive smoking and lung cancer in non- smoking women was based simply on the observation that the proportion of femak rson- smoking lung cancer patients living together with a smoker euoxded the proponion of male smokers as rcponed in the previous microaensus„ignoring intrr ala the fsct that in manyy families womcn hve with more than just their husbands. In the present sludy no significam relationship or passive smoking to lung cancer incidence in lifelong rwn-zrnokers was seen, either in the analyses based on the information, eofkcted in hospital or in subsequent inquiry of the spouses or both It must be pointed out, horevcr, that the number of lung cancer patients who had never smoked was rather small' so that, though our findings are eonsistent with passive smoking having no effcct on lung cancer risk at all, thcy do not eacludc the possibility of a smalli increase in risk„ though the upper 95% confidence limit of 1.50 for the atimate of ©.80 (Table IV)i in ral'ation to the spouse smoking during the whole of the marriage is not consistent with some or the larger increases eJ'ainxd by Hirayama (1981, 198t) Trichopoulos tt d. (1981. 1993) and Correa rr al (1983). Though the number of lung eaneer patients who had never smoked is small, varying around 30-50 depending on the analysis, this number is sot very different from that rel+orted in a number of trilrcr studies, e.g the findrnev of, Correa rt a!. (19g3), were based on only 30. whik those of Tttichopoubs tr a/. (1981), even when updated (Ttichopouios tt al., 1911.1) rrre hased on only 77. The difGctt)ty of obtaining an adequate sample size is uxderGnod when one considers that in our study the 14 never smoking lung cancer patients who completed passive smoking questionnaires in hospital were extracted from a total or 792 lung cancer patients. it would need a very large research efTort to incruse prt+cision substantially, and even then one would have to take care that the magnitude of any biases did not esQOd the magnitude of the effect one was looking for. The two major prospoctiive studies which have so far reported frndings on passive smoking (Hirayama, 1981, .GGarfinkel. 1i981)' were not actually designed to investigate this issue and; as a resu)t, could only use spouse's smoking as an index of exposure. Our study; on the other hand, though not able to monitor exposure oblcctivcly, as would have been prcferablc, was able to look at passive smoking in a wider contest, by, asking about the extent of exposure at homc, at work, during travel and at kisurs. Although the answers to these questions were subjocti'vc, and could have exhibited some bias, their inclusion perhaps allows grtater confidenor in the conclusions. It was intemting that, of the 59 patients for whom spousc's cigarette smoking habits were obtained from both the spouse snd the patients, there were 9(15Y.') patients for whom there was disa$ttement as to whether the Ipousc had' bren a smoker at some time during the marriage:. It seems reasonable to suppose that some of these were in fsct smokers and may have been erroneously elassified' as non-smokers had only one source of inCormation been used. It was also notcworthy that there was quite a stronF correlation inour study between sctivc and passivc smoking As illustrated in Table V1, current smokers were considerably more likely to be exposed to passive smoke exposure at home (from sources other than their own cigarettes) than were never or ex-smokers. As noted abovc, this eorrelalion, coupled with some misclassification of smokers as non-smokcrs, may spuriously inflate the estimate of risk related to passive tanoking. Itis import.nt to carry out further qudies to obtrin more accurate information, on reliability of statements about smoking habits because of this possibility of bias. Little other evidencc is available concerning the relationship between passive smoking and risk of the other smoking-associated diseases in (adult) non-smokers and much of this is open to eriticism. In his original pa=uer, Hirayama (1'981) presented relative risks of dcath for various diseases for non- snwking women according to the husband's smoking habits. Based on a total of 66 desths, a slight positivr trend for emphysema and asthma was twt sijnifscant, whik, based on a total of 406 deaths, tw indication of a trend at all was seen for ischaemic hnrt disease. In a later paper, based on

1M P.N LEfF rr 6il T.Yle VI Relatis ndds of having prawve srnokr o<rowre u home sanrdtnE to s pauent', own manuf'rcturcJ r.prenc smoktnE haMts Ittandardired for sEc barc - axnhrned cl.v I and 2 cvntrcdtt Rctatnr o1d. t9•S', r.afrdrwe Unutsl. O.-n uMd,rrrye Ao/nr. Malr` ftiwolr Never 1 1 f.s 1.25In Rb 1.911 1.26(0W(.1151 Current 4.0012 67~ 5.99) 2 51(1.74-3.62) CM-puared ktr trend tI Jf) 57,81 25.34 r <ot101 <o00t only a further 88' ischaemic hurt discasc dcaths. Hirayama (1984) reported a slight posiuvc trend in risk, but this was not, statistically signifxant: Garland rr al (1985). in a small prospcruvc study• reportcd a LS•fold higher risk of ischjcmic hcart discasc in non.smukmg C.bfurnian, womcn whosc husbands were currcnt or former smokcrs eompared with those whose hushands werc never smokers, but this enormous and implausible relative risk was only significant at the 9ft°-. confidence kvcl and' had ver7 wide confidence limits, bcrnF based on only 2 deaths in women whose husbands wcrc current smokers. Sandlcr rr al. (1985), in a casccontrol study, carried out in North Carohna, rcported a strong relationship bctwccn risk of cancer of all sites and passive smoking This study has bermcriticisrd by Lee (1985) who notes that it is basically implausible that passive smoking should increase risk of cancers not associated with active smoking Lee also criticised the method of analvsis, showing that no association, with canccr risk would be found if a more standard method of analysis was uscd Vandcrbroucke rt o/. (Ii984). bascd on a 25 ycar follow-up of 1,0?0 Amsterdam married ooupks, nxently rcportcd that passive smoking was associated with some decrease in total mortality. Therc is evidcncc indicating that young children whose parents smoke have an excess incidence of tespiratory symptoms and some reduction in pulmonary function. Revrewing this evidence. l.ee (1984) noted that the interpretation of the association is fraught, with difficulties and that other possible t:planations, including social class related factors, parental' tsegekt. nutrition; eross- infection and s+...:i;.: ':.rr.; wrcgrwncy, had not been taken into aoeount adequately, so that a causal effect of passive smoking could not be inferrt+d: The relevance of theae frndinss to chronic bronchitis or other diseases inn adults is in any easc not dcar. Our analyses showed no significant tdTect of passive smoking on lifelong non-smokers as regards risk of chronic bronchttis, ischacmtc heart disease or stroke. In all the nalyses relating the various indoccs of passive smoke exposure to these diseases, no significant difTerences were seen and' slight decreases in nsk were as common as slight increases. Whilt more data would be desirable for thcsc discascs, lung eaneer continues to be the major smoking associated disease for which passive smoking comes under suspicion: Since all the difTiculttcs of carrying out good research have ekarly stilt not yet bren overcome, further research is eertainly, necdcd. Our findings appear consistcnt with the general vievt, based on all the availabac evidcner, that any eRect of passive smoking on risk of lung eanocr or other srnoking•associatcd disc2scs is at most quite small, if it exists at all. The marked increases in risk noted in some studies are more likely to be a result of bias in the study design than of a true effect of passive smoking. Any viers eaprtes.ed in this prpsr are those of the authors and not of any other person or company. This study was fueded by the Tobacco Resorch Council (nor Tobacro Advisory Council), to whom wc anr most Rrateful pr Atderson was the h•older of the Cancer Rryc.rch CampaiEn endoweQ Chair of EptdcmioloFv at the tnstitutc of Cancer Rracarch during the period of the study dearpt and field work. Mr. 1. Marks from Research Surveys of Great Britain provided advKZ in the planning phasr arrd was responsible for the mtcrriewers' vital u+ntnbution to the tpudy We thank thc wuny ebnicians st Ihc 46 particiNttnf Iw.ptali who permitted as to trontan their patients and all the p.tirnts and tpouses who answersd the questions.. Ck R. Wang. who held a antnh Council award for the period 1'9ia 1913. 'as well as a aumber of other oolkagues provided mcfutadvioe at various stagn of the study. Mrs /Jt. IForcy provided in.aliubk aarraanoc in oarrying our the statntitat analyse. t

PASCIVE SMOKING AND SMOKINC',•RF.LATff7 DISTASfS 10 Rete.e.ea ALDP:RSAN. M R.. LLt_ P N. 1 WANf;4 R(19R5). RrskiOf lung oncrr, drcunK Dronchttrs. tschaernK htart disryse and turoke in relYtwn to lyrc of c.prcttr smoked l. t pairm, Cowrn NGA „7!, 296 RRESLOM N:E a DAY N.C (19E0) SrosLrtfni k/Nkodr w C.owR-r Rrsrorrk t qf I- TRr Anofi-su of Cav-fnrtrol SrrdKS International AFaK7 for Research on Canccr; Lyon •UFFLER. P A_ PICKLE, L W.. MASON. TJ ! CONTANT, . C. (1981) Thc ouses of lung eanar in Tesas M C.we Cancer Caavs and Pnt•.Mr«.n, Mvsll. M i Corres. / P. (eds) Vl.1aF Chcmie International Inc . CHAN;..M•C (1982). Z.hkn aw HonFkont. A/rwrA. lfrd t M'ork, 1X 16. fORREA, P.. P1CI:LE. L W., FO*TNAM. E, LIN, Y. i . HAENSZEL. W (198)). Pasw.c smoking and lung 5 OnCet. LanfTt, U,.395. ~ DOLL R t PETO. R(197E) Gprelte wmokin8 and s t bronchial orcanoma dose and timc rdalionshiqs amont rcEular smokers and lifelong non.snwken. J. E4hdrnr Cow.n. Mlrk. 22: 30?. GARFINKEL, L. (1981) Ttmc trends in lim8 Onoer monahty among non-smoken and a oote on p.aive smoking l.,Arot! Conrn. but., ib, 1061, GARFINKEL, L., AUERRACH, O. A lOUlERT, L(1985) Involuntary smoking and' lung onoer. A essesontrol qudy, J. /V.r! Cmrrr+ /+su.. 75. 463. GARLAND„ C., DARRETT-CONNOR., E. SUARE2. L. CRIQUI. MH A WINGARD. D.L (1985); Effects of passive smoking on ifehemic k~cart dtseasc tswrtaUty of tson-snwkcrs A prosrecti.e study. Awrrr: J. 12f, 64'5. GILLIS. C R, HOLE. DJ.. HAMTHORNL: V.KI L pYLE. P. (191U) The effect of environmentali toEacco srnokc in two urban oommunitles in the west of Seotland. fvrnp. J'. Ittsf Dts.. iS. (SuM+l 133). 121.. HIRAYAMA. T(1981) NonamoLGnF wives of heavy pnoken hane a MEher risk of, lung tanoer- a atudy from lapan. Ir. ilhrd J..1:2, 193 HPRAYAAtA. T('1981) Lung nntxr in Japan effeirts of nutrition and passive smoltnE Iln lyne Cwrn, Caxv. and L..enrwwn, Mirrt1.,M & Correar, P. (R1a) Verlag Chemx Internultonal Inc HUGOD, C. HAx•KINS. ILH 1 A.~RItP, T(197Jt1 Espusurc of pautve smoken 1'n tr.h.acr, sms,kc oonstttucnts Mt. Arck' ChW . FArNan IIhA. 42. 21. JARVIS. M l RUS.CLI.L. IA A I/:. FEYF.RARfNp, C A• othen (1985) Panivic e.f+osvre to tuol?aceo .mokc plhra cottnine CarKlCnlrfUonsin~ a R(+RlfntaUvC Pnfwhtion tamrlc of aon-snwktnp tMc+olthddrrn. Ar Med l.. 291, 927. KABAT, GC • M'YNDL'R. E.L (191t1) Lung urno" in .K+n-smoken Cowrr. S3. 1121• KNOTH. A., KtHN, H & SCHMIDT„ F. (19R3) Pauivc nnokinF as cause of lung canoer in fcmak oom eoken. klrd A/rr. 71, 54 KOO.I.C., HO, 1N•C a SAW. D(191/i); Is fusaive smnktnF sn addod risk futnr for IunF onoer m Chmex .romcn+ J. Eslr. Cwr Conrrr Rr, . 2. 277. LEE. PN (19111) Pasurc SmokinF In Swrntr+F and rkr l.nR CumminE. G dc RonsiEnorc, G. (eds) Plenum PuNishinE Corf,oratKm LEE. P.N (1985) Lifettme passive stnoLinS and onccr risk Lwcn: L 144 LEHNERT, G.. GARFINKEL. L.. HIRAYAMA. T a4 othen. (19M) Round tabk discussion. /,rr. Nfrd:. 1X 7)0: SANDLER, D P.. WILCOx. A l a EVERSON, R R ('J9N5) Cumul.tive effects of lifetime smoking on csnon tisk Vncer. L 312 TRICHOPOULOS. D.. KALANDI!)1, A.. SPARROS. L & • M.rMAHON, R(1981) Lung onoer and Pamvc smoking M1. J. Caser7. 27, 1. TRICHOPOULOS., D, KALANDIDI., A& SPARROS, IL (1983) Lung cnnar, and, passivc nnoking Conclusion of Greek study Lan, rr. ii. 677. VANDERdROUCKE, ll.. VERH[LSEN, l.tl H.. DE BRUIN, A.. MAURIT7. RJ VAN DER HL1Rf•WESSCL. C t VAN DER Hl-.IDL. R M (19n1) ACtuvc and pasaive pnoktn8 in marrxd awPles results of 25 ytar follor tap !•. klyd J.. 2><It, I lla 1