Tobacco Documents Online

G. a. C. u. r a.. ! C o+. poe. 1.11- -' 1 Passive smoking and diet in the etiology of lung cancer among non-smokers Anna Kaiandidi, Klcl Kztsoupanni, Nelly Voropoulou, George But~u, Rodolfo Saracci, and Dimitrios ?richopoulbs (Receined 4 Apri11990; nceived in revited jorns 1 May 1990; .ucepred 3 May 1990) A -aol ancr waU anderrJlta in AtMas m aoore the eds of posc'rve smo7anj and dier in the e~uaacoa of kna= aacei. by bistoloyic type. in aon-umokiag womm. Amoas 16° .roensn wirb lua; -aaer admicrd ro oos off ae.aa major baspirsL in Geeusr 1dnem bs,...n 1997 aod 19119, 1S4.rere mta.iewed in penont of dcse iaiesriewed. 91 wen Gfe-bnr nos-bmohcs Among 160 ideod{ed eoounls wicN $r=uars at othu oethopedic epndirions.14S Mar inoer.is.ed in penant aP mow iam.ien.d. 12° wsss Gfe-lon= aam-.mokets Azziapoi a imoi~fos:d, womaato.a.xbi~bd.e s mlctarloeianeaofzI (9sOr.e°o6deac4iorawllQly 1-t.1j:_ ' smu8'__ ~~ . , ., . . aa~ br'a~. Bmh.od aod liba of ioipa:e o° b~ri s~n= . ,~liaarlr e~eoqc ,i~::ILvs..a. eo a.ideam of soT asrodation .ith amosnee to smeiont af orLrr honsehold membra, aatd tbs asmciation with expaore w pas4.v smokiq= ac wotit was :mall and' aa samoolly sifaionat. Diooey dsra taZkead taron=n a aani-qeaeuititi.e food-fssyueaeyy qoetaoaoaiee amdie.ud thx bi:h oomamption af fnWa rru ineaneyrebted m tAe a6k oE4Wt aoaes (0s miadre cpk baweeo aueme qmetiil/s w=017 (CI 0.10-0.74)). Neitba reSenbi.a nor aor orher food:roup had an additioaal pcoteetivs dfee: fnrthcmuorc, tbs appaxnt procati.e sfi~a'[ of vqsuble..at nor doe co cuooeooid rinmlo A oaoxae and wss onlr, pas1y e:piriaai in naas of riamin C. I3e aaoaaaom of 1un= noar mic wirh papiva . .,,, .... amoiciag aad ndvoed fiuil imkt .ae iedeptodeai and dd na coafoaad neA odrcs. PaKi.t smoEin= .:ai .omnd'widi an iatswe of t0s of aII hisroloyc e"M of omoer,, ahboa06 eha .l~+rar saoee m°dss For admomcin=. +`> Kry urosdr: Lung cancer. passive smoking, diez, air poltution, vitamin A, vitamin C. euocenc. Inuodrlcdon The association betweca it:volurlrary exposure to tobacco amoke and lung cancer was fitst rcported in 1981;" by the end of 1989 more than 20 epidemiologic snldies had eMmined this issue.;-10 In about one-third of these Rud'les there were statisvcally significant positive usociadons between passive smoking and lung cancer as1c. .rherrst in another third' there were positive but sratistically non-sigaifiaat relacionst'lips, and in the remaiaiag third the associations were mint~lal or pon- =stesu. Overill the associauotl betweea passive snokiag and lung aacer is highly sigaiScult and. for practical purposes, chance can be elccluded' as a possible acp2am don+ On the basis of biologic plausibility and cpidemi- ologic evid'eace, aasaiity appears the mosr likely exphnation of the empirical association, but this view has rlot been uniformly accepted. It has been suggested -t5at the association may reflect misdassifiacion of eY-smokers among non-smokers (cx•salokets arc at increased rislc of lung cancer and are more likeiy to be malried to smokers relative to life-long non-smokers), or to confounding effcct(s) of unspecified fictor(s). One such factor could be nutrition (e.g., nutrition poor in crotenoid sources of aitarain A) since passive smokc:s may be less healtb-conseious than non-csposed peaons. The pnsenc study was undertaken to examine the role of d'ier and pusivc mmoking in the ausuiorl of lung cancer in non-smokers, by histologic type. Special emphasis was given to the exclusion of ex-smokers from any ulalysis concerning lifelong non-smokas.. M2teaals and methods All women hospitalized during an 18-month period (1987 = 89) in scven hospitals of the Greater Adu:ns area Dir K.uCs+d'tafi, KauaL7 mri, Voropoylos. r1 Boarr ms ia tAs Drpsrn.at of F7Jgie.t osd Fpl4-uioJom. UrEwrriiy of Asknsr AG1rc.! St.iool. AAbe.r, Gireu. LL.Lmss'irot rir 1L,~A'3.waX fo,Rsrrxd oo. Cas+ov. Lyw.. Fraeci: D, TircwoparlAt ti ii rb, Dypsrwesr ofEPidc+R+olel7• tl"avd Scioof of Psilu 1i'sslsi. Rrpnnt mqrru,l6orld i6 rnr ro Gr Tric&opoaJoJ ,t 673' FlrmASYron eissaw., Borrow, 9Ll 02111, lJ3A. Tie nw.reb wr rrppn.a,rlry t,i.l.u.nuro,.d Agrncj /6. Reuanb o: Csesr --ad ree Co..Aairrios of R.be &emp..n. C.Owxrsirrsr. It it par ol rlu ELrBOP.4SS'collaoorws, ZEC pro*u. e l970 ,tapd' co+emunMaC;ow or odorJ Ld 15

A. KJmrduh'a et al. with a de:iaitc diagnosis of lung cancer formed the ca.se series. The hotpitals included a!i three cancer hospitals in this uea, the only hospiral fnr chest diseascs in Acbns, and the diree largesc ucuvessuy genml hospitzls. Women Wae included when chere was a positive histologic or eytologir esansiaaaon or when bronchoscopy was con- sidered diagnostic of prirnary bronrhogcnic euonoma. A total of 160 cases were idetziried. Conrrols were 160 women hospicalixed in the orthopedic depurrneats of the samr Txspizzla or the nearby hospital for orthopedic disordez, to which most accident cases from Greatcr Athens and the surrounding ares aze admitted. Controls wce randomly selected from those admitted within a week afte+ the identification of a corresponding case and had to be 35 yeats of age or over. Among the concrol' women, 102 had fraetures and' rhe reaiaiaisig 58 had' other truuaatic or orthopedic condidons. AU cses and controls svc= interviewed in person in the hosaiml wards, as soon as a definire diagnosis was esralalisised., by one of five interviewers who eac.h inter- viewed the same proportion of cases and controls. There were no rmiuals unong cases but six were too ill to be- intcvieared. Among contcols, 12 wese in a condition tliat did not permit interview, and thrce rcused to partici- pate. In the uuerviews, patients were asked co indiczce in dctail their lifelong smoking historics. their exposure to passive smoking-from their husbands, from otherr household members and at work-as well as a number of other demographic. socioeconomic, and mediczl chuactrrisdcs. Subjects were also asked' to estimute the avetage frequency of coasumption (per monrh, per week. or per day), before the onset of the preseat disesse, of 47 food iacros or beverage categories. These iteans were selected from an extensive list of 120 items, using the aizerion that the xiectcd items should cover, colleccvdy; more than 3056 of die inrske of e.uh of the energy-gr.us- uing nutrients as well as of vitasa,in A. This ctiterion was established on iaformarion from control groups in a number of case-control studies undertaken in Athens to explore the mIc of diet in the causation of cancer u rutous sices. t L l.ifetime tzposure to air pollution was controlkd in the analysis on the basis of inforiazrion about the lifelong residmdzl and employment ad'dmsey of all subjerss. The srczs of residcace and work were divided into five categories according to their estimated outdoor air-pollu- tion levels. For the Greater At}Kns area. sir•pollurioa levels by borough were cakuiucd on the basis of the meaa yearly mcaslurmenta (1983 - 85) of smoke and N00.. as recorded in 14 monitoring stations dispersed throughout the area. A line for zero air pollution awss dr3wn at the highest points of the surrounding moun- taisu. For esch borough, the calculsted zir-pollution level was the avesagc of the measurementx of the three neamt 16 stations. or the cwo nearest and the zero air-pollucion Line, weighted by the inverse of the distaace from the borough's center to the melsurcmeac poinrs. Boroughs were then dividcd into four categories, catcgory 4 beitig the most polluted with dailysmoke values frequently in eseers of 400 µg/ my and racegory I being the Icss polluced wirh daily peak-smoke valucs ruely exceeding 100 µg/m3. Past residences in niral or seaYi-urban areas (population less than 10;000) were considered as category 0, whereas past residences in other cities of Greece were cilssified in categories I or 2 according to the recorded or presumed lcaels of ait pollution, Fiaslly, for every individual, a time-wcighted sum was cahulared assuming 40 hours per week working tinme for individuals woridag outside the home. For housewives, their home residence formed rhe only basis for estimating theiu air- polludon exposure. When all subjects had their air- pollution exposure usessed, they were distributed into four groups based on the margii~i' auartiles of the air pollution index distribution. Since it is possible that the hospital cxrchrncnt areas are Larger for cancer patients than for patients with minor fractures and tssumas, the possible air-pollution aysoczarions were taken into scoauat only in order to control for possible conioundiag (generated; by either genuine causal effects or through seleccion forces), and not for assessment of causality. Among the 154 casas. 91 had been life-long non- smokers (less tiian 100 cigarettes in their l"u`ezime); among thurs, 44 we:e diagnosed histologically (48%), 34 cytologically (38%), and 13 (14%) t,b.rough broncho- scopy. Among the 145 controls, 120 had been lffe•long non-smokers. The analysis was confined to l'u"e-long non- smokers. Three sources of passive smoking examiacd in the prescnt study we.^c: husband's smoking; smoking of oth.er household mcmbers; and exposure to smoking at worJt. Exposure to husband's smoking was considered eo starc ac the tisue of marriage or when the husband srarzed smoking (wFsichever came second) and to end.vhen the husband stopped smoking or died, or the couple separated (tvhictiever came first). Change of busband ans considered equivaknt to change in husband's sraoking habits, .vheras single women were considered as unecposed to husband's smoking. Ycass of esposure to husband's smoking and average numbcz of cigarettes smoked daily by the busband were separatcly examined in the analysis. Exposure to the smoking ofhousehold members other than the husband was astesscd' by multiplying the reaz< a woman livcd in cach of her homes throughout her life. with the number of smokers in the corresponding homc (excluding the husbznd) and by summing chcse prrduc• rerms. Subsequently, 211 women were diScributed iarc four groups: one conrsining those who had never bet: 2023382529

Lung causr amoag swn•snro,Errt etposed to passive smokiag from rnembca of their houschold; and three corrsponding to the terciles of k==,ing household cupoaue. Finally, ecposure to passive sraoking in the workplace ,wn calculmd u the rime-arcighced sum of exposure to smoking at work. the expowrc beitlg bssed on the nuraber of smokers among people working in the Slme dOJCd spice.. From the scmi-q*'= *3"nvr food-frequency quesdon- naire. energy intake snd inakes of vitamin C snd of v'tt7:air1 A and its cotutitucus (Ieuotene and rerinol) were esrimated' by multiplying the nurrierlc content of the selected portion size for each specified food item. by the frcquency that the food was tonsumed, and adding these estimsta for al1 food itaas. Food-composuion dzta were based prin>adly on values obrxiacd.&om the University of massachusetrs Nutrient Dat2 Base.tl Analyses were also undertaken in order to ecmine food consumption (tntile: than nurriau intakes). Subjccts were distributed into marginal quutiles by total frequency of consvrnption of food itens belcnging tc specific food groups (s.g., meats. ftuits, and vegetables). Multiple logistic-regression models we:e used for the ststisticni examinzuon and summarization of the data. In the analyses. a core model was used' which included age (as a cate;orical' variable in ten-year groups), years of schooling (quantitasively), and iintesviewrcr (four indicator vsriables). All confidence intervals shown are 95 °e i;ntervals. Analyses were done using the GL statisrical package (Numerical Algoriuusu Group Inc., Rede2se 3. 1978). Tabi.1. Dismbudon of 91 noa-smokint ,ornen ..izb dsmg nncer and 1?0 oon-smolan8 campaawn women by sda'std demoeraphic charactnums (lriammta in patentboex) Chux=ristic C.asa ConunL P Age <s0 ymn 1} (16.}) U1 (14.2) 30-391'on 18 (19.8) 22 (14.3) Q.36 60-69 3an 27 (29.7) 31 (23.1) 70• Yacs 31(34.1) 50 (41.7) Sc6noCq <I pear 18(19.8) 27 (22.7) 1-6 yean 53 (38-2) 72 (60.0) 0.42 7. Years 20(22.0) 21(17.3) Curreat saidence Gnaur Aduas 48(52.7)67 (37.8) Or3iu urbaa 11 (12.,1) 9 (7.5) Semiiuosn 7 (7.7) 9 (7.3) 0.99 flursl 25 (27.5) 3$(29.2) Oaupacon E.W employed 67 (73.6) 90 (66.7) Hauew;ir 24(26.4) 40 (33.3) 0.88 Stsrinl snrus E+er maaied 83 (91.2) 109 (90.8) si:tt{e S (t.y) 11 (9.2) 0.99 'P nlue foa linear atnd. Table2. Dum'bucon af 91 nen.imekin`.omm.irL lung ancQ and 120 non-smolcm8 comp.ri- woman by aieaed puamscen of upoautt ro pastive smoltia8(per+easaga ia puendhests) Quncn=Eic Results Table I shows the distribution of cases aad controls by selected demographic charscteasria. Therc are no signifrsat differences with respect to zge, years of schoollne, cnnear residence and oa:uparion, evea though Husband's smoking Gprem/4sy neveruno)xd 1-20 21-40 41* these variables were controlled for in subsequent multi- ~'"Q10k"`e tatiate analyses. Table 2 shows the distribution of t~ses D"°"0° °f °p°"1fO ae.a smoked aqd controls by selected parameaca of esqnsure to passve <20 yeao ~rtokin '. . . . &• .,~.~~ ~ l~ni 20-29 }rao ng wtrh ~ ash., but the ~* 3~ dfffereaces are nor large enough to be iacerpretabl~e _. wirhout controlling for confounding effecrs. Tsb1e 3'bd,,,, ym-haw "exposote Compares the distribution of cases and controls by lifdong Now ecposure to outdoor air pollution. The two diszribarioes Lo. (la oeruk) are almost identical. Finally. in Table 4 the disuibutioo 1edium Of oses snd controls by frequency of consumption of HO (3°d-Cile) specified fexx! zroups and nutrients is pnsenced. The:e is no clear ur suggcstive difference between cases arld controls with respect to any of the indicated nutritioual variablcs, except for cereats (P . 0.04) and fruits (P . 0.11). The association with cereals is positive but is not Esposum a work Houuwift E/(Airml somt biobgically c rediblc. is not stlpporued in the literaaue. IP.aluc for linar eiend. Cases Concrolt P 26 (2a.9) 46 (39.7) 34 (37.s) 39(33.6) 0.16 22 (24.4) 22 (19.0) a (A9) 9 (7.8) 26 (2i.9) 46(39.0) ts (16.7) 21 (17.5) 15 (16.7) 20 (16.9) 0.07 17 (1t.9) 15(12.7) 17 (18.9) 16 (13.6) © CS (16.7) 24 (22.0) C,.i Cj 29 (32.2) 26 (22.0) 0:60 GO 24 (26.7) 27 (22.9) 22 (24 4) 0) 39 (33 . 24 (27.0) . 40 (33.9) Ul W 0 52 (31.3) 68 (37.6) 0.13 13 (14.6) 10 (8.s) 17

A. JCalsardid% cr al. Tabk 3. Dirribudua of 91 aao-anokiAj .wmen with luni aaar and 12D non-smo)oag totnpaown women by indrt of lofcien8 -posurs m «udoot air poAurion (pementaga in Pu'a+th-) Ait pollutioa iodex Gua Conuots Lt Quorsilr, wery low 32 (33.2) <3 (35.8) 2ad quardk: lo.. 20 (22.0) 26 (21,7) 3rd quaetik: modenre 18 (19.8) 22 (18.3) Ath quara7ea higii 21(25.1) 29 (24.2) P for liaar aead - 0.99. Tzbk 4. D'uvibuaaa of 91 ooa-smokin8 woiaen xiut lun; noeer sad 1M oon.aaokia8 cosapan.on.rogncn b+sppemda= aaprn!'q-nites or mriIa of' frcqueney of consumpsion of sQeeisied food 8ton- ead ouaientx (peteenn;p in PutnrteeasS) Food sroup Quutibts Pfor of aurriear 1 (LOW) 2 3 4 (H+gA) uncaf asad. Cecsslt asa 27 (29.') 16 (17.6) 34 (37.4) 14 (33.4) coacsnk di (35.8) 34 (28.3) 33 (27.3) 10 (8.3) 0.04 Boncoey ' axs 7 (1:1) 1 8 (19.8) 33 (58.2) 13 (14.3) wntrois 1~1 (11.7) 27 (': ,5) 57 (47.3) 22 (18.3) 0.57 sussa esus 28 (90.8) 24 (26.4) 26 (28.6) 13 l,14.3) conaols :4 (36.7) 30 (25.0) 31 (23.1) 15 (12.3) 0.40 Fultss uact 50 (54.9) 36(39.6) 5 (5.5) soauoL 91(67.5) ' 29 (24.2); 10 (8.3) 0.26 Vegesnb.let cam 27 (29.7) 22 (24.2) U(19.8) .24 (26.4) sooao)s 34 (28.3) 36 (30.0) 29 (2<.2) 21 (17.5) 0.44 rhan wa 35(36.5) ', l9 (20.9) 1'3 (16.5) 22 (24.2) <onvois :2 (1i.)) 44 (36.7) 24 (2A0) 30 (23.0) 0.11 IGaa, S.h..m ma 26 (28.6) 23 (25.3) ' 21 (23.1} 21(23.1) e'ofitmli )9 (32.3) 27 (22.5) 31 (2s.8) 23 (19.2) 0.57 3./SIk and au7k paoduts cam 21 (23.1) 29 (31.9) 16 (17.6) 23 (27.7) toorrois 27(22.5) 32 (26.7) 30 (25.0) 31 (23.8) 0.76 fsrs and oils cam 31(27.5) 24 (26.4) 33 (36.3) 9 (9.9) coavoli 35 (29.2) 30 (25.0)' 33 (27.5) 22 (12.3) 0.66 cob uses 70 (76.9) 10 (11.0) lt (12.1) mnrmh 97 (79.2) 16 (13.3) 9 (7.3) a44 Tab1e t masieuud Food Qvup Quasn'ks P for ot auaKnr 1 (L.) 2 3 4 (F4~igh) UAM aend Orfic aoasltCJfolie bevaates csses 44 (48.4) 30(33.0) 17 (18.7) mnaoil 69 (57:3) 26 (:1.7) 25 (20.3) 0.52 V'rmmin A cam 23 (25.3) 21 (2'3.1) 20 (22.0) 2,(29.7) aoouok 30 (23.0) 32 (26.7) 32 (_6.7) 26 (21.7) 0.48 leticd (Frefonrud) csus 22 (24.:) 23 (25.3) 21 (23.1) 23 (27.5) ooaavk 30(25.0) 30 (25.0) 12(26.7) 28 (23.3) 0.72 A,marocene eaxs 25 (2?.5) 19 (20.9); 20 (22.0) 27 (29.7), coaaois 28 (23.3) 93 (27.5) 33 (27.1) 26 (21.7) 0.68 Vhnmi4 C cam 30 (33.0) 16 (17.6) 23 (25.3) 22 (24.1) tpnuois » (1a.3) 39 (31.7), 29 (24.2) 31 (25,8) 0.30 Tasl eerrYy casa 23 (25.3); 25 (27.3) 18 (19.8) 25 (27.5) tonuols 30 (25.0) 27 (22.3) 35 (29.2) 28 (23.3) 0.93 is not parricuhrly marked. and may well be acphtined by the multiplicicy of comparisons made; it was not futther explored. By conrrast, the negarive asaoci.uion with consump'uon of fruits is biologirslly credible given their high content of vitamin C znd some carocenoids, and has beea foaad in mzay oiher mldies. Among the non-usociauons, cozzl energy indce deserves partimilu attention bersuse it indicaro Cb1r, in quanricarive cOmos. dere is no diffcroausl dierery reporung between cam and coaaals. The auociauon of lung cancer with exposure m pusive smcking through _marriage to smoking husbands was further -xlined' by mukiple Dogisac negression, controlling for age, yeus of schooling and interviewer. (~1u rdacive ~~~8&~ conassria~wnrnea anstiied _- msrriediw on=smoGawm 1.92 .vafs 6ilpbe:: _ ~ 4C~9. _Tbe eff= of the avera;e number of W he tuisband and rhe dursrion il b ci k d d y y t garecoes smo a e of marriage to sraoliers ovcre eval=ced in tae diffcrear ~ models, controlling for ches:me core vz~dlbles zs:bove ~ and introducing, shernaiively, the daiiy number of ~ W ciprccta and the duracion of aoarriage as quantitative tezms inreractiag with the smoking stuus of the husband ~ «la Ihe.luag ess~cer asic smoag no~a~ (tlie 'rau• device) . a 4 ~ 2 a ,e= of~ . to t~~isla"- soktng snd bp ~~(i foreveiy' addicioelal ~ ~ack of cig2iettcs sm ohed daily. These estitaaces ue low sad siatisticaJlv non=signiflani - probably bcnuse. wbile is

sz 42 7: fla I; ed iot •an en is, he l:u s. ;es :ve ss .n, cr. to :th of on .tu -ve of Ve ad a• ire ta! )w i1C the smoking status of the htubaad is validly sscerr~incd. the quastIIrsave aspects of the exposure are difficulr to ~ss accurxtely. Simiiar models were used to assess the effecrs of c,cpoaure to tobacco smoking by other househoid mem- bers or at the workplace. The reslelcs were qualiativdy similar to those presented in Table 2. There was no evidexe of any e$':xt from exposure to smoking of other household a»sbers, whereas the effecr of exposure to passive smoking at work was very small and not srarisc- ically signi6cnnr (the AR between ottrcrae quartdes was 1.08 (0.24 -4'.87)), Controiling for air pollution bad no effetx on any of the rrlulrivaziate anal'yses. Table 5 shows multipie logistic regrtshion-derived RR for lung cancer among non-smokers between eztremc quartiles of selected food groups and autrienu. The RR estirnstes are adjitsted for ape, years of schooling, interviewer, and total' energy iatake. There is little, if any. conioundiag between rhe two indicated food groups (vegetables and ituics) or among the three indieared nutrients. The.ore, the data suggest that fruits, but not vcgetlblts, protect against lung cancer, and tiLarvitzmin C alone cnnnor esplaia all or most of the protective t$ect associated witil fruit cosuumption. There is also evidence that retinol (preformed' vitamin A), far irom being procective, may actually be associated with an increased' risk for lung cancer in this series. Finally, esposure to husband's tobacco smoking, aa& fruit consumpuon. wcrr simultaneously introduced in another model (together with age, ,vears of schooling. interviewe:, and total energy intake) to cxplorc whether the passive smoking effea is confounde4 by inackquate ;=Llce of fruics, and vice versa. There is no such evidence. In &ct, the RR associated with exposures to husband's tobacco smokiag inc=sed from 1.92 to 2.11 and the relative risk associated with high al. low consumption of ftuits detreucd from 0.33 to 0.27. linaoduction of cetehts to the last model iud no effect wiuress the study was Table S. Mulriple lorytie eegftssi;orsdeYisad tr!at;ve rislt iot iuof - sewn= ooa-SMoken benreea e=eme qsardks of elowd ~od'$loups or nrasiena Food jroup or wutrieat Relsti.e ridt bet.een actteme qwrulb' Cootidarcs inmaal• P wlus J!-Carotene 1.oL o.G4 - 1.3a o.96 9=4W9 cPmlot~l 1.31 0AM- 1.." 0.06 Vitamin C 0.67 0.42 -1'.03 0.0! Veteabks 1.09 0.44-:.df 0.36 Fnuis 0.33 0.13 - 0.36 0.02 lrrng cmusr arnong son-neso.ltn wu aulall LU illuw mccruingful A.rsasacua uf iuLCi+.uve effects, if any. The lzst model was also applied sepuarely for adeno- eueZnomzs, on one hand, and for squamous, sraall- and lsrge-cell csrdnomas, on the other, using in both in- staaces the total set of controls. The results uc shown in Table 6. It aapezrs that the e$ects of passive smoking are more evident for squamous small- and large-ce'.f car- cinomas taken together. than for adenocarcinomss. although the diffe•.ence is not statistically signiucsnt. On the contrary, the nutritional factor(s) associated with fruit consumption appear to be eqesaliy strong in both groups.. Discussion Three major reports have coaeluded chat ciwe existing daa staoogly support a causal relation between passive smoking and lung cancer. 11 - eb There have been 160 more than 10 epidemiologic studies assessing the role of nutrition in the etiology of lung eaaee:. In a critical review, Willctr' summarized the evidence as being reltSarklbl4 consistent in suggesting an 1r1vG'SG 1iSOU2t1pn bc:weca carotenoid sources of vitamin A and the risk of the disease. Since bonc espostuc to passive aaoking. and a dier poor in fruit and'vegetablesd may re$ecr inadeqture health eduezezon, it is conceivable•that each of the two factors couid' confound the relation of the other to the risk of lung cznce.. The resent stn4Y suggests that this is nsche n5e;,~th {cct of ~assive smokuLg snd diet zppezr`w be ind'ependent' l~CSidual confounding on the basis of a conceivable association between husband's smoking of high tar (rather thaa low tar) cigarettes and iaadeqture fniit intake by his wife is unlikely, because high- and low.car ciYarcaes confer siailar exponues in the context of passive smoking (tar intake depends primacily on the fslrer used). Furthermore. the special cfforr to eor.lude ex•smokca from the study of lung canca among non-smokers provides assurancc rfsar the rrsults Tabk 6. i1CltiplC* b&ciC.:R9=ioededred duii~e r& (9yY. waude= inta.alt) for 1unR nncerby 6ismbf ic type amon= non. staoiers aaording to baaband'a tobocco smoklng sasus- and ro higa st. loa quarsile of ituit toeenumptiona Hismfo`ical r~ Hvsband' Ynoka Ftuix ccnsumption vt. non-tmotoer high rr. low quuriJe .4ll',lung esncz: '14i#04-4.001 i 0.217 (0.10-0.74) Ade•aoardaornas :.0i (o69tl-d.2d) G.,, (0.G7-0.73) sqwraow. sm.l1, 1 . largrceu 2.St (0.88 -'.77) 0.20 (0.04 - 1.36) 'Canuollur; riir a1e. yesra ofxlroo{ing, inrervre.Kr. snd totai ea"y {oermlling for sge. years ofscSooGnr. interviever. snd tonl encrp innke. ineake. b`~S%. "For L3 of the 4L cases histoloKic t.rpe waa not sva4-4bic.. 19

A. K•Jlaxdidi et a1. of passive smoking do not rtflcct bias generated from misclassifitacioa of ex-smokcs among the non-smokcrs. The present study hu advantages and disadvantages. A11 women were interviewed in pe=n by medically qualified interviewers in the hospital wards; there were .ery few nfusals, and most of the lung cancer tues were histologirslly or cytologically confirmed. However, the study is a hospical-bastd' case-conrrol investigation, and the possibility of selection bizs cannot be excluded. Nevcrthd'ess. the patticipuing hospizals admit morc than one-half of the lung cancer czses and accident pauenta hospitalited in the Grr•atcr Athens arez; rsses and controls were very similar with respect to the demographic and socio-economic characteristics examined: the RR linking sccive smoking to lung cancer (3.3) in women is very similu to the couesoonding estimates derived from ocher lsugcz studics: ts and the air-pollution findings in the present study, suggesting chac there is no association between air pollution and lung cancer risk, are in line with the collective evidence in the international liccrsturt. cv =° The appropriateness of using patients with fracmres and other traurn.uic conditions as cont:ols may be questioned on the basis of a postulaced associ- auon between diet and osceoporosu and the well-known link between alcohol intake and risk of accid=t. The link between diec and osteoporosis is tenuous, however, and there is no evidence that increased intake of fsuit or vegetables can Ic•ad to osteoporosis or increase the probability of a traumaric lesion through another mechani-, The•.e is adso no r+eason to bdieve thu alcohol intake deserves importznt coasider.uion in studies of mostly elderly Greelc women, and there is no obvious link between high slcohol' intake on the one hand, and low eupostue to passive smoking or high intake of fiuirs. on the other. ?b.e association between diet and lung cancer has been exhausrively and critically reviewed by CVilltrr. 1T It appears that physiologic considerari'ons and some results fram animal studies hxve pointed to vitamin A as a possible protective faccor against several turaors, in parucular lung ancer. Most of the epidemiologic studies, however, provide no evidence rhat re=inol plays an important role in the etiology of lung ancer, whereas the majority of chese studies undicated chst a high incalre of fruirs and veYeabla is inversely related to the occurrence of the duease.17 It has been assumed ttutcally that the etfecr of trutct 3nd vegetables could be ecpl- ned through their high content of carocenoids, some of which are eventually converted into vitamin A. Howevcr, earotenoids cuv have other acuons that are not shared by vitamin A - notsbly quenctting singlez oxygcn and free radicals chat could otherwise iniciare harmful biochemical reactions like lipid peroxidarion.21 Gradu- ally„ a consensus has begun to develop th=r carotenolds (and in particular, beta-carotene) are impoaant themselves rather ehan as precursors of vitamin f Although this view is certainly compatible with tl cmpiricsl evidence whichsrrongly suggests thar fiuia u veEenbles procect against lung atncer, it is not the on credible hypothesis. At Willea" and ochez auciioa ha. wamed. it is possible char other components of certi fruits and' vegetables, such as earorenoids unrelated E vitamin A, or indole compounds,u are the procecti~ faaor(s). The f:ndin;s of the present study point towu( this alte:nacive incr.precation of the oversll evidenc. They indicare that fruirs, rather than ve;cmbles, sre tl imporunt protecrivc factors and that their effect, if rea is medtated neither through betl-carotene. nor cxdu ivelv through vitamin C. !t is possible. of course, that the results of this stud; noL'w1tllsrOSlding their stal3sLtcal'siqniflClnce. are d33e t chance. It is also true thac:he ucilized'semi-quancicacin food-frequency questionnaire was limited to only 47 foc ittzns and char the available auuicnt database was nc spediscally developed to represent the Greek dieru~ . inakes. iYcverthdess. both issues cannot readily ocphi the partern of associarions sern in Table 5. Furcheaaorc these findirtgs are compatible with the relatively to• incid'ence of lung cancer in the Greek population - popuLzcion with the highest per rspira tobacco car surnpuon in the world, but with a very high &u. consuraation as well. Z; The rindings of this study with respect to passic smoking and histologic type are compatible with findinr preriously rcportcd. It is now accepted that acuve smoY ing csuses all histologic forms of lung cancer buc ris: the s.ssociarion is weaker for ad'caoc2=oma.24 pauiv smoking has been linked to adenocarcinomaa•n as we. as to other fouus of lung csncer,'•' but the associirio has been demonstrazed less consistently for th formcs.'•=' On the concruy, the dirrsry amociado noted in the presenr study does noc appear to dcpen- on histologic type, in line wicn the evidence emergln from othcr.$ though not all'.~r previous studies. Rcfcrea.cGs 2023382533 t. Trichopoulat D. Kslandidi A. Sparos L. 3fac_'JsRoa P Lung czncer and psssive uooking- WJ Csxes.1981: r 1-4. 7 WimysTa T HAft.VRlflllflg wives of lfC3vy sQifJCCfShl't a higixr risk of lung cancer. a study fiom Japan. Bs ale. 11931; 282: r83-S. 3. Trichopoulos D. Patsive smoking and lung cancer. Th lDscn Cecture 1997. Sesnd j Soc Ai.d 3988; 16: 75 - 9 4. Saracai R. Riboll E. Passi.re smoking and' huig cance: cutrenc erideace and ongoing srudia ac cbe lnxrnacion% AYtney for Research on Cancer. bfatstios Rsumc6 1985 22: 117-27. S. Lnm TH. I:ung ITM. Wang CM. er a Smoking. passis

Lrxg namer serorrg aoA•rnoRSrs Ip.. to isc od tot aty ain ac, ow -a :in- 331t :ive ags .olo- :hat sive ndl ion the tion end tinS have d[sd The •-9. .neer. 'iotul i 989; nsive tmoking and' histologiai types in lung cancer in Hong Kong Chinese wrcnnen. Br f Cacrr 1937; 56: 673 - 8. 6. Koo LC. Ho JH.c. Sap D. Ho Gy. Mezsutcmeats of pz,eive smoking and estimares of lnng eancer risk among noa- smoking Cluceoc fcmales. Grt f Cp.crr 1987; 39:162 - 9. 7. Gae YT. Blac WJ. Zheng W. u m! Lung cancer among Chinese women. 1w J Caocc+r 1987; 40-- 604 - 9. 8 Inoue R. Hirayama T. Passivr smoking and lung cancer in women. In: Aoki M. Hisamichi S. Tominsga S, ads. SwoLrirg aad Fd'ia461987, Mascestaa: Fscerpta 34edica, 1988, 283 - 5. 9. Gen GY. L'ang ZH. 2hang AY. Wu GL On the rchrion- sbip be7w+cra smoking and female lung auxer. In: Aoki S(. Hiaaraicai S. Tominigs S, eds. Sso.Eisg ssd Flserltb 1987, AmstercLara: Exeetpca Medica. 1988; 483 - 6. 10. Shimiuu H. .lelorishiu :vl. htizuno K, st &L A esse•coarrol stvdy of lung cancer in noa•smoking women. Toboiu ~ Exp :fitd I9$8; 154: 389 - 97: 12. 1Cat:ouyaaai K. Willetc W, Trichopouloi D. sr zL Risk of bresst cancer among Gredc women in relation m nucrienc intake. CWcts 1988: 61: 181- 5. 12. atiettinen OS. Ta.artica! ;oidenialogy. New Yod;: Wilep Mcdic-li 1985: 216-44. 13. Trichopoulos D. Hsich,C., 3tacMahon B. tt xt Age at any 5irth and breast aneer risk.lar f' Caacrr 1983; 31: 701- 4. 14. National Research Council, Commitcee on Passive Smoking. E"barrnrsNsat Tobccco SModic: M*anwirg Expoumr arrd: iuzsring Hsmltb E, ffiut. Washiagcon DC: Natibnal Academv Press. 1986. 15. Surgcon General. Ths Hsslth Coxa+qveawr of lasr,/rrs=7 Smoiinq: . R.pon. Rockville. Maq1Ynd: US Department of 1-feali,h and Human Services. Public Heahh Service, Centers for Disease Controi; Center for Hcait}i Promoricn and Education. Office on Smoking and Health. 1985; DHHS (CDC) 87.8398: 332. 16. World Health Organirstion. Tobacro Smo4rag, IARC Monographs on the EMSluadon of the Caneinogenie Risk of Chemicala to Humans: Vol. 38. Lyon. France: World Health Ocganisacion, IARC. 1986: 17. Willett W. Nnairioxal Epidsnr-ologr. New York: Oxfocd University Press. 1990: 292 - 310. 18. Surgeon General. Tbr Healrb Coouiqua.csi of Smo.4firg jor Womsx:. A Rsport. Rockville, atarylsnd: US Depart- mcrrc eQ Heakh'and Human Setvrices, Public Health Service. Offie.e of the Assisnnt Sccrctuy for Health. Office on Smoking and Health. US Govemraent Ptiating Offue, 1980:0-326-003. 19. Ven. JE. Air pollution as a risk faetar in lung caneee. Am f F,pidsr.io11982; 116: 42 - 56. 20. Hseaszed W. Loveland DB. Sirkca MG. Lung catuer mortality as related to residcace and uaoki.a; histories. L Whitc maits. JNCT 1962; 28: 947 - 1001. 21. Peao R. The marked differences between carotcnoids and reanoids: methodological itapliadonr for biocFieraid' epir3emiology. Cs.csr Sam 1983: 2: 237 - 40. 22. Waetenberg LW, Loob WD, inhibiiion of polycyclic .nomacic hydrocsrbon-induced' aeopizxia by natually occurring indolcs. Corcar Ret 1978; 38: 1410 - 3. 23. Sasco A. Dalla-Vorgia P. SkaHcidis Y, Kauouyaani iK, Trichopoulos D. A+. Emltwiox of tBs Effsrrisuett of Tobacco-cosnnt IsPistari.r Palioiat iA EEC Ms:arbss Stsur. 1948 -1987. a report to the Camaussioa of dse European Cornmunirics. Arhea:: Europe Against Cancer. 1989: 141- 64. 24. Surgeon General. Bsdusirrg the Hssltb Coarsqaexess of SAVO4ag: ?J Ysars of Arotrut: a Btposr. Rockville, bianiaad US Department of Health and Human Se:vices. Public Hnlrh Scrvica, Ccntca for D;swe Conuol. Cmacr •.'or Chronic Diseaw Prevencion and Health Promodon. Offue cor Smoking and Health. 1989: DFIIdS Publication ICDC) 89 - 8411. 25. Bno..nsan RC. Reir••JS. Keefc TJ, ferguson SW, Pritzl JA. J Risk factors for adenocarsiaoma of the lung. Am Epidaaiol 1987; 125: 2 5 - 34. 26. Dalager :1e1. Pickle LW. Mason 17 sr ai: The rrhdon of passive tmokiag to lung cancer. Cxscrr Rsi 1986; 46: 4808 - 11. 27. Pashagen G. Hruber Z. Svensson C. Paniresmokias and huig c~eet in Swedish women. AnJF,prr4,a,icJ 1987•,125: 17 - 24. 28. Wu AG., Hendcaon BE. Pike MC. rt a!. Smoking and' other risk factors for lung cancer in women. JNC1 1985: 74: 747 - 71. 21