Tobacco Documents Online

0' In,. l. Cancer: 27, 1-4 (1981) LUNG CANCER AND PASSIVE SMOKING Dimitnos TRICHOPOULO§', Anna: KALAND1Dt'. Loukas SPARROS"and BriamMACMAHON-"s ' Departrnent of Hygiene and Epidcmiology. University of Athens School of Medicine, At/uns., Greece; :Dtpanmtnt of Epidemiolog), Harvar•d School of Public Health„ 677 Huntington Avenue, Bosron- MA 02115, USA. Fifeyone wornen, wfth kx+g cancar and 163 od+er hos- pital patients were intsrview.d~rKardinj the smoking Mbits of tl+ernsetves and tAetr husJaar+de. Forty of tMe kin= cancer cas.s and 149 of tAe other patients wen rqersmokera. Among tAe non-smoking women there was a etatistkalty eignificant d[fferwrce between do cancer os.s ard the other patients with reeqett Oe thefr IwiDardt' amoking habits. Estimates of tKe relative r*k of kx+j cancer aseociated with having a/nuw+d who •mokes were 2.4 for a smoker of leu than one p.ck and 3.4 for worni.n wPwsc txnbands smoked mo.e than one pack of tirarettrs per day. The limkations of the data are enarnin.d; it is evident that further ihvesti=ation of this isue is warranted. Acute and chronic effects on lung function and the cardiovascular svstem.have beemnoted in non-smok- ers involuntarily or passively exposed to the ciga- rene smoke of others (Aronow, 1978; L.enfant and Liu, 1980) We report observations suggesting that the effects of such exposure may include the most notorious health consequence of smoking among smokers themselves - carcinoma of the lung. MATERIAL AND METHODS This is a case -control ~ study. The cases were all of the female. Caucasian patients, registered as resi• dents of' Athens, who were admitted to any of three large hospitals in Athens. between September 1978 and )une 1980, with a final diagnosis of lung cancer other than adenocarcinoma or terminal bronchial (alveolar) careinoma. The hospitals were the largest chest hospital of Athens ("Sotiria"), the largest cancer hospital ("Agios Savas") and the only other hospital exclusively for cancer patients ("Agii Anar- gyri"). Of the 51 cases identified, 14 were histologi- cally and 19 cytologically t:onfirmed, while in 18 the diagnosis was based on clinical and ndiological evi= dence. Diagnosis of adenocarcinorna can confidently be excluded in the 14 histologically confirmed cases. It is possible that some adenocarcinomas are in- cluded among the 19 cytologically diagnosed cases and probable that there are some among the 18 clini- cally diagnosed patients. However, even in un- selected clinical series of lung cancer cases among women in Greece, adenocarcinomas and alveolar carcinomas-do not represent more than one-third of cases (Papacharalampous, personal communica- tion): the number in our series is therefore not likely to be more than seven or eight. Comparison patients (controls) were hospitalized during the same time period' in the Athens Hospital for Orthopedic Disorders (KAT). This hospital is located in~ the same area of: Athens as those which were the sources of the cases. The hospitals from which the cases came were considered unsuitable as sourc4s of controls because of the high proportion of patients with other diseases of the lungs and othen smoking-related diseases: we did not wish to have the interviewer judge, on a case-by-case basis, the suitability of a patientfor control purposes. Six timess during the time-period of the study, the same physi- cian who interviewed the cases visited the Hospitall for Orthopedic Disorders and interviewed all the av- ailable adult women patients in two departtnentss of the hospital. Non-Caucasian patients and patients ttot registered as residents of Athens were not in- cluded. Of the 163 controls so ascertained. 108'were being treated for fractures, 18 for ostcoarthrosis and 37 for other bone and joint diseases. All cases and comparison patients (controls) were interviewed by the same physician. They were asked about, the smoking habits of themselves and their husbands. Specifically, they were asked when they started~smoking. if and when they, stopped and what was the average number of cigarettes smoked daily; the same questions were asked about their hus- bands. Those who had stopped smoking 5-20 years before the interview were classified as ex-smokers; those who~had stopped smoking within 5 years of the interview were considered as current smokers; and those who stopped smoking more than 20 years pre- viously were classified as non-smokers. For the com- putation~of the total number of cigarettes smoked by her husband, a woman's exposure was considercd to start with~ her marriage and to end when she was divorced, or when the husband died or stopped smoking. A change of husband was considered as a change in the husband's smoking habits (if the two were in fact different), and singleness was consi, dered the equivalent oT marriage to a non-srnoker.. Statistical significance is assessed by the X= for linear trend in proportions, as described by Armit- age (197i); tttstnTs Demographic characteristics of the cases and con- trols are compared in Table I. The groups are similar in age, as indicated by the distributions in Table I and' means of 61.7 for cases and 62.1 for controls. Duration of marriage, occupation, socioeconomic status (as measured by years of schooling) and re- cent residence are not notably or significantly diffe- rent between cases and controls. It is, therefore, not necessary to stratify for these variables in the analy- sis particularly since none is significantly associated 'To whom reprint requests should be addressed. Received: October 15, 1980.

2 TRICHOPOL'LOS ET AL TABLE I DEMOGRAPHIC CHARACTERISTICS OF T}tE CASE AND GONTROL PATTEt.7S humErrr Pcrcimtyt Char.nernir Caaes Convoh Cues ControY Total number 51 163 100.0: 100;0 Age: <50 years 7 21 13 7 1', 2' 9 50-69 years 30 98 58.8' 60.1~ 70+ years 14 44 27.5 27.0 Never married 1 15 2.0 9.2 Duntiomof marnage'' <20 years 8 33 16.0 22.3 20-39 years 31 70 62.0 47.3 40 years I7 45 22.0 30.4 Occupation Housewife 32 96 62:7 58.9 Agriculture or labor 12 44 23.5 27_0 Schooling of 6+ years 19 71 37.3 43.6 Recent residence': Urban Semi-urban - 34 3 101 13 . 66.7 5.9 62.0 8.0 Rural 1,4 49 27.5 30!]' ' PercentaEesof themamed -` AlB pattemss rere rcpstcred as.restdenttn Athens. but sotee had chanted residence recentn. perhaps m mnnenson .tith thrtr need tor tnedKal ure. Classified acmrdtng to standard classibcauon of the Greek I:atronat'Statuttcal Smsce, with~smoking in these data. The duration of school- ing of the husband was slightly longer in controls than in cases (65.0°'c 6 years or more, compared to 54.917c ) but again~was not related to smoking habit. Among ttie 51 women with lung canceo, 11 were smokers. whereas among the 163 control women J4 were smokers, giving a relative risk associated with, smoking of 2.9. These 25 women. were excluded from the following analysis.. The mean age of the remaining 40Jung cancer patients was 62.8 years and of the 149 remaining control women 62.3 years. Among non-smokers, control women were of only slightly, higher socioeconomic status than the cancer patients - 63 °k of their husbands had finished prim- ary school, compared to 58% among the controls. Table II shows the distribution of non-smoking women with lung cancer and of non-smoking control women according to current smoking habits of their husbands. 'i7~erC`is'a staustically tion betweca :d>te~Mnb~pd~r;. ~z altiti~:~r~ntria~t."~' lung cancer risk : A noa-imoinnj ivomin'whosc Mii=° band is oker°has a:rtsk otfd~,veloping l~trg• h1.~twice as.higft as tfiat of i noii= , sin'i7l~ing womars*tt amed~toIg-non-sin'olEtr. Table II1 shows the distribution of non-smoking women with lung cancer and of non-smoking control' women aceording to the estimated total number of cigarettes smoked by their husbands by the time of the interview. It may be noted that there are only 64 women in the "zero" category since the husbands of three women with lung cancer and of 15 controls died; or divorced their wives, or stopped smoking, more than 20 years ago and thus were classified among the non-smokers in Table 11. There is a statis- tically significant association between total number of cigarettes smoked by the husband' and a woman's lung cancer risk. The associacion between husband's smoking habits and wife's lung cancer risk was ex- - amined separately for patients with or without cy- tological confirmation of the cancer. The slope of the linear trend was practically identical in the two - groups. TABLE 11 SMOICAIG HASrrS OF HUSBANDS OF NON•SMOKAIG WOAFNWTTM LUNG CANCER AND OF NON•SMOlQNG CONTROL WOMEN IT.gDaa+u N • a E k Gprenapn. d.y (cunrnr aaaotetrl. ioup oe smo m a-Mo en 1-10. 11-20 21-)O 31+ Touli Lung cancer 11 6, 2 13 4 4 40 Controls 71 22 9 32 6 9 149 RR' 1.0 1.8 2:4 3.4 ' Relatrve rnk - the ratio of the rsk of lung eanoer.amon1.0 rxn rho.e husLandc hebna to a parucular,amokma orepry,to thatamona romrn whose hucbanQs are oon•moken- - X=(baeartreod).- 6'.a.. p(2-tad) .<D.D2-

LL'1NGCANCER AND PASSI%'E SMOKf^G 3 lt was note&abovc that the proportion of nevtr- marned women is lower among the cases than among the controls, and. since single women have been classified with those whose husbands were non, smokers, the associations in Tables 11 and 11I are stronger than would have been observed if concern were limited to ever-marraed women. In Athens. in the age-groups involved in this study. never-marned women tend to have the traditional values and habits associated with singleness in elderly women and for tMs reason are, we believe, correctly classified in the extreme group of women never having been exposed to a husband's cigarette smoking. However, if the single women are excluded; the association remains. significant (X2 - 4.6; p=0:03) and relative risks of 1.5, 2.0 and 3.0 art observed for the three categories of husband's smoking for which relative risks are shown in Tabfe 11. unusual opportunit, to investigate this issue. Until about 20 years ago,smoking was unusual among wo- men: whereas it was alreadl quite common among men (Greek Cancer Society, 1978). It is therefore easier to discover an effect of passive smoking among Greek womemthamamong men or women in, other Western populations, since in the latter groups the overwhelming effects of active smoking. to- gether with the high correlation between smoking habits of spouses. will confound and conceal the ks- ser effects of passive smoking. It is. on first cunsideration, strange that the rela- tive risk associated with passive smoking in this study (2.4 for all categories of smokers combined) is only slightly lower than the figure of 2.9 associated with active smoking by the women themselves. However, the numbers are smallland the confidence TABLE III DISTRIBL'TIO!, OF NON-SMOKI"1G wOMEN WITH LUNG CANCER AND OF NON:SMOKING COr.TROL WOMEN ACCORDING TO?FZ ESTIMATED TOTAL NUMBER OF CIGARETTES SMOKED BY THEIR Hl9SBA.'JDS BY THE T1M'E OF THE IN'rERVIEw Dup+ost K TaNnumber oVeyarena 1111 tMVanQs 1 ~ aoap 0.. 1-99. 100-199 200-290 10a399 dOD+ Total Lung carxaer 8 4 6 9 6 7 40 Controls 56 21 26 ~ 16 12 18 ~-- 149 RR' 1.0 1.3 2.5 3.0 "See footnotc to~Tabic I~~..XJ~~(Gneartrend)'- 6.S0. p~(2~-tu1) .<Q.02. DISCIJSSION. This study has obvious limitations and is offered principally to suggest that further investigation of this issue should be pressed. Most seriously, tlte numbers of cases are small. Nevertheless, the associ- ation is in the diTection expected - if any association were to be expected - and is unlikely to.be due to chance. There is a high percentage (35%) of cases laciting cytology,,but the association existed both in those with and in those without cytologic diagnosis. That the comparison group was taken from a diffe- rent hospital from those of the cases may also raise questions. However, the ratio of smokers among the cases themselves to that among the comparison pa- tients is about as expected from previous studies of smoking and' lung cancer in women (Hammond. 1966; Doll u ol., 1980), and no major d'etnographic difference between cases and controls was found, other than in the proportion of single women. The difference in the proportion of single women is con- sistent with the hypothesis of a meaningful associa- tion between lung cancer risk and husband's smok- ing, but in any event cannot explain the difference observed within the group ofmarried women. Against the limitations of the study must be put the fact that the Greek setting provides a somewhat. limits of the latter figure are broad (9S°Jc, 1.3-6.8). In the only other controlled study of this matter in Greece (Kanellakis at al., 1976), smokers of less than one pack of cigarettes a day had a 5-fold and strookers of more than one pack per day a 20-fold itxzease in lung cancer relative to non-stnokers. These are the risks appropriately compared with our estimates of 2.4 and 3.4 associated with husband's smoking of similar amounts. Further, active "smok- ing" does not have the same connotation in men and women. Women smokers tend to smoke less heavily than male smokers but have lower relative risks of lung cancer even for a given level of smoking (Ham- mond, 1972). The explanation appears to be in the facts that duration of smoking is an important deter- minant of risk, women in the current lung cancer ages commenced smoking at a later age than men of similar age and have therefore been smoking for shorter periods, and' substantially smaller propor- tions of women than men inhale (Wald, 1978; Doll er a1.,,1980): These factors complicate a comparison of the risks associated with active and passive smoking, but at least one of them - the frequency of inhalation - seems likely to operate in favor of a relatively larger eHecr for passive than for active smoking. other components of the exposure being equal. Fi- nally, it, has been observed that smokers tend to clus-

4 TUCHOroVt:os El AL ter together social)y, (Reeder, 1977), and the smok- ing hablts of a woman's husband may be an indez of a broader exposure to cigarette smoke than that which emanates from the husband himself. ACKSOIWLEDGEMEtiTS Thi, work was supported b) grants from the Greek Mrnistr` of Soclal'Services and the U.S. Na- tional Cancer institute (5 P01 CA 05373). REFERENCE.S AKrsrrAOe. P. Starurica!'merhodr in rnedical nuearch Black- well Scxntific, O>dord ()9711, Aaorcow„W:S.. Effect of passive smoking on angina peetoris. N. Engl.1: Med:, 299, 21-24 (1978). Dont. R., Ga.Ar, R.. HAfnrea, B., and PE7o, R., Mortality in relation to smoking 22 years' observation on feeuk British doaors. Bru. rned. !:, ?,0, 967•976 (198Q). Gttux CANcEZ< Soetrrt_ Research on the srirudes ofrhr public towards cancer, pp. 48-51, CanocrSociety, Athens (1978) (iln Greek ). HAr,oaon-D- E.C., Smoking habits and air pollution in relation to lung cancer In: D.H.K. Lee (ed.)„Environmrnm! Janors in nrspiratory diseasr. pp. 177-199, Academic Press. New York (1972) H.u,awr,v, E.C.. Smoking in relation to the death rates o(one milllon men and women. In. W: Haetsuel (ed.), Not. Csncrr Aur Monogr. No 1.9, pp. 127,244„Supetintendent of Docu- trxnts, Wuhington, D.C. (1966): IG.wEu.AKls. A.. Talcr+orout.os. D. MICHALAKOrouws, l:., MARAGOUDAKIS. S.- KANELLAKI. K.-. XIttOUCHAKI, E., and KALAron+AKl. V.,The relationshlpbetweensmokingof Greek eigarsnes and the developttxnt,of lung eancer. Mauno Med Gnrca, 4, 351-355 (1976) (In Greek. with an English summary ); Ler+rAxr, C., and Ltu'- B.M.. (Passive) smokers versus (volun- tary) smokers. N. Engl. J. Med:. 302, 742-743 (1980). RttEtsEn. L..G., Sociocutiural factors in the etiology of smoking behavior: an assessment. In: M.E. Jarvik. 1.W. Cullen. E.R. Gnu. T.M. Vogt. and L.1'. West (ed:). Research on smoking Aehovior. Narionaf lntnnur on Drug /16use Research Monogr. 17: DHEW Pubhcation No. (ADM) 78-581'. pp. 186-20t1,. Superintendent of Documents. VVashington D.C: (]977): WA1D, N.1.. Smoking as a cause of disease. In. A.E. Bennett (ed. ). Reeenr advances in, convnunirv medicnne. pp 73-96- Churchill Livingstone, New York (19'78).