Tobacco Documents Online

I CORRESPONDENCE Re: Environmental Tobacco Smoke and Lung Cancer Risk in Nonsmoking Women The reportt by Stockwell et al... (1) adds little to the data on environmen-, tal tobacco smoke and lung cancer 1. Various biases could have contributed .. to the associations noted'between lung cancer and some indices of environ- mental tobacco smoke exposure. One . concern relates to the use of healthy control subjects who were obtained by i, random-0igit dialing. resulting.in pos- I sible recall and nonlesponse bias. Another concero, perhaps more so than in other lungg cancer-enviton- ', mental tobacco smoke studies, is information bias. Thus, all contml subjects provided dataa directly, but surrogate.cs provided data for. 67% of case patients, many of whom were dead. There were also notable~case- I control differences in theproportions of interviews conducted' face toface, , by telephone.. or by mail. Much at- ~ tention hars been given to bias from the misclassitication of smoking j habits (2),.but, although Stockwell et i al. (f) refer to this. misclassification, they attempt no statistical adjustment and do not pre.9entt a comparison of smoking status as recorded at various stages of their study:.They also fail to consider confounding by. diet. This failure is remarkable, since in another paper (3), apparently based on the same study; they report a strong pro- tective. effect against lungcancerg amongg nonsmokers that iss associatedl withh total vegetable consumption and with intake of carmene, and, as.i have reviewed elsewhere (2),, a reduced consumption of vegetables is agi:oci- ated withh marriage to a smoke0 Ad- justment for this source of bia$ alone 'could well render the reported asso- ciation, between lung cancer and ex- posure too environmental tobacco :smoke not statistically significant. , There are also severe problems. !regarding presentation of results in the report by Stockwell et al. How . can one update mela-anatyses for differing indices of environmental tobacco smoke exposure when risk estimates are presented only for those 'indices for which an association is reported? What were the odds ratios and confidence intervals for environ- menWl tobacco smoke exposure at work or during social activities? Meta-analysis would alsobe assisted by presenting,, as other investigators do (4,5), numberss of cates an wnlrols by exposure. Another di t- ulty_ is the unusual method of Vcis for spousal environmental to- ~cco smoke exposure, wit)i the r4ferent groupaot,p as is to omary (Z7, marmi:d women whose ebands didF not smoke;, but instea~women (m vied and unmarried) unlxposed to housthold environmentAl tobacco 'smoke: from any sourceiThe relative risk es(imate is thus pat comparable with thyt for other ;[udies (2). Fur- thermore,C because of the strong asso- j cialion of the t ices of environ- mental mental tobacco s oke used with both marital stat`us a household size (uhe I larger the famiiy, the more Iikelyy is , exposure), thre are additional pos- sibilities of c~lfounding. The referent ' exposure up is idemical for alil I analyses ip~Tal 2 and 3 of the report by,Slockire6 et al. (1). mere- fore, theeZ2d `!ltisks for diffcrcnt indexes' of env nmenlal tobacco smoke~exposure art not.independent and could all be~ffecled by an unusually low proportion of unex- posed cases, perhapsVesulting from recall bias. In any. event, the sultss of the k studyby Stockwell et ab (J) have no real effect on the overall ta. Based on all data available before his.study, meta-analysis (Table f) ~Ives no overall s[etistically significaLt asso- ~l ciation of lung.cancer to.workplace or childhood environmental tobacco snwke exposure and only a small positive acsociation with a husband's smoking,. which, as I have previously shown, can be explained in terms of mi9clascification of smoking habits, co oundiag, publication bia9, and s ciffc study weaknesses (2)t. ~ PETER N. LEE, M.A. P. N.' Lee Smfi.tric.c und Crmnpaling Lrd. 17 Cedar Rd. Sutlml, Surrey SM2 5DA. &fg/and References (J) ST(NXwP.II. Hrl. GbIJIMAM AL LYarAn GH', nn .r.: Env7ronmenW.tobacco smMte and lung cancer risk in annsnmking .nmen..J Nml Cancer InM 84:1417-1422, 1992 (2) It.n PN: Envimnmemal Tohacco. Smnhe and Morfalify. Basel: S. Rarger,. 1992 (i) CANUer.nnn EC, Srncawet.t. HG, AsarsYarnq AW. nT nl.: Dietary intake and ri.vk of lung cancer in wnmen who never srmtked. Nulr. Cancer 17:263-270, 1992 (4) WAr.n NJ, NAnaUr6vk K,.Tnoansnn SC, r.Y nr.: Dnes breathing nther peryd<'s tohaecn smoke cause lung cancer? Br Med I (Clin Res Edl. 293:1217:-1222.. 1986(5) NAnnnnc ResowacoCrmnon_ Envimn- mental tubacrn smrdtt. Measuring ea- Imsures and aase<aing.heallh effects. Waali- ingtm: Natl Acad Preu. 1968 (6) FnnanAM E71. CrmaEA P. Wo-Wn.rrAMs A. F.t AC Lung cancer in nnnsmnking wnmen:A mMicenter caneKonrrol aludy. Cancer Epideminl Rionrarker. Prev 1:35- 43- 1991 (71 f]u 2.Y. Hn X7 Cuwrr.rAn. RSc Sm„king and rxher risk facurcx au lung# cancer in Xtmnwei, China. Inl J Epidemial 20:26-31. 1991 2nd Letter Stockwell ct al. (!): reported a relative risk of 1.6 for women who never smoked and who were married to smokers. This relative risk was adjusted for age- mce- and education but not for other poten(ial'l con- Table 1. Meta-analysia' of studies ..IC environmemal rnhacco xmnke and lung cancer among reryxtsllf lifelmg nonamukers Source of envirmmwl Na'pf smdie. tobacco amoke esrynaure Pmv'ding dau Workplace II Childhrwd rl flushand 30 Relalive risk 95% canncence eximmes interval 1.02 0.93-1.12 0.96 0.g5.1.09 1.19 1.119 1.31 1 • Based on data presemed in (2)) in TaMex 3J4F, 3,21, and 3.23, with the addition nf deta frnm twn smdies (6.7). Jrnunal of the National Cancer Inslitute Vnll. 85, No. 9:. May. 5,. 1'993. 748 CORRESPONDENCE

From the Surgeon General, US Public Health Service i Tobacco Contiul:As the former Sim geon General, the iasum Vist lie at the heart of tobacco contro availabil- ity,themarketing,thesale,: thedan- gets--®lleoncernmegreatlyN these issues most troubling when y per tain to our youth. And the ' of women, children, and mihority ups by the industry also troubles me. As a nation, wencedan action that.can help save millions of lives. plan must speak for our younger eh drea who have no say and no choice regardingtheirexposuretosecondhand smoke orihdustiy tactics. And the plan must give 5rst priority to protecting the health of nommwkers. During the past 3 yearss as Surgeon General, I traveled wadt to coast talk- ing about tobacco control. I have deter- mined that we must face-10 key chal- lenges if we are to gain control over tobacco use in this country: 1. We as physieiaosand.other health professionals must speak strongly with one voice and continually remind the public of the following. . Tobacco use is responsible for the prematuredeaths ofnearw one half miI- lion people every year in this country. . Smokingis the single mostprevent- able cause of death in our sooety,and it is oosting thiss nation plenty-over $65 billion peryear, more than a bi8ion dol- lan per week. . Tobacco is the only produdt that whenused as directed results in death and disability. 2. We must expose the seduction of ourchildren bythe tobacco industryand work proactively to counter its mes- sages.and techniques. More than one millioncluldren start to smoke in the United States every year-thatls 3090 per day. Ten percent of them Nart smok- ing by the fourth grade, and nearfy two thirds of them by the tenth grade. Each year, the tobacco indbstry gar- ners=1 million in profits from illegal sales of tobacco4o children.The upcom- ing Surgeon General's report on youth initiation of smoking will provide ample ammunition to wage this ongoing battle. 3. We must spur federal,.state, and local legislation that will limit minors' access to tobacco.,Each year, more than 3 million American children under the age of 18 years smoke 947 million pac of cigarettes. Although virtually states ban tobacco sales to minors, roany states have trou ble enforcing these laws, and only.Floridaand Vermomare known~ to enforce therru. 4. We must speak out against the se- BOB .1M.7A. AugusI 18. 1993--va 270. Nm. 7 duction of women by the tobacco indus- try and develop effective prevention messages to countertheibdusjry's mis- leading entieements: Sbme22 im7lion women in this oountryare smokers, and cigarette amoking prevalence amongad- olescents is about equal fur bochmales and.females-tietween 18%and 1 . In America toda_v,.more women di ~m o lung cancer caused by cigarette n(ak- ing than die from breast cance . Other than lung cancer, A erican women still have not fully un Lstood the consequences of smokfng at per- to them--oral, cerviat, tic, dbbddercancers,low-bhth eightin- coronaryheartdisease, stroke.. nmencontinue to exe their in- rights, we must a way to em glimpse the tru behind the our and sophis tion~in se- vertising and nt the un- ies~~.ofsmo g. talertthe blicaboutthe tobaoeosmoke nswhodon'E to endure the ke of others, is linked to the y 300o nonsmok- each year. The En- ion Agency'sdes- ~ des- h group Auman tronger regula- lic'ss health by blic places. widespread tobacco as a In 1991, op age eUS- n~e- ed di help false ductive healthy 5. We dangersof viromne (ETS).The . %ofour smokeshoul not h consequences especially whe death of approxi ers from lung can vironmental Pro ignation of ETS carcinogen will tions.to prote controlling a 6. We m use of smok "safe" about 7 mi and older the maju spite our, relation bacco dict p in control or "spit vetosmo on people 12 ed spit tobaa:o i of them~young ings and the d p between prolonged and oral cancer. Expe oral cancer epidemioin ecades if the current tren baceo use continue. e must recognize the impor[an ce in controllingtobaceo use, es ally amongyouth. The 1992 Surgeon eral's report estimated that for ev- ~ 10%increase.in the price of tobacco ~roducts, there will be.an approximate %.decrease in tobacco consumption. The Centers for Disease Control and Prevention recently reported that be- tween 1990 and 1991, adults 18 years of age and oldershowed no decline in smok- ingprevalencc the first year with no decrease.since 1965. Economic factors that may have contributed to this lev- eting include an increase in the avail- ability of discount cigarettes and a 10% annual inctease:in thealready morethan $4 billion spent on cigarette advertis- ing, rebates, tteesamples, two-packs- for-the-price-of-one offers,and discount coupons on cigarette rartons. The Coalition on Smoking OR Health has called for increasing the tobacco ex- cise tax by $2 per pack of cigarettes,, which will generate more than $20 bili lion in the lust year and is projected to result in close to, 8 million fewer smok- ets and 2 milkon.lives saved over time. We must ensure thatany increase in the federal excise tax on cigarettes includes a tax increase on all tobacco products. After all, tobacco is tobacco-whether one chews it, spitss it, or smokes it. 8. We must document, assess, and change soc~ty's attitudes and practices regarding tobacco use, e9peciallywhere our children are concerned. We mus4 improve our ability to coBect infosma- tion om morbidity and mortality-and particularly on usage and public opin- ion-that is mmprehensive, gender-spe- cific, and cultu ral ly specific. Equally im- portant, we must evaluate information on the economics of the tobacco indus- try to u nderstand the pervasive role that tobacco plays in our economy. 9. We must strengthen and harnessthe tools that we do have in controlling tobacco: legislation, taxation, and indi- vidual and public health education. We also should consider changes in warning labels to provide updated, comprehen- sive information on the health risks as- sociated with use 4 tobacco products. 10. Personal involvement, in preven- tion andin control, must be our watch- words. We knoww the devastating con- sequences of tobacco use,.and'we know how the tobacco industry promotes its deadly product. We must insist thatthe tobacco industry operate as responsibly as.any other consumer product manu- facturer in this country. And we must become betterrole naodelsand the cFiam- pions of prevention if we are to succeed in increasing public awareness and in promoting public action. Bolstered with this action plan.and ed with scientific integrity and a pur- eful sense of direction, we can ensure ~ theindustry will not decide who the ne~ emokersand chewers will.bel by Antonia C.. Novello, 11fD, MPH Former Surgeon General aEnw'aada:FU~eivqu,e6m91'.bfYw.YEbWR 't~ .nreewcnNteaa.ycewanew.ose.am. _y~[{suqw, Ge,Wai. nmm ]~6.e IYmqr9y aWang„ yWp.,p~n. oC 2n4oi: rebpwro (2a2199h)i® ~ from the suqeon C+eneral. usaHs

founders, such as diel, occupatirm- ,and puior lung disease. In fact, no mentionn wass made of data having been collected on those factors. How- ever, ever; in another repon. Candelora ef al. (2) discussed an analysis of diet and lung cancer in a subset of sub- jects from the same study:Among , other results, they reported strong inverse associations between lung cancer and both,total vegetable con- sumption sumptfOn and lotal carotene intake. ! For example,, the relative risk for the highest consumption quartile versus ', the lowest quartile was 0.2 for lotal vegetable consumption and 0.3 for tolal'l carotene intake. Although Calendori et a1. stated that informa- tion wascollecled on occupation, . exposure to known lung carcinogens, , personal medical history, and family ' history_ of cancer and respimtory '. diseases, the diet analysis was not adjusted for anyrof those variables or for environmental tobacco smoke exposure. Other studies (3,4) have noted inverse associations between dietary factors and lung cancer among.people whonever smoked,, and three studies (5-7) have reported inverse associa- tions beuweenn environmental tobacco smoke exposureand: 0-carotene intake among women. in the UnitedStates who never smoked. Clearly, diet is an important potential confounder in re:- ported associations between environ- mental tobacco smoke exposure and lung cancer. Since associations were reported between lung cancer and both environmental tobacco smoke exposure and diet ih the report by Stockwell etl al. (1'.), it would! be interestingtog know if there were associations between dietary factors and environmental tobacco smoke exposure that would give rise to confounding of the lung cancer asso- ciations. A multiple logistic regres- sion analysis that considered all of the ', potential risk factors for which data were collected would be useful in elucidating these relationships, A history of nonmafignant lung disease is another poteneiall con- founder in reportedd associations be- tween environmental tobacco smoke exposure and lung.cancer. In a large case-control study of women who never smoked, Bsownsonn et al. (8) reported a spousal smoking-lung can- cer relative risk of 1.0.This relative risk was adjusted'for prior lung disease, but the extent of the adjust- ment was not stated. In another report on the same study, Alavanja et.al. (9) estimated that 16%% of lung cancer cases among, women who never smoked were attributable to prior lung disease. The importance of adjusting esti- matesof associations between en- vironmental tobacco smoke exposure and lung cancer for potential con- founders-is emphasized by the weak- ness of the overall epidemiologic data. From a meta-analysis of the 13 currently available studies of U.S. women, including the studies by Brownson et al. (8)and'Stockwell et al. (1); t have calculated a summary spousal smoking relative risk.estimate of 1.07 (95% conffdence interval = 0.95-1.21)...Thisestimate was adjusted for smoking status mi'st]ussifrcation, using the.assumptions and methods of the U.S. Environmental Protection Agency (10).. Only two of those studies (8,11)adjusted for prior lung disease, and none adjusted for dietary factors. Both Sidney et al. (5) and Le Marchand er al- (6) estimated that confounding by (i-carotene intake could inflate environmental tobacco smoke-lung cancer relative risk esti- matess by about 10%; therefure, the very weakoverall. U.S. association could conceivably be explained by that single factor. laa recent Journa4 editorial, Burns (12) asserted Ihat a causal relatiunship between environmenlal tobaccasmoke exposure and lung cancer has been established withh what'he characterized as "scientific certainty." The. above considerations,.and many other uncer- tainties.in Ihe environmental tobacco smoke-lung cancer epidemiology, lead me to believe that Burns' conclusion is unjustified. MAXWFIL W. LAYARDI Laynrd Associrrres 2242 San Antonio Ave. Alraneda, CA 94501 References (/) ISmcKwmi HG, Cnn.ueuN AL, Lvkuw GH, er A: Envimnmcmal mhxuu .muke and lung c+rxur riak in nuwm,king women. J Nml'C:mcer lnw84:191R-1422, 1992 (2/CAnueLOxw EC, ST.aCKwuu. HG, AasuianNUAw,.ur AL, Dietury intake and risk of lung carwer in womea who never Mmd:ed. Nutr Cancer 17:267-270, 1y92 (J) Kou LC: Dietary habiu and lung cancer risk among Chiuese tetmies in Hong Kong who never srmrked. NWr Cancer 11:155-172. 19gg . (4) FRAsrn GE, Beeson WL. Punurs RL: Diet and lung cancer in California Seventhday Adv<nlists. Am J Epidemiul 133:683fi93, 1991 (5) Smner S, CwnnBl. FRreoarAn GD: Dktmyinlakc of caralene in nonsmokers with and without passive smoking at hume. Am I Epidemiul 129:1305-1,309, 1989 (6) Le MARCnAMn L. Wtl.l:ans LR, H.urBUt JH. sT AL: Dieuryy patterns of female nunsmokers with and without exposure m envimmnenul tobacco smoke: Cuncer Czuses Control 2:11-16. 1991 (7)SnmATA.A, PAUAnmo-Hni A, Ross RK, nr. AL: Dierary p-carwene, cigareue smuk- ing, and lung cancer in men.. Cancer Causes Control 3:207-214, 1992 (8) .Bxnwnwn RC, A.AVAmA MC, Hncx ET, en AL: Passive .moking and lung cancer in nonsuoking women. Am I Public Health g2:1525-l5D0, 1992 (9) . ArAVwnu MC. Buowqswr RC; BnR.'a JD lx, er nc Preexisting lung disease and lung cancer among nnn.muking wumea. Am 1 Epidemiul 136:623-632, 1992(10) : Euv1uow.mrrrAL Pvorcc-nuN Aunncr: Respimmry Health Effects uf Passive Smuking: Lung Cancer and Other Disur- ders. EPA/600/6-9tU0116B.. Washington, DC, 1992 (T/):GAxsrn.eL Lr.Time uends in lung cancer monaliry among nrmsmokers and a note on: pacive snuking. JNCI 66:1061-1066, 1981 (J21 Bunns DM: Environmental tobacco smoke:.The prim of scientihC eenainty. J Nall Cancer Inst 84:1387-1388, 1992 Note -Amhor's note:,The aurhur is a pmtner in Layurd Asxaoime.;.a smliyicul axuulting Inm. He arxnuitslur the Trdacco.lmtimm. 3rd Letter While another investigator (J) sees the report by Stockwell et al. (2) as an affirmation of a discernible link between environmental tobacco smoke exposure and lung.cancer, the evident inconsistencies pointed out in the Stockwell report should give one pause. For example, the adenocar- cinoma dataa show no statistically significant relationship to environ- mental tobacco smoke exposure and no pattern of dose response, in sharp Journal of the National Cancer Institure, Vol. 85, Nb. 9, May5, 1993 CORRESPONDENCE 749

distinction tei the large IJ.S. study by Fmuham etl al. (a) thaa reported an elkvated risk only for adenocar- cinomas. As. a second example, the apparent relationship between the ex- posure information source and the reportect risk is opposite tothato reported in the large study- by Gar- finkel et al. (4). The pointt is that hunting expeditions through the data , of an epidemiologic study can easily. ~, produce inconsistent artifacts where the exposure.effects, if any,. are likely too be very small.. My second point concerns the multiplicity of risk estimates in Table. 2 of the Stockwellreport, which were reported to be statistically significant. The impression some may. have iss thalt of repeated affirmations that strengthen thee claim of a consistentt effect. flowever,, the reported riskk estimates are oved'apping. 11 appears. that there may be only a single statistically significant result,-i.e., the reported risk associated with non- adenocarcinomas withtotal exposure greater than 40 smoke-years. This single result can accountt alsofor reported statistical significance for all lung.cancers, for adulthood exposure.. for childhood exposure, and for all the related significant P values for trend. Furthermore. we were given no '1, information regarding any possible . association between childhood ex- posure and adulthood exposure. [nevitably,, choices were made in both the conduct oflhe study and the reporting of the data. It appears that some or all of the stated conclusions could be affected by inclusion, exclu- sion,, or redistribution of a small number of cases. We should' be told to what extent the investigators' choices could have affected such I conclusions. Examples are choices ~ related to geographic and temporal ~ cutoffs for the selection of cases, the definition of exposure cla.c.ses, the choice of adjustment variables and adjustment procedures, and grouping or splitting of cell typesandre.spond- ent categories. fdeally, science woulc[ be better served if thee study protocols and reporting procedures were published in advance of data collection. The final report of the study could then distinguish between planned and tln- plannedd material. Perhaps this Journal could help to promote such prestudy' publication. The possible impact of other ptnen- tial biases inn addition to the potential . selection biases described above,.de- serve discussion.. For example, smoker-nonsmoker misclassification errors were mentioned, buttlieir im- pact was nott assesced. Possibly of greater importance, there wass no discussion of preferentially lower en- vironmental tobacco smoke exposure among nonrespondent case patiente.or among case patients excluded' because of inadequate information on active smoking. tt would have been helpful if Table ~ 2 of7he Stockwell.report had included I an additional column indicating how . many aases fell into each- of the exposure categories as well as the number unexposed. Such direct re- porting of observedfrequencies, while they are not demographically adjusted relativee to the centrols, provides a fuller appreciation of the underlyiing data. Finally; the Stockwell report notes in a single sentence that the study also looked att environmental tobacco smoke exposure at work and during social activities and' found no statis- tically tically significant estimated increase '. in Iung-cancer risk. This failure to report in detail'l is a finee example of a publication bias where practically. no mention is made of a negative result, and it is therefore unlikely that this sludyy would ever be included in a meta-analysis of workplace exposure studies. PAUL SWITZER„ PIi.D.t DeJ)qrtnlent of S1nlislied Stanford Unn•eeaih' Stmiford; CA 94305-4065 References f/l nvans DM: Envimnmenul rohacco smnkc: The price M.cicmirr eertainly: J Nail Cancer Lrlat x4:13fi7-13ax. 1992(2) SrllenwFl.L HG. Gr/LnamN AL.. LYr.IAN GH. m[ nr.: Envimnmenml tohacro slm,k< and lung cancer ri41 in nonsmoking wnmen. J Nail cancer Inn'a4:1417-1422, 1992 (3) Pnwrnenl ET, CuxveA P. wu-Wnr"nxls A. sr nr.: Lung cancer in mmsmnking women: A mullkemer, casesontmi srudy. Cin er f pukmLd Rn mvkern Prev. I t0. 43 i wl • (41GARrINKII. L. AIrRn.Hlr G, JnnPmT L. Imr hlmary smnkmg atd lung.ancer A naa-.nl I .IUdy.JNCI ]1.<N7-469- 19x3 I Note IAmhm'k nmec.These cmmmenta hare been ' prepared aI the request ofthe Tnhaacn Insliive I and represent the views of the aWMn, nM ruceasarilylhnse nf the Tobxco Institule or of Slanfnrd Universiry. Response Ihl responsee to comments regarding the potential for dietary factors to confound the relationship between environmental tobacco smoke ex- posure and the subsequent risk of lung cancer (J), it is. ilnponant to recognize that Ihe possible role of dietary factors is important for both I smokers and'nonsmokers. The ques- tion being addressed inn our report,, however, was whether nonsmokingg women who were exposed to environ- mental tobacco smokee had an in- creased risk of developing lung cancer compared with women whowere not exposed. Our results indicated that expnsure to enviionmental tobacco smoke can increase thee risk of de- veloping.lung cancer in nonsmokers. A question that should be considered separately is whether dietary factors can exert a protective effect, i.e.. .. reducingg the risk of lungg cancer among those exposed to tobacco smoke. Analysis of our data on this question is not yet. complete, but the results should be available shortly. Itt was suggested, in the correspond- ence by. Layard- that prior lung disease may havee contributed to the lung cancer risk in these women. As tobacco smoke is known to contribute to (he development of both malignant and nonmalignant respiratory diseases in smokers, a shared common ex- posure to environmental tobacco I smoke would appear a more likely I explanation. In the correspondence by Lee,, the use of data from surrogate respond- ents wass questioned. Because lung cancer is a rapidly fatal disease, the use of data from surrogate respond- ents was necessaryin sonte cares. These data were presented in Table 3 of our report 1L), showing Ihe results 750 CORRESPONDENCE Jralmal nfi the National Cancer. Inslitutc. V/d. $5, No. 9. May 5. 1993

of analyses performed separately for selrre/KMsan}I for surrogate resprmd- ents. 'Che odds ratias associated with environmental tobacco smoke ex- posure weraactuallye greater when the analysis was limited'to living sub- jects. This finding.suggested that, had it been possible lo interview all case palientsdirectly, thee data might have indicated an even stronger association between lung cancer risk and ex- posure to environmental tobacco smoke than theyy did when surrogate respondents were included. Lee also indicated that he considered it to be unusual to use women who Aadd not been exposed to environmental to- bacco smoke as the referent group because most previous studies had used only married women whose husbands had not smoked. Consider- ing spousal exposure as the only source of household tobacco smoke, however, ignores the possibility of exposure from other household mem- ben. Janerich et al. (2) reported that exposures to high levels of household smoke during childhood and adnles- cencedoubled the riskk of lung cancer among nonsmokers. To consider only women married to a smoker as exposed, regardless of other reponed exposures to smokers in the house- hold, could.resultin Ihe tnisclassifica- tion of exposed woolen as unexposed, possibly causing an artificial reduc- tion in Ille o(kJk ratio.. Also, consid- eration o~ differences in household size, whic could have an impact on the numbeij of potential smokers in the home,. did not vary by case or control statual, Lee also stated that all associations -betweenn environmental tobacco smoGle exposure and lung cancer from all available data can be explained by issues in study design. However, it must be noted thatt this smdy. (J) increases the total number of studies with positive findings be- tween environmental tobacco smokee exposure and lung cancer,, and as these studiescontidue~to bereponed (3), dismissal of all sr7chfindingy becomes increasingly.diffieult. I 1'M latilALOG OFF.GOVERNMENT BOOKS I The U.S. Government Printing Office has a free caaalog of new and popular books sold by the Government. Books about agriculture,, energy, children, space, health, history, business, vacations,, and much more. Find out what Government books are. 211 about. Send.for your freecatalog. FJ'ee Catalog P.O. Box 37000 Washington, DC 20013-70Gn ilHAllIr:R G. SttICKWF.I.I. AI.I.AN L G(l1lIMAN CI/ARIi.1:5 1.. Noss. EI.IZAnF.TII C. CANDmARA. ADAM W. ARMtiTRONC.~ Department of Energy J Uf~re nf Nea&h /d Germanrmrn, Md. References (1)Snrnwti.l HG. Graw.uN AL, LxruN Gil. eT Ae: Envimnrnemal lobacco snude . and lung cancen risk in nunareuking wwnen. JNall Cancu Insl 84:1417-1422, 1992 (21JANEx.n DT. TrwuPSON WD, VAavaw LR, er At.: Lung eancer and exposure lo mbacco smoke in the househald. NEngl L Med )2J:632-676, 1990 (3) Frm'nInMhT, Gos.r.c P. Wu-Wn.uAas A. Er AL; Lung cancer in nonnnuking women: A mullicenter. caseKOnlrul swdy. Caocer Epidcmiol Biomxrkers Pn:v1:35- 43, 1491 i•lote CurretP.mdsnce rur Heather 0. SlockwelG.. SaD.. Depanmcrn of Energy; Offim uf Htallfi;, EH-02„ 19901 Gemumown Rd., Gennmuwn,. MD 20874. Journal ot. IheNationul Cancer Instiwtu, Vol. 85. , No..9, May 5, 1993CORRESPONDENCE 751