Tobacco Documents Online

PERSPECTIV ES IN PREVENTION SERIES EDITOR/hLARGARET R. SPITZ,.MD. 1N1PH Environmental Tobacco Smoke and Lung Cancer ELIZABETH T. H. FO\TH.-1J\f, DRPH; PELAYO CORREA,,MD; PATRICL-~ A. BUFFLER. PI-iD; R.-~YINIO\D GREENBERG,,41D, PHD;,PEGGl REYNOLDS, PHD; A.NA «'t'A'VILLL-MMS. PHD The N,ear 1981 was a landmark one irrn which two separate reports linked exposure to emironmental'tobacco smoke t:ETS rwith lung cancer in nonsmokers_ `-- A cohort study of Japanese females and a case-controll stud\ of Greek \\omen reported an increased risk oflung cancer among nonsmoking women whose husbands smoked. These initiali reports and most studics t}hatha+e been published during the decade since thenihave cornpared temaJe nonsmokers married! to smokers with female nonsmokers married to nonsmokers. The assump- tion is that the findings for female nonsmokers arc generalizable to all nonsmokers. The causal association, of lung cancer \.ith acti~-e smoking was established in large part, by stud', ics that included a preponderance of males. This was not because of any expected gcnder-rel~ited differences in exposure response, but simpH• because when this relationship was und'er in- tensie e im-estieation in, the 1950s and 1960s most longtime smokers Eli:,aocth T. H. Fonrham„DrPH; Rrlavo Corrcn• .VD:-Drpartmcnr of Patholoqv; Louisiana Srntr L'ntvcrrin Mcdica/ Crn- tar; \'cw oYltans, Loutriana; Patricia A. Bufjlrr, PhD, School of Public Hcalrli; Unirrrsirr of Calif'ornia at Brrkclcv, Bcrkrln; California: Rnvmond'Grccnbrrq;, MD, PhD. Emon, C%nwcrsin School of Public Htalih, AtJanta, Grorqia; Prqqr Rcvnolds. PhD. Calitonoia Dcparrmrnr of Hraltb Scnvcrs, Ernm1•illc„Cali%ornia: Anna Wu-lf'rNinms. PbD, Dipnrrmrnt of Familv and C:ommuntrr.Sfcdicinc, Uniwcrsttv of Soutbnrnj CaJi(ornia, LosAnarlrs. Cnlr/nrnrm. A FEW FAST FACTS V Second-hand smoke is now clas- sified bv the US Environmental Protcction Agcncy as a group A carcinogen. d Only 23 substances have been dc- darcd known human carcinogens. V The article below provides fur- ther impetus for the passage of workplace smoking restrictions by locall and state governments. V On an annual basis, 3,000 lung cancer deaths arc attributed to second-hand smoke. V Sccond-hartd smoke is also im- plicatcd in childhood respiratory illnesses and increases the severitv of asthma: attacks in childrem -M. R. S. Cancer Bull T993;45:92-94 0199377. fSwnrrn?r ejTmi M. D: Adinoe. CG.m Gntn ..,ere male. Today the majority of lifetime nonsmokers with lung cancer are females; hence, the concentration of studies of women. The most studicd' ETS association has been derived from the spouse'ss smoking history. Partially this is be- cause it is usually a physically cJbse relationship with shared1iving quar- tcrs an&bccause recall of'thi's cxpo- sure should be optimal. The chemical composition of sidcstream smoke is the same, of course„ regardless of who is smoking the cigarette. In terms of the rcgulation of ETS, exposures outside the home have greater rele- vance, but few studies have examined these exposures. Multicenter US Study In late 1985, a multicenrcr US stud% was initiated to evaluate the associa- tion of ETS with the risk of.lune can- cer in nonsmoking females. The studv was designed to minimize some of the method'olbgic problems that have been discussed in re\ie%\• of ETS lung cancer studies by the National Re- search Council and the International Ageno- for Cancer Research, among others." These included misclassiti- . cation of smoking status, inaccuracy of case diagnosis an&cell typc, recall bias, ETS exposure from sources in addition to the spouse, andlinadequate control of potential confounders. The findings of the first 3ycars of this stud\, are summarized hcre.` This was a population-based case- control study of lung cancer imwomcn who never used any tn-pe of tobacco: Tobacco use was detincd' as _100 cigarettes smoked or any other to- bacco used for >6 months. Gaograph- icalh• the studv included fi\•e metro- politan areas-Atlanta, Houston, Los Angeles, New Orl'eans, and the San Francisco Bay area-represenring a population base of approximately 18.5 milllon people. Study Methods Rapid case ascertainment proce- dures, which included ~%•ceklv or bi- M•ceklv review of patholbgv reports from hospitals in the studv, were used to idcntifj•potentiall\, eligible lung cancer cases. On completion ofthe case ascertainment, >1'-,500'potenaal- 1y eligiblc cases .,.•ere screened. These cases included English-, Spanish-, 92 llHr CkNCr.a 6~_'L1.eTnNN2ti5. \<> l. 1993 2023513075 sa~ur4~'/~.. ia6rsra/~

or Clunese-speaking females from the aces o1'20'to 79 \~~ho had'a histopath- ologicalli confirmed diagnosis of pn- mam•carcinoma of the lung pnorto death, no historv of cancer, and were lifetime nonsmokers. Two control groups were selected to evaluate response bias, which is al- \van•s a consideration, in comparisons of ill cases and healthyl controls, The first control group (population con- trols) was selecte6bs randomrdigit dialine and supplemented bv random sampling from the Health Care Fi- nancine Administration of women aees 65and older. Controls were tTe- quency-matched to cases based on age (<50, 50 to 59, 601to 69; and >70) in a 2':1 control-case ratio: The\ met the same elicibilit\• criteria as cases for ace, residence, language, and tobacco use. A second control group selected consisted of femal'es from the ages of 20 to 79' who had a diagnosis of pri- man• carcinoma of the colon and met the same language, histor.• of cancer, residenca. and tobacco use criteria. This group was matched to cases by 10-\•ear aee groups and race. A multistep procedure was used to d'etermine lifetime smoking status. The medical record of each potential- Ir elieible lung cancer case and coloni cancer control \vas reviewed for in- formation on tobacco use. Patients identified in the medical record as current or former smokers were con, sidered ineligible. The physicians of IunQ cancer cases and colon cancer controls \\ ho were still considered po- tcntiallv eligible were then contacted for addinonal1 information on tobacco use. Potential studv subjects with lung cancer and colon cancer who \%,erc identified as current or former smok- ers by their physicians \vere consid- ered ineligible. This screening proce- dure not unexpectedly eliminated a larger proportion of lung cancer casess than colon, cancer controls. All rc- maining potennally eligible stud\• sub- jects (cases and both t\-pes of con- trols) were contacted by telephone t&elicit information onitobacco use. Women \\•ho reported smoking ?100' cigarettes or using any other form of tobacco for >6 months \&•cre con- sideredl1ineligiblk. At the ume of the intervle\\, these same questions \\ere repeatedlto confirm eachistud\• sub- ject's reported nonuse of tobacco:. About 4% of othenvise eligible \rom- en \+ithilung cancer \+•ere found to be nonsmokers after this multistep screening procedure. All of the interviewers \\•ere non- smokers to eliminate the unlikell- pos- sibilirn• of contaminating anv urine sample. At the time of the interview, a urine sample was requested by the interviewer from each studv subject. The specimens \\•ere stored at -20 CC until shipped to the American Hcalth. Foundation for analysis of cotinine and creatinine. This biochemical de- terminadon was used as a verification of current smoking status to exclude study subjects who \rere likely to be current>Iy active smokers. Overall,, about 2% of the study subjects were excluded from further analysis be- cause of cotinine/creatininc levels of ?100 ng/mg. This level was chosen to eliminate persons most likelk• to be active smokers and allow• for thc pos- sibil.ity of very high ETS exposures.° Of thc population controls that were tested, 0.8% of lung cancer cases, 2.6% of colon cancer controls, and 2% of population controls were de- termined to be ineligible by this cri- tenum_ The lower proportion ~ of cases with cotinine levels >100 ng/mg in our stud\+, compared to colon cancer controls and population controls, is likely to be the result of prior med- ical record reviews andiphysician queries used to screen out smokers from the lung cancer and colon can- cer control series. The information was more commonly available fr= these sources for lung cancer cases; therefore, a larger proportion of po- tential cases \vere exclud'edl as ineli- giblc prior to obtaining the urine specimen. The extensive procedures taken in this study to screen for to- bacco use minimizcd the misclassifi- cation ~ of smokers as nonsmokers. In addition to obtaining pathologic reports for each case, representative slides \\-ere requested from the hos- pital to confirm and uniformlv classi- ft` each case. Slides were available for 80`/0 of the cases and \vere reviewed by one pathologist who specialized in pulmonan• patihology: After re-ic\v,, 2% \cere found to he InellLible anl' \\•erezxcluded from the stud%. One rori the specific aims of the stud'\ \cas too evaluate the histologic specificln• of the ETS-lung cancer association hy examining the relationship for each of the main histologic types. A high proportion of the cases were adenu- carcinomas (around 75/o consistent across all stud\• centers); therefore; the report considered the association of ETS with all priman• lung cancers and priman- pulmonar\• adenocarcinoma. Results The estimated risk of lung cancer in nonsmoking women that: is associ- atcd with living \+ith~a spouse who smokc&\vas approximately 30°10, re- gardicss of which control l group was used in the comparison. An~increase in the risk of approximatel\. 50°ianras observed for adenocarcinoma of the lung compared to each control group. Separate analyses were conducted for subjects who personally responded and for those whose informadonwasn obtained from~ surrogatc respondents. The findings \,.•ere consistent for self- and proxy-respondents. -IJl odds ra- tios \.,crc adjustcdl for age, race, gco- graphi'r region, respondent n7 c, in- come, and education. An approxi- mate 30% risk of lune cancer associ- ated with spousal ETS exposure per- sisted after an additional adjustment was made for the consumption of vegetables (the most significant food or, nutrient factor), family histomy of lung canccr, and cmployment ini high-risk occupations or industrics. Household radon le\•cls W.ere deter- mined in a sample of case and controll homes under separate funding. Radonm levels are quite low in all of the areas included in our stud\•: < 1% of all homes tested had lc.x15 of ?4 pCi/L. The observed increased risk of lune cancer associatcd!\\itlt ETS cxposures is unlikel\, to result from confounding by radon,,diet, or other such factors. A positive dose response \vas ob~ scrvedl for all lung cancers and pulL monarv adcnocarcinomas With sc\•- eral measures ofspousal ETS, iriclud- ing dose, duration, and pack-\•cars. (~Table 1)i The trends \\•cre similar but tended to be higher for pulmonanY THE CANCER B CL'LE77X: \'i1L i'a; \c) d l. 1993

t Table 1. Adjusted Odds Ratios for All Lung Cancer and for Adenocarcinoma of the Lung Associated with Pbck-Years of Exposure from Spouse(s) (95% CI)' Pack-Years AII'Lung Cancer Adenocarcinoma of the Lung 0 1.0 1.0 <15 0.96 (0.72-1.29) 1.03 (0.73-1.46) 15-39 1.13 (0.8T-1.59)' 1.26 (0.85-11.87) 40-79 1.25 (0.86-1,81) 1.49 (0.98-2.27) >80 1.33 (0.68-2.58)1 1.70 (0.82-3.49), P.,.,,d - 0.07 P,...d < 0.01~ • Adjusted for, age, roce, study area, annual income, and education. q indicates confi- dence interval. Table 2. Association Between Risk of Lung Cancer and Adulw Exposures to Cigarette Smoke Among Nonsmoking Women: Adjusted Odds Ratio (95% Cl)' Exposure Source All Lung Concer Adenocorcinoma oFthe Lung Spouse 1.21 (0.96-1.54) 1.38 (1.04-1.82) Other household members 1.23 (0.97-1.56) 1-39 (1.05-1.82) Qtcupotional 1.34 (1.03-1.73) 1.44 (,1i.06-1.97) Social* 1,58'(1.22-2.04) 1.60 (1119-2.14) •Adjusted for oge,,race, study area, annual income, and'educotion. Q indicates confi- dence interval. adenocarcinoma than all lung cancers combined. Exposure to ETS from various sources dunng adult life was evaluated, in addition to spouse-relatcd expo- sures. Exposures to cigarette smoking from other household members, onn the job, and in other activities of adult Ilfe-ie, "social exposure" of }2 h/w,k from~ sources other tham occupati'onal or houschold-wcrc associated overall uith 40% to 60% significant elcvations in the risk of ad'cnocarcinoma of the lung and all lung cancers, combined (Table 2). Significant positive trends of increas- ing risk Ait}i increasing years of ex- posurc were foundin each exposure setting. No association was found between the risk otllung cancer and childhood exposures to cigars, pipes, cigarettes, or all tobacco npcs combined. These exposures ~%-ere limitcd'to the first 18 years of life, after which exposures wcrc attnbutcd to adult: lifc. Summary The findings of this study-which included methods to evaluate recall bias, minimize misclassification of smokers as nonsmokers, ensure accu- racy of diagnosis and clgssification of lung cancer, and adjust risk estimates for potential confound'ers-are con- sistent v.ith and extend the findings of numerous published reports that did not address all of diese issues. The overall 30% increased nsk associated with ETS exposure from a smokinc spouse is remarkably close to the ?5°o to 34%p csrimatcs of the cvaluauon off relevant studies in the 1986 report of the National Research~Council. The significant posiiive dose response to exposure to tobacco smoke within households, in~occupational settings, and in social scttings during adult lifc strongly supports an edologic role of ETS in lung cancer in nonsmokers and extends thc findings from the home into the workplace and public settings. References 1. Hira}•ama T. Non-smoking wivxs of heavy smokers have a higher risk oflung, cancer: a study from lapan.,Br.Ned J, 1981;282:183-185. 2. Trichopoulos D„Kalandidi L, Sparros L, McMahon B. Lung cancer and passive smoking. Inr. J Canccr. 198ll;27:1-4. 3. Environmrnta! Tobacco Smokc: Measuring Fxporurct and Assessing Health Tffccts, Committee on Passive Smoking; Board on, Environmental Studics and Toxicol- ogv; National Research Council. Wash- ington, DC: National Academy Press; 1986:1-337: 4. O'Neill 1K, Btunncmann 1LD, Dodct B, Hoffman D, cdi. Emironmcntal car- cinogens: methods of analk-sis and ex- posure mcasuremcnt., L4RC Sci Publ. 1981;9: 5. Fontham ETH, Correa P. Wu-«illiams A, et al. Lung cancer in nonsmokingg women: a multicenter case-control study. Cancer. Epidcmiot Biomarkcrr Prm. 1991; 1:35-43. 6. MattesemMD, Boyd G, Byar D; et aJ. Passive smoking on commerciallairline flights. JA1KA. 1989;261:867-872. 7. Fontliam ETH, Coatcs R Dillev A, et al. Lung cancer in nonsmokng «omen: dietary andoxid'ants. Cancer Epidcmiol N BiomarktrtPrcr: 1992;1:?SQ. Abstract. N W ~ N W a 94 THE C.iNCERBCLLET7X. VOLl45, No 1, 1993