Tobacco Documents Online

ERRATA Cancer Detection and'Preoention Voltune 14 / Issue 5, 1990; pp. 497-503 Dr. G. H. Miller, the author of "T'he Impact of Passive Smoking: Cancer Deaths among Nonsmoking Wornen" rcquested that a revised Abstract be printed to clarify several points. - ABSTRACT In order to obtain an estimate of the impact of passive smoking on cancer mortality, a retrospective study was conducted examining the cancer mortality of nonsmoking wives with no known or minimal exposure in contrast to nonsmoking wives with moderate to life-time exposure to tobacco smoke. The study was based on the data from 906 deceased nonsmoking women who resided in Erie County, Pennsylvania, who were divided' into the following three categories: I. No known exposure 2. Exposed nonemployed 3: Employed (assumed to be exposed to environmental ~ tobacco smoke in the workplace). The data were analyzed by the retrospective case-control method using cancer deaths as the cases and non-cancer related deaths as the controls. Also, the data from 401 smoking women were used for comparative purposes of the total percentage of cancer deaths among three groups: 1. Nonsmoking, nonexposed women 2. Combined nonsmoking unemployed and employed exposed women 3. Smoking women The major finding from the study are 1. Only (2.2%) of the total deaths reported among the nonsmoking women with no known or minimal exposure to tobacco smoke were due to cancer of any site i ., ~~ ~ 2. ^`No eass§'of !wt .cancer deaths werc_. ' -.ed for the` nonex report posed:'nonsrrtolung women, and eightaurig,cancer deaths;w~e. - ~- , . , ~. `'Farnong the nonsmoking women who were exposed to passive smo{ting.'Also, for this small group of 179 nonsmoking nonexposed'women„ there were no reported cases of breast cancer~ genitourinary or lymphatic cancer. 3. Employed nonsmoking women experienced proportionately more cancer, deaths (34.3%) than both nonexposed (2?%) and exposed nonemployed wives (18.9%). The combined groups of exposed nonsmoking wives (nonemployed and employed) contracted' 25.5% cancer deaths.. 4. Age-adjusted data showed similar trends. 5. Cancer death rates for women smokers was 35.5% of the total deaths of women smokers. Public health officials should consider requiring that the workplace be free from tobacco smoke since these data imply that passive smoking has a very detrimental effect upon nonsmokers. Also, smokers should be made aware of the potential dam- age they inflict on others in their home as well as the workplace. 688 . Volume 14, Issue 6

Cancer Detection and Prevention The Impact of Passive Smoking: Cancer Deaths among Nonsmoking Women G.H. Miller, Ph.D., CPC ABSTRACT In order to obtain an estimate of't6e impact of passive smok- ing on cancer mortality, a retrospective study was conducted examining the cancer mortality of nonsmoking wives with no known or minimal exposure in contrast to nonsmoking wives with moderate (up to 19 years) to lifetime exposure to tobacco smoke. The study was based on the data from 906 deceased nonsmoking women from Erie County, Pennsylvania, who were divided' into the following three categories: 1. No known exposure 2. Exposed nonemployed wives 3. Employed wives assumed to be exposed to environtnental 1 tobacco smoke in the workplace The data were analyzed by the retrospective case-control method using cancer deaths as the cases and non-cancer related deaths as the controls. The major findings from the study are 1. Only 2.2% of the deaths reported among the women with no known or minimal exposure to tobacco smoke were due to cancer of any site. 2. No cases of lung cancer were reported' for the nonexposed women, and eight lung cancer deathc were reported among the nonsmoking women who were exposed to passive smoking. 3. Employed women experienced proportionately tnore can- cer deaths (35.3%) than both nonexposed (2.2%) and exposed nonemployed wives (25.5%). 4. Age-adjusted data showed' similar trends. Public health officials should consider requiring that the workplace be free from tobacco smoke since these data imply that passive smoking has a very detrimental effect upon non- smokers both at home and in the workplace.. Key Words: passive smoking, lung cancer,, breast cancer. I. INTRODUCTION Whether or not passive smoking is detrimental to health has been extensively considered in the last decade. The first reports on the effects of long-term exposure to passive smoking ap- pearedin the late 1970s.'" Prior to that tiine, it was generally assumed that passive smoking was not of much consequence. However, most health professionals were aware of reports of moderate-to-severe eye irritation as well as allergic reactions to tobacco smoke.'" Then: were also reports of the serious health consequences observed in animals after exposure to high concentrations of tobacco unoke.' In the 1970s, research re- ports discussed the effect of increased respiratory diseases in children of smoking parents.I These were followtd by the 1980 Surgeon General's report on the hanmful effects of cigarettes on the health of smoking women„ including research reporting lower birthweights among children of smoking morhers.' Pas- sive smoking has been associated with deleterious effects on the fetus.'-s In 1978' and 1979,' at the Third W orld Conference on Smok- utg and Health, Miller''0 proposed a passive smoking classi- fication based on three different exposure levels: (1) short-tetm - a few minutes to a few hours; (2) moderate-term - less than 2 decades; and (3) long-term - 2 decades to a lifetime. He suggested that studying the consequences of long-term ex- posure provided the best opportunity for observing the effecu, if any, of passive smoking. He reported a 4-year earlier av- erage-age-at-death of nonsmoking women exposed totttei>• hus- bands' cigarette smoke compared' with nonsmoking, nonex- posed women. White and Froeb9 in 1979 reported severe lung dysfunction following long-term exposure to tobacco smoke. Hirayama,10 Tricopolous et al.," and and Correa et al.'= noted a two- to threefold increase in lung cancer in exposed wives when com- pared to nonexposed wives. Gilli's et a1i13 and' Schmidt etal." have also shown that passive smoking is detrimental to the health of the nonsmoker. Miller,'s Sandler et al.,16 and Repace and' Lowrey" each showed a two- to threefold increase in total cancer deaths following long-term exposure to passive smok- ing. Garfinkel et al;," who had previously found no statistically signi6canrevidence of the detrimental effects of passive smok- ing, now reports a two- to threefold increase in lung cancer.'v Wald et al'.20 and Pershagen et a1.21 most recently have provided support for the hypothesis that passive smoking increases lung cancer incidence, largely among spouses. Two reports provide a detailed review of the research on passive smoking, noting its detrimental effects: the Surgeon General' recommended a smoke-free environment, and the National Academy of SciencesZ' recommended no smoking in the home environment of chil- dnen: In order to obtain additional information on the effects of passive smoking, this study compared different levels of en- vironmental tobacco smoke exposure of married nonsmoking G.H. MHler, Ps.D.. CPC, Studies om Smokin`.,tnc., 125 High Saeet: Edinboro. PA 16412. I 1990 497 • .. . . . _ ... s ~

The Impact of Passive Smoking: Cancer Deaths among Nonsmoking Women 11 women with cause of death. The causes of death include& in the study were the usual causes of death such as cardiovascular. cancer, and respiratory diseases but excluded traumatic deaths such as accidents. suicides, and homicides. Men were not included in this studv since so many men are exposed tc en- vironmental tobacco smoke in the workplace. II. MATERIALS AND METHODS A. The Population The population of Eraie County. which includes Eric. the third largest citv in Pennsvlvania. was 263.654 in 1970. It is primarily a middle income population (the average family in- come for 1970:was listed as S9:380) with a low migration rate of 7% for the 1950 to 1970 time period as reported by the Pennsylvania Department of Commerce. 8.Methodoloqyi The Northwestern Pennsylvania Study on Smoking and Health (NPSSH) began to gather data in 1973 on the smoking habits of deceased male and female residents in Erie County by in- terviewing the decedents' next of kin. Death notices for the years 1972 through 1982 were obtained from the local news- paper. which lists the names of approximately 95% of deceased Erie County residents along with information on surviving rel- atives. Telephone numbers of one to three surviving relatives were obtained for approximately 85% of the death notices reporte& in the newspaper. Deaths from accidents, suicides, congenital anomalies, and persons younger than 30 years of age were not ineludedbecause an age bias is introducedwhen those classified in these categories have their lives curtailed and therefore would not provide an accurate estimate of the age at which cancer actually occurs for the passive smoking-exposed women. Be- cause of the reduced life expectancy. the inclusion of accidents and suicides would, lower the incidence of cancer since those in the lower age cohorts might have contracted some type of cancer had they lived long enough. These individuals com- prised about 10% of the total deaths. A questionnaire designed for telephone interviews was con- structed with the assistance of the local branches of the Amer- ican Cancer Society. the American Heart Association, the American Lung Association, and by smoking and health ex- perts in the Pennsylvania Department of Health. A more de- tailed description of~ this questionnaire has been published.'s lnterviewers explained the purpose of the study to the iden- tified surviving relatives and solicited their cooperation. In- formation was collected from them on each decedent's cause of death, age, occupation (including inforrnation on whether or not the wife worked outside the home), exposure to known sources of pollution (including environmental tobacco smoke), smoking history, and whether or not the spouse and parents smoked. The interviews were conductedlb% thr dir«tor it the studv and by interviewers trained bc him. The questionnaire was revised in 1975. and :,Jdttion,,l items were included to obtain more complete informatiom , in ., ~rx)u.r " smoking habits such as type and quantit% ut none:,:o u,r.J. details on smoking cessation. the age at the time or de:uh ,,rn1 the cause of death. the current age if livtng. the trur 1r ,lei.,JC of death if deceased. Information on whether an, other mem- bers of the household smoked was also added. Detailed data on passive smoking from the re% i.ed yut,- tionnaire were gathered on the decreased from, 23 months in the years 1975. 1976. 1979. and 1980: these aere the unl% data considered in this study. Due to logistical problem.. the interviews for deaths occurring in other months were not :nm- pleted at that time. The totaGnumber of deaths amon¢ re.idents of Erie County for these 4 years (Pennsylvania Department ofi Health) was 10.131 (5478 men and 4653 women). Among the 3538 relatives contacted. 3361 (95%) provided information on 1863 men and 1.198 women. Of the 1498 deceased %k omen for whom information was obtained. smoking exposure hi,toriass were available for 1423 - 906 nonsmoking wives. who are the subjects considered in this paper. plus data on 401! smokers. Of the remaining 116 deceased nonsmoking women. 63 ne.er married and 53 did not have sufficient information on passive smoking to include in this study. Fon example. the statement that a husband was a smoker was not considered, sufficient evidence if details on amount and type of tobacco use wore unavailable. resulting in the exclusion of the wife trom this study: These data are summarized in Table 1. In this study. a nonsmoker was defined as a person w ho had' smoked fewer than 20 packs of cigarettes during his or her lifetime. A nonsmoking. nonexposed wife was one `who had no known reported exposure to tobacco smoke from any source or minimal exposure (only occasional exposure to pipe. cigar. or cigarette smoke). A nonemployed exposed wife was one who was exposed to cigarette smoke by a family member Table I Total Deaths from All Causes, Totali Cancer Deaths, and' Number of Intervtews from the Erie County Population Data Base for the Years 1975, 1976, 1979, and 1980 Intervkws Intervkws Atl All Totat passive used in esuses eaecer Interviews smokln` study Men 5.478 1.116 1.863 1.793 Women 4.653 995 1.498 1.423 906' Total 10.131 2.111 3.361 3.216 • This figure includes only, twnsnwtcing women with detailed informaoion on passive smoking exposure. It excludes 401 smotting wives and 116 who were never married. single. or of uncertain passive smokmg sutus. 498 Volume 14, Issue 5

Cancer Detection and Prevention. (husband. children, or relatives) or a nonfamily member. or had reported long-term exposure to srnoky rooms in non-em- ployed activities outside the home. An employed wife wass identified by a surviving relative as a person engaged in an: occupation outside the home for more than 3 years. In order to insure a clearly defined category for nonexposure. all those who may have had any potential moderate-to-heavy exposure were eliminated from ahis category. Thercfore., those not included in the nonexposed spouse category, in addition to the 401 smoking women, were (I) women who were married at any time to a smoker; (2) women exposed by family or friends or exposed in nonemployed activities; (3) all non- smoking employed women. Although some nonexposed~ individuals might have beem eliminated by these definitions, the nonexposed category as defined here is as free as possible from the inclusion of those who may have been exposed ~ for longer periods of time. The nonsmoking wives were categorized in the following three groups based upon the information reported in the telephone interviews with relatives: 1. NX - nonexposed (no known or only minimally ex- posed) wives 2. ENE - exposed nonemployed wives 3. EW - Employed wives who were assumed to have some exposure at work The data were analyzed using the retrospective case control techniques described by Fleiss.2' The "cases" were defined as deaths (among nonsmoking wives) due to cancer of any kind while the "controls" were noncancer deaths among nonsmoking wives. These deaths included cardiovascular. respiratory, kid- rtey, and other noncancer diseases, but ~ excluded traumatic deaths such as accidents and suicides. The classification of the cause of death as recorded on the death certificate was provided by the ICD (International Clas- sification of Diseases) codes listed on computer printouts pro- vided by the Pennsylvania Department of Health. The causes of death reported by the next of kin were not used since the ICD codes are considered more accurate. However, the infor- mation provided by the surviving relatives was very close to the ICD codes when comparison was made between the next of kin's report of the cause of death given during the interview and' the ICD (Death Certificate) disease designation. Details on these data will be reported separately. In order to be sure that all those classified as having no known exposure to carcinogenic compounds of tobacco smoke were classified correctly, very close relatives of the deceased were located ~ and interviewed if the first interviewee was not a close relative. or close relatives were interviewed again to obtain greater accuracy in the classification of the nonexposed. During this interviewing, it was discovered that two women classified as nonexposed who died of breast cancer were ac- tually exposed to cigarette smoke by their husbands and chil• dren.,This reducedrthe total cases of canceramong nonexposed women from six to four. Because of the small numbers with different types of cancer, only the major cancer categories were considered as noted in the standard International Classification of Diseases: Oral (140-- 149). Digestive tract (150-159), Respiratory (1160-163): Breast (l74)„ Genitourinary (180-189). Lymphatic (200-209): and All Other Cancer Sites. Although specific details on cancer deaths due to smoking are not considered in the present study, a comparison was made with the nonexposed (NX). the exposed nonsmokers (ENE and EW). and smokers (employed and nonemploved) to provide comparisons betwe.en ~ the nonexposed. exposed nonsmokers.. and smokers. In order to check for potential bias due to age. the data were age adjusted for all cancer causes and4or the major categories. of cancer:: oral, breast. lung. respiratory. genitourinary. an& lymphatic. 111. RESULTS Analysis of the data in the two by three contingency table (tTable Il--two categories: cancer cases and noncancer cases compared with the three categories: NX (nonexposed wives). ENE (exposed nonemployed wives), EW (employed wives)j Table II Deaths among Nonexposed and Exposed Nonsmoking Wives Cancer '.Yoneancer Total Percent cancer deaths deaths deaths deaths Plonexposed. 11t nonemptoved 4 175 179 Exposed. , , (2) nonemployed 78 334 412 18'9 t3i Employed log 207 315 }4,3 Total 190 716 906 21 0 Chi, square for above 3 by 2 matnx 12 DF1 - 72.64 Chi square evaluations for'_ by 2 compansons i 1 DFi Odds Cht ratio square cl ivs i2t iNonexposed vs exposed 10:: 290 nonemployed wi%esi ~ i l Ia 43) INonexposed Vs. emplo.ed ='.8 6h' u wivesi: 121 vs (3t (Exposed nonemploved wives -._ 3 vs. employed wives) 111 vs INonexposed vs. exposed IF.0 1t'_) - 4311 nonemplo.ed and employed wives) 1990 499

The Impact of Passive Smoking: Cancer Deaths among Nonsmoking Women showed a Chi Square value of 72.64 with 2 degrees of freedom (p <€0.01D. Table II also shows four different comparisons of the three subgroups: (l ) nonexposed wives (:VX) with exposed nonem- ployed wives iENE): (2) nonexposed' wives (NX) with em- ployed wives iEW): (3) exposed nonemployed wives (ENE) with employed wives (EH'): and(1) nonexposed wives (NX) ,l compared with the combination of exposed nonemployed wives (EtiE) and employed wi.:es (EW). J " data;from Tabl'e ~ 'kttow that exposure to envitonmenta ~ ' " asso-~ s ~7~' a,"~y..,-w~r".~'~ .ti•ai . ra:~E ~~' V.. ip4,1~^'"~ . tam ~tatoir"`wlt}~ Inc~a~ed" cance mo a s hn 1 an od'ds t ratto$~C3'~ tten'"c m pang nonexRose =on~e~ "' ployev`&Mves"an odds4ratio of 22.8 when comparing nonex-, posed"tviVef wtth emploved wtves,-an odds ratto of 2 2 when . companlf~ e>lpc3sid`eonemployet'`wi'Tie e )'~by~d w ve's ` andaii~`rattqof' 15:0_when companngttonexpos%c~w'es~ with exposed nonemployed wives and employed wi've.sw r Proportionally far fewer nonsmoking noneXposed wives died of cancer than the exposed wives. The comparison of the ex- posed nonemployed wives with employed wives also showed' highly significant excesses in cancer mortality among the em- ployed women. Table IIl shows the mortality from cancer by primary site for the three different exposure groups (NX - nonexposed wives: ENE - exposed nonempioyed wives, and EW - em- ployed wives) for themajor categories of cancer (oral, digestive tract. breast, genitourinary: lymphatic, and other sites). There were no reported deaths from oral. lung, broast, genitourinary. and lymphatic cancer among the 179 deaths within the nonex- posed women. One death from digestive cancer and three deaths from "other" cancers were reported for the nonexposed. Eight deaths due to lung cancer were reported among the 727 deaths of exposed (nonemployed and employed) nonsmoking wives. Figure 1 shows the percentage of deaths due to cancer among nonexposed wives, exposed wives (both nonemployed and em- ployed), and smoking wives. Although smoking wives are not considered in detail in the present study, the preliminary data are provided here for comparative purposes: I. Nonexposed' nonemployed wives - 2.'_17c 2. Exposed (employed and nonemploved) wi+es -_5.5cic 3. Smoking wives (employed and nonempltiycd) - 35,39, In addition, the data for cancer deaths were age adjusted by the standard age-adjusting methods. The cancer deaths were adjusted to cause distribution of all female decedents in the years considered in this study (1975. 1976. 1979. 1980) from Erie County based on the data from the Pennsylvania Depart- ment of Vital Statistics. The results of the expected and ob- served were analyzed by the two-way probability based on the Poisson distribution and are reported in Table IV: The Poisson distribution was used since the nonexposed' group had so few cases. Table IV shows' the results after age adjusting for all cancer cases, digestive cancer. and the results for the combi- nation of orall lung. breast. genitourinary. and lymphatic can- cers which appear to be associated with effects of passive smoking. IV. DISCUSSION While many epidemiologists prefer to conduct prospective studies. the retrospective study has the advantage of allowing one to obtain estimates in a short time. Thus, this type of a research provides mortality data related to information on the smoking habits of the deceased. Because nonsmoking women may be exposed to numerous sources of tobacco smoke. this study tried' to eliminate as many of the sources of tobacco smoke exposure as possible in order to more accurately classify women into a "pure" nonexposed category. It is probable that persons in this "pure" category have had some small exposures to tobacco smoke during their lives, because minimal exposure is difficult to avoid. The present study was designed to obtain smoking history data on all nonsmoking wives who died during the specified years. While it is difficult, if not impossible, to obtain data on all members of a population, there is little reason to believe that the 23-months' sampling of the 4-year population was Table Ill Major Types of Cancer for Nonexlposed and Exposed Nonsmoking Wives Type of nncer Oral t>ti{nt1.e trsct inpirawry { B.east Genitoarinary LymObatJc cllher tdtn Total AveraBe age at dat6 ICD clusificuion 140-149 150-159 160-163 174 174-185 200-209 NX' 0 1 0 0 0 0 3 4 84:5 ENE 0 28 3 14 9 a 16 78 71.1 EW 0 38 5 23 14 10 16 10/ 67:8 Taul 0 67 8 39 23 18 35 190 Noae: ICI). International classification of disease; NX:' nonexposed M ives: ENE: exposed nonempbyed wives; EW: employed wives: Total: exposed and nonexposed'wivea. Volume 14, Issue 5

Cancer Detection and Prevention srrrt er t~trtis FIGURE 1. A companson of mortaliy rates of smoken. passive smokers. and the nonexposed. Table IV Analysis Using Age-Adjusted Values for Cancer Sites (All Cancer, Digestive, and Passive Smoking- Associated Cancer) for the Three Major Nonsmoking Groups Diseasc Group EXP OBS Probability All cancer Nonexposed 24.0 4 0.00002 Exposed nonemployed 73.1 78 NS Exposed employed' 70.6 108 0:001 Digestive Cancer Nonexposed 9.3 1 0:002' Exposed nonemployed 24.2 28 NS Exposed employed 20.4 38 0.01 Pusi.+e smoking- usociated cancer onexposed .2 0 .0002 Exposed nonemployed 35.7 34 NS Exposed employed 38:9 54 0.1 biased in the types of cancer deaths that would' have occurred during these years. Analysis of the data of previous nonrcs- pondents showed the same basic distribution of cancer deaths as the analysis of original respondents. Therefore, the data from this population study are likely to be representive of the actual population. ~ther smokers in the household, or in outside activitiesl Pas 2' higher probability of dying of cancer of all kinds n women *'h0 have-no known exposure. 1 e Cit~C~elµ>17ai~~)n,rlOli~ , ~smoking Eric County wives with no known exposu're to t~o'b"fae/'co smoke is very low _.-only 2.2% vs. 25.55rr~qr exposed women and 35.3% for smoking wives (Figure I/: ~ The data imply that the absence of passive smoking could lead to even lower rates of cancer mortality for the total pop- ulation. Thus, these data tndicate that passive smoking is vers; sdetrimenaQu~~alteDf`ttLRo sinokef`".";'- 4r . The total proportion of deaths due to cancer for the 401 smoking and 906 nonsmoking wives in this studN is 23.9rK. This result is close to the proportion of deaths.due to cancer for women in Erie Counry, (21.8%) for the years considered in the study as reported in the Pennsylvania Department of Healtti- Division of Vital Statistics. This agreement sug¢estss that the data are representative of the total population. The observation that the nonsmoking employed wives had an odds ratio of 2.2 of dying of cancer when compared with exposed nonemploycd wives suggests that workplace expo- sures may involve important hazards in addition to. or apart from, tobacco smoke. Over 80% of the employed wives in this study were either office workers„ in sales occupations. teachers. or nurses, who (aside from the nurses) should have had little contact with carcinogenic agents other than tobacco smoke.. The doubling of the rates of cancer for employed women com• parcd with nonemployed women are in agreement with the data of'Rcpace and JAwr}•'-3'and'the estimates made by V4'ells.'O It may be thar work place passive smoking exposure is higher than at home. Another possible explanation is that the em• ployed wives may receive additional tobacco smoke exposure from smokers at home such as from their spouse. i.e.. their husbands may be more likely to be smokers or heavy smokers. Also of interest is the observation that onlv' 8 deaths due to lung cancer were reported among the total of 727 passively exposed nonsmoking wives. Thus passive smoking does nott result in large numbers of lung cancer cases when compared to active smoking. This difference is in agreement with the results of several studies by Hirayama.10 Tricopoulos et al.." and Correa et, al.1e showing two- to threefold increases in lung cancer for exposed vs. nonexposed wives and a much larger increase for smokers. In addition. a study by MilJer on the Amish - a nearly completely nonsmoking population - showed no lung cancer cases among either male and female nonsmokers for the Lancaster County: Pennsylvania. Amish for the 1970 to 1980 decade. There were no reported deaths from lung cancer. breast' cancer. genitourinary.or lymphatic cancer among wives with no known exposure to tobacco smoke among the total ofi 179 deaths in the nonexposed category: This contrasts with the distribution. of cancer mortality for women in the general population of Erie County for the years under consideration: lung cancer - 12:7%: brcast cancer - 20.7 K. and lym- a _Tlese daii..DrQVIQe6ldeMeO't)Iatolt~flStaft n ex• smo fi rl~lated°Eb"5t'!~_ W)&.'"ig t990 501

The Impact of Passive Smoking: Cancer Deaths among Nonsmoking Women phatic cancer - 8.7%. These Erie County percentages of cancer causes by site are close to the national averages. Thus it appears that the types of cancer reported in the nonex- posed group are very different from the types of cancer re- ported for, both passive smoking and~ active smoking groups. While ir has been shown that passive smoking causes some lung cancer ca,es. the amount is small compared with that of those who smoke. Whether or not breast cancer is caused by active or pas- sive smoking has become a controversial issue. Previously. it was thought that breast cancer was not active-smoking re- lated. For example. no editions of the U.S. Surgeon Gener- al's Reports from 1964 to 1989 show any subsections de- voted to active smoking and breast cancer. However, the Bibliography of Smoking and Health and Index Medicus provides reports of many studies on the topic of smoking and breast cancer. The most recent studies appear to be equally divided between those concluding that there is no overaff positive association with active smoking and breast caneer'1-" and those that conclude a positive and, in some cases, a sig- ni6cant association between active smoking and breast can- ,-_,s cer.•- It has been onlv recently that studies have taken into account the potential effect of passive smoking and its possible asso- ciation with breast cancer. Two of the mosu recent studies were completed by Sandler et al.J6 and'Honon." Two other major cancer categories have been reported as being associated with passive smoking exposure - lymphatic cancer, by Wells:=° and eervical cancer. by Slattery et al.a' The data from the present study agree with the studies noted above. The possibility of passive smoking being associated with cancer sites was also verified by the highly significant values obtained after age adjusting for total cancer, digestive cancer. and the combined' passive smoking-related cancer sites (refer to Table IV). Since many studies during this decade have come to the conclusion that breast cancer is not smoking related, this new data might appear to be questionable. However, past studies failed to allow for such passive smoking categories as employed women and others smoking in the household. Therefore, the results of the NPSSH study should provide a more accurate estimate of passive smoking exposure. A recent article by Bailar and Smith," pointing to the continued increase in cancer mortality despite our apparent gains in early identification, as well as improvements in the treatment of eancer, shows that cancer is still increasing in the population. It may be that the cumulative long-term ef- fects of both passive smoking andactive smoking are con- tinuing to have an impact on this increased cancer mortal- ity. These issues have important implications for both the home and the workplace. More research is needed to validate these observations. New studies require: I. Better estimates of exposure to passive smoking. The nonsmoking category should exclude emplo}ed~women. women exposed at home by individuals other than their spouse. and long-term exposure in outside nontmplo-~ed activities. as well as exposure at home by their spouse. 2. Long-term exposure information - preferably a mini- mum of 3 decades-to-lifetime exposure. Studies tmolv- ing a few weeks or months or even a few years of'exposure are inappropriate for an analysis of the true long-term impact of passive smoking. 3. Large enough samples to provide valid data. For comparative purposes with the results of this studv, the author would like to see more population studies instead of nonrandom or nonpopulation based samples in order to over- come the biases that are inherent in these types of sample designs.. ACKNOWLEDGMENTS The author is extremely grateful to Dr. Marvin Schneider- man of the National Academy of Sciences for his statistical and editorial assistance, to Dr. Robert DePue for his complete computations on the age adjustment of the data, and to Charles E. Chittenden, Dr. Victor Hawthorne, and Dr. Thomas No- vomy for additional editorial assistance. Partial assistance for this study has been made possible by grants from the ERIE COMMUNITY FOUNDATION'and the ITT LIFE INSURANCE COMPANY. And thanks to the many thousands of cooperating relatives who made this study possible. REFf REN'CES I. 2. The Health Consequences of Smoking - A Repon of the Surgeon General. U.S. DHEW. 1972:119-138. The Heahh Consequences of Smoking - A Report of the Surgeon Gened: Chapter IVytnvoluntary Snwkin8): U.S. DHEW. 1975:83- I12. 3. Smoking and Health - A Report of the Surgeon Gcnerai: Chapter 4. 11, (Involuntary Srnotun8), U.S. DHEW. 1979:11-3 to 11,-40: The Health Consequences of Smokinlit for Women - A Report of ~ S. the Surgeon General. U.S. DHEw. l980:189-239. Geeenberj ItA, F-ttel RA. Haley NJ. Exposure of the fetus. neonate. and' mused iafantm nicotine and continine from maternal smoking. N Fa81 J Med 1984: 311(10):67Z. W W 6. Paniv Roekning och smaa barn - Nya ForslwinYsroen (Passive 7. smoking and the infant - New nesearch findings)'- Editorial - Tobakken Och Vi 1984; 29f3):7-9. Miller GH.1Te Pennsylvania Study on Passive Smokin8.1Breathin8 N ,~.. (Illinois Lung Assoe. - Sprin88eW. IJ-L.) 1978: 41151'S-9. >. 8. Fourtli World' Conferenoe on Smoking and: Health - Atisuacu - ~ 9. 1979. Stockholm. Sweden. White JR. Froeb HF. Small-airways dysfunction in nonsnaken chronically exposed to tobacco smoke. N En81 I Med' 1980: 302:720- 723. 502 Volume 14, Issue 5

Cancer Detection and Prevention 10. Hirayama T'. Non-smoking wives of heavy smokers have a htgtier risk of lung cancer. A study from Japan. Br Med 1 1,981'Jan, 28.:18 ti- 185. 11. Tricopoulos D. Kalandidi, A. Spaaos 1- McMahon B. Lung cancer and'passive smoking Int l Cancer, 1981 Jan; 27:1-4. 12. Correa P. Pickle LW. Fontham E. Lin Y. H>tertsul W' Passi.e smoking and lung cancer. Lancet 1983; 2:595-597. 13. Gillis CR. Hole DJ. Hawthorne VM. Boyle P. Environmental tobrc;o smoke. The effect of environmental tobacco smoking in two urban communities in the West ofScotland. Eur 1 Resp Dis 1984.65 i Suppl 133):121-126. 14. Knoth A- Bohn H. Schmidi~F. Passive smoking as a causal'tactnr of bronchial carcmorm in female smokers. Med Klin 1983: 78t. t 6a- 69. 15. Miller GH. Cancer. passive smokinp and rtonemployed and empb.~ eJ wives. Western 1 Med 1984: 140:632-635. 16. Sardler DP. Wilcox AJ. Everson RB. Cumulative effects of lifetime passive smoking on cancer risk. Lancet(6424),1985: 1:313-14 . 17. Repace JL. Lowrey AH. A quantitative estimate of nonsmokers' lung cancernsk from passive smoktng.,Environ Int 1985; 11:3-32. 18. G.rfink'eI L. Time trends in lung cancer mortality among nonsmoker.m with a note on passive smoking. JNCI 1981: 66:1061-1066. 19. Gtufink'el L.,Auerbach O. Joubert L. Involuntary smoking and lung cancer A case-control'stud%. JNCI 1985; 75;463-469.. 20. Wald NJ. Nanchafial K. Thompson SG. Cuckle HS Does breathtng other people's tobacco smoke causes lung cancerT Br Med J 1986. 293:1218-1221 . 21'. Pershagen G. Zdenek H. Svensson C. Passive smoking and lung cancer in Swedish women. Am 1 Epidemiol 1987; 125td 1:17-24 22. The health consequences of involuntary smoking - a report of the SurgeonGeneral. Washington, DC: Department of Health and Hu- man Services. 1986. 23. National Academy of Sciences: Environmental tobacco smoke - Measuring exposure and assessing health effects. Washington. DC Nuional Academv of Science Press. 1986. 24. Fktss JL. Statisttcal Imethods for rates and proportions. New York - Wilev. 197153-66. 25. Repace JL. Lowrem AH. Modeling exposure of nonsmokers to am- bient tobacco smoke. Ann Air Pollut Control Assoc 1983. 26. Wells AJ. An estimate of adult rnorsalitv in the United States from passive smoking. Eroviron Int 1988: 14249-265. 27. Miller GH. Lung cancer; A comparison of the incidence between~ Amish and non-Amish in Lancaster County. l/ndiana Med Assoc 1982: 76t21:121-124. 28. Bnnton LA: Schaircr C.Stanford JL. HooverRN. Cigarette smoking and breast cancer. Am J Epidemiol 1986 April: 12384e614-622. 29. Hiatt RA. Fireman BH. Smoking- menopause, and breast cancer. 1NCl 1986 May: 76t5c833-838. 30. Stockwell HG. Lyman GH_ Cigarette smoking and the risk of female reproductive cancer. Am 1 Obstet Gynecol:l987 July; 157f il 1:35-40: 31. Adami HO. Lund E., Bergstrom R. Meirik O. Cigarette smoking. akoliol consumption aod the risk of breast cancer in young women Br J Ctmcer. 1968 December: 38(6):1832-837. 32. Schechter MT. Miller AB. Howe GR. Cigarette smoking and breast cancer a case-control study of screening program participants. Am 1 Epidemiol 1985 ApnL 121t4e479-487. 33. W-atnnatie S. Ochi H. KobayuhiY. Tsugant S. Anrnoto H. Kiugawa K. Frequency of multiple primary cancers and risk factors for lung and breast cancer patients. Int Symp Princess Takamatsu Cancer Res Fund 1987: 18:275-28, 34. DantelliHW. Increased lymph node metutases at mastectomy for breast cancer associated with,host obesitf. cigarette smoking. age., and large tumor size. Cancer 1988 1July IS: 62/2t:429--f35. 35. Brownson RC. Blackwell CW. Pearson DK. RtynaldsRD. Riohen> JW 1r. Papermaster BW. Risk of breast cancer in rclawun t., ci.aronr smoking. Arch Intern Med 1988'Januarv: 149111 l4u-las 36. Sandler DP. Everson RB. Wilcox AJ. Pusive smoking in sdultho,xl and cancer~ risk, Am J Epidemiol' 1985: 121 t 37. 37. Horton AW. Indoor tobacco smoke pollutton - A maior rr.A tactur for both breun and lung cancer. Cancer 1988: 6211 i,6-1-t 38. Slanery ML. Robison LM. Schuman KL. et al. CiFarene .mal;ine and exposure to passive smoke are risk factor, for centc,l ;ancer JAMA 1989: 261' 1593-1598. 39. BailarlC III. Smith EM'_ Progress againsncancer' tiew Ensl I Ved 1986: 314i J 91:1226-1232. 1990 503