Tobacco Documents Online

Time Trends In Lung Cancer Mortality Among Nonsmokers and a Note on Passive Smoking 1 L.awr.nce Gatflnk.l,, M.A. 2.3 ABSTRACT-Lunp uncer mortNity rates wer* Computed for nonsmokers in the American Cancer Societys prospective study for three 4-year periods from 196o to 1972 and in the Dorn atudy of veterans for three 5-year periods from 1954 to 19%: There was no widence of any trend in these rates by 5-yeer ape groups or for the total groups. No time trend was observed in nonsmokers for cancers of other sNaaed sitas except for a decrease in cancer of the uterus. Compared to nonsmoking womin married tononsmokinq husbands,., nonsmokers married tosmokinp hus- bands showed very little, it ar+y, increased risk of lung canoir.- JNCI 1981: f16:1t]61 -1 p6f3:. Mortality rates from lung cancer in men in the United States have been rising steadily since 1:930 (the first year these cancers were classified separately),and in women since the mid,1960's. It has generally been accepted that the major reason for the increase has been the cigarette smoking patterns which began in young men around' World War I and in young women in the 1930's and 1940's.. A large body of evidence from epidemiologic and pathologic studies on smokers con- firms this conclusion (1). A recent estimate of the percentage of cancers attributable to smoking in men was 34.5% for totali cancers and 82.8% for lung cancer. In women the comparable percentages were 5.4 and 43:1% (2). This analysis was based on data from the large epidemiologic study of the ACS and covered the period 1967-71. It was based on a number of assump- tions that would give slightly different figures if the smoking distributions in the study population differed from those of the general populauon or if smoking distributions changed in the late 1970's compared to the late 1960's (as they indeed have in women): There has been a suggestion, however, that the lung cancer trend in nonsmokers has also increased in the United States over the years. Enstrom (3) stated that "a more complete understanding of lung cancer etiology is needed." This analysis indicated a large relative increase in lung cancer mortality in nonsmokers in both white men and white women between 1914 and 1975 on the basis of an interpretation of data 'un samples of national mortality statisucs and several epidem'sologic studies in different periods of time (3). Enstrom recognized that most of the increase occurred between a 1914 survey; of death registration areas in 24 states and national mortality statistics reported in 1935 and that most of that increase was probably attributable to incompleteness of reporting lung cancer and' to changes in diagnostic criteria. Nevertheless„ the possibility exists that lung cancer is increasing in nonsmokers who have had increasing exposure to other factors-occupational exposures, MOTfCE This materiali may be protected by copyrigM Ilw (fiqa 17 U.S. Codej; general air polluuon, and perhaps even to passive smoking (inhaling the smoke from smokers). Even if these factors were related to the alleged increase in lung cancer, they could have had only minimal effect on the upward trend for lung cancer in men, since the mortality rates among smokers and nonsmokers differ so greatly. Moreover, in the last 50 years and until recently, most men had a history of cigarette smoking. Among women lung cancer rates remained low up to about 1960. Since then, there has been a threefold increase in rates attributable in large part to the changes in smoking patterns among women during the preceding two or three decades. In this paper, infotmauon is provided on trends for lung cancer (and cancers of several other sites) in nonsmokers over a 12-year period (1960-72) from data in the prospective study of the ACS. In addition, data for nonsmokers from the Dorn study of veterans for the years 1955-69 are given. While such data do not provide evidence over a very long time span, they are based on the two largest prospective studies in the United' States and cover a 1734-year period from 1955 to 1972. MATERIALS AND METHODS Procedures in the collection of data in the prospec- tive study of the ACS have been presented in a number of publications (4-6). There were 94,000 male and 375,000 female nonsmokers at the start of the study. In the ACS study, a"nonsmoker"' is one who reported he or she had never smoked or smoked only occasionaTly but had never smoked' regularly. Classification was made as of the start of the study, and very few nonsmokers reported that they staned to smoke on any of four later questionnaires. Enrollment of subjects in the ACS study began in October 1959 and extended through March 1960. Fol- low-up was complete for 98:4% of all subjects through June 1971 and 92.8% complete for the 12th year of the study. Deaths were reported by the ACS volunteers, and death certificites were obtained from state health de- AaanavuroN us[n AGS -American Cancer Society. Received October 23. 1990; accepted January 26, 1981. 3' Departtnenr o(' Epidemiotogy and Suusucs. Amencan Cancc Society, 777 Third Ave.., New York, N.Y. 10017. ' C thank Eugene Rogot for supplying the data for the Dom stud• of veterans and Henry Vasquez and Linda Merlino fbr asstsung it the processing of the data in this study. 1061. JNC1.. VOL. 66... NO 6. JUNE 146. ;? 202 351 2608

17M2 Gsrflnk.l partments. Mortality data for this analysis begin with observation starting on July 1, 1960. Data are presented fos three 4,year periods: period 1, July 1. 1960, through June 30„ 1964; period 2. July 1, 1964, through June 30, 1968; and period 3, July 1, 1968, through June 30,, 1972. Person-years of observation in nonsmokers and deaths at single years of attained'ages 35-89 years were computed and' combined by, 5-year attained age groups. In the Dorn study of veterans, questionnaires were mailed starting in January 1954 to 293,000 veterans holding U.S. Government life insurance. About 65% of the questionnaires were received over a period of several months. In January 1957 a second questionnaire was mailed to those not responding to the first mailing and the replies raised the total to 85% (7), About 54.000 of those who replied were nonsmokers. The same classification of nonsmokers was used in this study as was used for the ACS study. Person-years of observa- tion and mortality by single years of attained age were computed starting with January 1. 1955, for the re- sponders to the first mailing and starting with January1, 1958; for the responders to the second mailing. Death certificates were supplied to the Veterans' Ad- ministration in support of insurance claims through 1962. For the period 1962-69, death certificates were obtained through field work at health departmencss by AC.S personnel (8). Death rates by 5-year age groups were adjusted to the distribunon of the stationary population („L.) of white men and white women of ages 35 years and over in the abridged life tables for the U.S. population in 1965 (y). Differences in death rates for periods I and! 2 and periods 1 and 3 were tested for significance at the P<0.05 leveli by the Mantel-Haenszel procedure (10). RESULTS Titne Trends Nn Lung Canc.r Mortaltty Amonyl Nonsmok.rs Table I shows the 5-year attained age death rates for lung cancer among nonsmokers in three periods of time. The table includes men and women in the ACS study and men in the Dorn study, of veterans. There were 195 deaths from lung cancer among male non- smokers and 564 deaths from lung cancer among female nonsmokers in the ACS study during the 12- year period. There were 168 deaths from lung cancer among nonsmokers in the 15-year period in the Dom study of veterans. Some of' the rates computed for 5- TABLE 1-DeotA ratb /rvm !wp carter per I00.000 person-yeass among nonsmokers. ages JS-89 yeart. by time period.• ACS proepecnve study and tAe Dorn, stLdy of nettroni ACS prospective studyb Attained atRe lrroup, yr' Period 1: Period 2: July 1960- July 1964- June 1964 June 1968 Males Period 3: July 1968- June 1972 Dorn's study of veurans' Period 1: Period 2: Period 3: Jan. 1955- Jan. 1960- Jan. 1965- I)ec. 1959 Dec. 1964 Dec. 1969 35-39 40-44 - (8.7) (14.3) (103.5) 45-49 (4.0) (5.1) (8.6) 50-54 (5.3) 8.8 (8.8) W59 10.5 11.6 8.3 (12.0). 8(-64 17.0 17.3 17.5 112 (10.7), (48.0): 6b-89 18.8 29.4 34.3 25.1 16.9 43.5 70-74 32.3 26.4 19.2 39.9 40.5 382 7(5-79 32.7 41.5 58:6 (37.8) (15.0) 47.2 80-84 (47.9), 106.8 51.9 - (200.6) (204) Sb=89 61.8 152.7 (69.9) (595.2) - - No. of deaths 52 74 69 38 52 78 Age{tandardized death rate 12.5 18.5 15.8 189 13.4 19.6 Females W39 40-44 - (3.5) (3.5) 45-49 6.9 (3.3) 50-54 5.2 7.7 (3.0) . 6559 A0-64 7.4 14.0 8.0 - -12.3 6.814.5 86-69 15.6 162 17.7 70-74 19.4 21.1 22.0 76-79 37.3 30.5 36.3 80-84 51.5 45.1 40:8 86-89 53.4 44.5 59.5 No. of deaths 175 184 205 Age-standardized death rate 13.8 12.9 13.1 ' Some 5-yr aQe yroups were combined in the standardization of rates to avoid 0 caea in these Qroupa, ~ htumbers in parentheses indicate <3 deaths in Qroup: JNCl! VOL. 66. NO 6., JUNE ,1M, 2q2351~609

Lung C.nc.r Tr.nds in Nonsmoken iv year age groups were small and subject to considerable sampling variatiom There was no appearance of' any consistent increase in the lung cancer death rate among nonsmokers with time by, 5-year age groups. The age- standardized rates for males shown in table I and inn text-figure l showed no trend. The rates for women were based on many more cases, and the age-standard- ized rate was. vinually., the same in all three periods. The differences in rates between periods 1 and 2 and periods I and! 3 were not statistical'ly significant in both the ACS study and the Dorn study; of veterans. The analysis was based on the underlying cause of death on death certificates. The death rates for the three periods were also standardized to the distribution of the stauortary• population of white men and women combined, of ages 35 years and over, in the abridged life table for the U.S. population in 1965. This standarditation raised the r2tes for males slightly and decreased the rates for females slightly; but it changed the pattern of the trends verylittle. An atterimpt was made in the first 6 years of follow-up in the ACS study to obtain confirmation of diagnosis for all cases with cancer from physicians who signed the death certificates or from hospitals in which death occurred! Information was received confirming the primary site of cancer in 78% of the cases, and microscopic confirmation was obtained in 69% ob the cases in the first 6 years (6): Table 2 shows a comparison of the death certificate diagnosis and the final diagnosis from the medicall report. Among nonsmoking men. 74 were reported to have died of lung cancer according to the death certificates. Six of these (8.1%) were reported to have died of cancer of another site on the final report. However. 9(0:8X) of the deaths reported as being due to cancer of a site other than lung on death certificates proved to be due to lung cancer on the final repon. Thus among nonsmoking men there were 74 deaths from lung cancer reported on death~ certificates and 77 deaths from lung cancer according to the final medical report. ~ 1= ~ 0 TABLE 2-LunQ caneer deaLUU amonQ nonsnwkerZ in fiTef 66 years of study on dta.th orrttJitatu and wr Jinal rrporte Final Death certificate diagnosis repon Lung cancer Other cancer ditQnosis No. Percent No. Pereent. Maler Lung cancer 68 91.9! 9 0.8 Other cancer 6 81 1,153 99.2 Total 74 100;0 1,162 100,0 Females Lung cancer 169 83.3 10 02' Other can cer 34 16.7, 5,160 99 :8 Toul' 203 100.0 B,170 10U;0 so I- ~ DORN _n~ It,G's. lTUD'1~ :' 0 20 °o IIC.i sTUOY, 1POMEN a In women the picture was somewhat different. Two hundred and three cases of lung cancer among non- smokers were reported to be lung cancers on, death certificates, and 34' (16.7%) were reported to be cancers of other sites on the final medical. reporo. A smaller number, 10 (0:8%), of those cancers that were reported as being of a site other than the lung on death certificates were reported to be lung cancers on the final report. Thus on death certificate reports, 203 nonsmoking women were reported to have died of lung cancer in the first 6 years. On the final report, 179 (a decrease of 11.8%): were reported to have died of lung cancer. About one-third of the 34' females whose causes of death were attributed to lung cancer on the death certificates and changed on medical conf'irmation died from breast cancers. However, breast cancer was under- diagnosed on death certificates in nonsmoking women. There were 1.310 breast, cancers reported on death certificates in the first 6 years of the study and 1,37'1 on the final report. Table 3 shows the age-standardized rates for total: mortality for all cancers and for cancers of selected sites among nonsmokers in the three time periods. Overall mortality, in men decrease& 3% from period 1 to 3. This slight difference was statistically significant at the P<0.05 level because of the large number of deaths involved. None of the differences in total cancer or in cancers of other sites in men in table 3 between periods P and 2 and' periods I and 3 were statistically signifi- cant. Women had an 8% decrease in total death rates between periods 1 and 3. The difference in rates was statistically significant. The decreases in totalancer and uterine cancer between periods 1 and 2 and periods I and 3 were statistically significant. None of the differences for cancers of other sites were statistically significant except for the 29% decrease in cancers of the buccal cavity, pharynx, larynx, and esophagus between periods I and 3. YEAR TEXT-rIGVRE t.-Lung nncerr mortality niesin three 4-yr periods for nonsmolners in the ACS prospecuve studT ind for nonsmokers in three S-yr penods in the Dom .tud; of vecerans: P.sslve Smoking N © ~ ~ A number of studies have established that non- ~ smokers exposed to smoke from cigarettes in a poorly ~ vOL. 66. NO 6. JUNE 1981

1064 Ctirtfnk.l TABLE 3.-Trends in moriality rates from conctrs of selteicd'siiu iw tArre time periodalon *orvmok'rra .! CS p+wyeriive study. 1960-7 Ps.nmeter No, of d th Period lc July 1960- Period'2: July 196.4- Period 3: July 1968- ea s June 1964 June 1968 June 1972 Males Total deaths 19,805 1.608.7 1,588.6 I .6b9:9 Tota cancers 3.151 247.8 252.4 251.6 Cancers of buccal avity, pha.ryru, larynx. and esophs." 62 6.86 6.79 6.46 Cancer of colorrrectum 636 51.9 45.0 50.4 Cancer of pancreas 199 15.0 17.6 14'.0 Cancer of prostate gland 573 69.5 63.1 59.6 Females Total deaths 62 .966 1,49C5 1, 485 . 8 1.374.2 Total cancers 13,275 917:9 304.6 298.1 Cancers of buccal tavity, phsrynx, larynx, and' esophagts 159 4.88 4.21, 3.48 Cancer of colorr-rettum 2,429 68:0 69.8 56.7 Cancer of pancreas 688 17.4 16.2 14.8 Cancer of breast 8.186 69.3 _ 68.0 76.0 Cancer of uterus 833 22.1 18.4 16:0 ventilated room will show increased levels of carbon monoxide in their blood. These higher levels of carbon monoxide can result in deterioration of psychomotor performance. Many nonsmokers have acute eye and throat irritation responses in the environment of ciga- rette smokers (11). One paper reported changes in lung function tests in people classified as passive smokers compared to nonsmokers, and these changes were interpreted as demonstrating a greater reduction in the function of small airways (12): 1FJirayama (13) reported lung cancer mortality ratios in Japan ranging up to 2:]' in nonsmoking women with husbands who smoked 20 or more cigarettes a day compared to nonsmoking women with nonsmoking husbands. Trichopolous et al. (14) reported similar findings in a case-control study in Greece. A similar analysis was made of nonsmokers in the ACS study, even though classifying nonsmoking women on the basis of the smoking habits of their husbands is not an accurate measure of their degree of passive smoking. Moreover, exposures in Japan and Greece may be very different than they are in the United States. Lung cancer mortality among persons who were marrie& to cigarette smokers was compared with the mortality among those married to nonsmokers. A total of 176,739 nonsmoking women were identified who were married a) to men who never smoked, b) to men who currently smoked cigarettes regularly but less than 20 cigarettes a day, and c) to men who currently smoked 20 or more cigarettes a day. Most husbands had smoked for 20: or more years before the study began, and presumably their wives were more likely to have been passive smokers than were the women married to nonsmokers. Twenty-eight percent of the husbands of nonsmoking women were nonsmokers compared to 21% of inen in the total study population. Table 4 shows the results of this analysis. Expected numbers of deaths were based on the lung cancer rates for the 12- year period (,1960-72), by 5-year age groups of the women with nonsmoking husbands. No attempt was made in this first analysis to adjust for other possible confounding factors. The observed versus expected lung cancer mortality ratio for women whose husbands smoked less than 20 cigarettes a day was 1.27; for those whose husbands smoked 20 or more cigarettes a day, it was 1.10. Neither of these differences was statisticallyy significant arP<0:05 by the Mantel-Haenszel procedure. A separate matched-groups analysis was made of the lung cancer deaths among the same 3 groups of women to eliminate the possible effects of potential confounding factors. The women in the 3 groups were matched by age (5-yr age groups); race (white, non- white), highest educational status of husband or wife (not a high school graduate, high school graduate, or higher), residence (rural, not rural), and husband occupationally exposed to dust, fumes, or vapors (yes or no). The analysis was restricted' to nonsmoking women who were not sick and who had no serious disease at the start of the study. "Adjusted" numbers of deaths for each matched diad were computed, as described in other publications (;S, 16). In this pro- Test.t 4:--Ob.erurd' tersvs espeeted' haip cnrcer deaths eenorp wor naoiriwp uvsss+t ,eith eiparstie nnokfrp Ausbaadr A CS study, lDe0-tt` Husband Husband Husband Parameter did not smoked <20 amoked ;=2Q smoke eisareass/ cigsrett~ea/ day day Observed deaths 65 39 49 Expected deaths 66.00 : 30:67 44.67 Mortality ratio 1.00 1.27 1.10 ' Ezpscted deaths are based on the lung cancer rates by S-yr at;e Yroupa in women with nonsmoking husbands applied to the peFson-years of women with smoking husbands. The 95% confidence limit for women with husbands smoking <20 ci¢arettea/day' was 0:86. 1.89 ' for women with husbands smoking ;'20 cisarettes/day; it was 0.77. 1.61. JIVCT vOL. 66. NO. 6. Jt'ttiE 1991

Lung C.nc.r Tt'.rtda In Nonsrnok.n 1c T,s1F 5.-Marshed arv+cp .rudy.• Adjwud' tinw fanerr deatAr amonQ wamen w-itJi nonamokinp, Auubawdl .eatched svitA wonu+c witA emodi np A'subandi Group No. of sdlusted lung cancer deaths Ratio P`' Nonsmoking husband 25.6 1.00 Hsuband .moked <20 35.0 1.57 NS cigarettes'day Nonsmoking husband 84.6 1.00 Husband smoked 220 36.8 1.04 NS cigarettea/day ' See text for definition of adjusted' deaths. s Matched on the basis of a) wife's b-yr av group: 6) hnsband'i ocups.tional expoeun, e) highest educational level of huaband or wife, d) raa. e), urban-runl residetues and f)absence of serious d'uease at the stsrt of tlie study: ' N S -not siQttifiesttt. cedure women whose husbands never smoked were compared to women from each of the 2 groups in which the husband smoked cigarettes. The number of lung cancer deaths in each matched diad was adjtasted to the proportion of persons for each group and summed over all groups to give an "adj'usted"' number of lung cancer deaths. Variances were computed; for each of the matched groups and summed'over all matche& groups, an& probabilities were computed under the null hypothesis of observing no differences. The results of this analysis are shown in table 5. Th- ratio of adjusted lung cancer deaths in women whose husbands smoked less than 20 cigarettes a day to those in women whose husbands never smoked was 1.37. The comparable ratio for women whose husbands smoked 20 or more cigarettes a day was 1.04. None of these differences were statistically significant (P>0.05). DISCUSSION Data from the two prospective studies reported in this paper indicate that the age-adjusted mortality rate for lung cancer in nonsmoking men 35-89 years old' was between 12 and 19/100;000 in the 1950's and 1960's. The observed rate for women was about 13/ 100,000. The rate may actually be about 10% less because lung cancer in nonsmoking women may be over-reported on death certificates. The lung cancer rates shown in table 1 may be slightly different from those shown in other publications because different years, age groups, or methods of standardization were employed. The rates for male and female nonsmokers by age group in this analysis were in about the same range as that of the 1958' rates for nonsmokers in Haenszel's report of a 10% sample of death certificates in the United States (17, 18). The 1966-68 estimates derived by Enstzortt from several sources are not directly com- parable because of a different classification of non- smokers ("never smoked cigarettes"), (3): The male rates in the period 1968-72 are about one-half those reported by Enstrom for active Mormons in 1968-75 (19). Enstrom defined active Mormons as a cohort that can be considered "almost entirely as white males who never smoked," and he used this cohort to serve as the nonsmoker lung cancer rates in the 1968-75 period "in lieu of recent national mortality data on nonsmokers." The mortality rates for lung cancer in both male and female nonsmokers by 5-year age groups showed no consistent trends over the period' in this study. Long-term effects of passive smoking are difficult to establish because of the problems in classification. It may be misleading to classify, a women as a passive smoker or not on the basis of her husband's smoking habit. Wives of nonsmokers may be more exposed to cigarette smoke of others than wives of cigarette- smoking men; wives of smokers may, be very little exposed to the cigarette smoke from their husbands or others. In addition, 13% of the women nonsmokers who died of lung cancer in the ACS study reported that they were previously married, and the classification of their exposure to their husbands' smoking may, not be perunent. In autopsy studies of cigarette smokers, there was a dose-related spectrum of histologic findings„ including basal cell hyperplasia, metaplasia, and cells with atypi- cal nuclei in the mucosa of the tracheobronchial tree that may, lead to invasive carcinoma. In contrast, advanced histologic changes in specimens from the tracheobronchial tree, such as lesions with six or more cell rows, lesions having 50% or more cells with atypical~ nuclei, and carcinoma in situ, were found in less than 0.1% of' the slides of nonsmokers (,2Q), Since there is such little variation in the appearance of these histologic changes in nonsmokers of different age, sex, and residence, it seems doubtful that those nonsmokers who had been heavily exposed to cigarette smoke from others in their lives could have had many more precursor lesions for the development of lung cancer than nonsmokers not so exposed. Therefore, there is evidence from these studies that passive smoking cannot play more than a very small role in the development of lung cancer. Mortality ratios for male smokers of less than 10 cigarettes a day compared to those of nonsmokers range from 2 to 1 in Japan to nearly 5 to 1 in the United States. Mortali'ty ratios in women are even lower. It appears unlikely on a biologic basis, therefore, that wives with husbands who smoke 20 or more cigarettes a day, can have mortality ratios that approach those of regular cigarette smokers. To obtain data on passive smoking in nonsmoking women, an epidemiologic study should be specifically designed to measure their exposure as accurately as possible. This is very difficult to do. Neither the Japanese study nor the ACS study was designed to obtain definitive information on passive smoking. Data for lung cancer risks in occupationalh• exposed nonsmokers compared to nonexposed nonsmokers are JNCI. VOL 66. NO 6. JL'ti{ 1961

pf~e Garflnk•I not very extensive. One study showed an increased risk in heavily exposed asbestos workers on the basis of a smaM number of cases (21), It would be interesting to continue studies of lung cancer trends in nonsmokers over a long period of time, but the,~,major«publie, heakh,; problem in lung eancer is with cigarette smokers. Cigarette smokers who are occupationally exposed to asbestos have a greatlyy elevated risk compared to the risk among cigarette smokers not so expose& (21). Lung cancer rates are rising at an alarming rate in women who smoke cigarettes. Educational efforts should focus on smoking- cessation programs for these groups and particularly on persuading young people not to start. Even if the estimates from this analysis are in error and there was a slight increase in lung cancer trends in nonsmokers, it did not appear to be an important problem in the overall picture for the time period of this study. REFERENCES (1)', Public Health Service. Smoking and health. A repcxr of the Surgeon General. Washington. D.C.: U.S. Govt Print Off. 1979 [DHEW publication No. (PHS)79-50066J.. (2) HAMMONDEC.. SEIDMAN H. Smoking andancer in the United States. Prev Med 1980: 9:169-173. (3) ENSTROM JE. Rising lung cancer moruliiyamong nonsmokers. J N CI 1979: 62:755-760. (I):HAMMOND EC. Smoking in relation to the death rates otonr million men and women. Natl Cancer Inst Monogr 1966: 19: 127-204. 0)HAMMOND~EC. GArtrrNKEL L Coronarrhearcdiaease, stroke and aortic aneurism. Factors in the etiology, Arch EnviTon Health 1969; 19:167-182: (6) GARFtNREL L. Cancer mortality in nonsmokers: Prospective studv by the American Cancer Society. JNCI 1980; 65:1969- 11"r3. JNQ. VOL. 66. NO. 6.. JLNE 1981 (i )' KAHN HA. The Dorn studv of smoking and morulitv among U.S. veterans: Report on elght and one-half, years of' observa- uon. Nytl Cancer Inst Monogr 1966,, l41-125. (8) RoeoT MA. Mt;RR tr JL. Smoking and causes of d'rah among U.S: veterans: 16 years of observauon. Public Health Rep 1980k 95:213-220. (9), Public Health Service. Life tables. In: Nauonai Crnter for Health Statistics. Vital statistics of the United, Sutes-1965. Vol !1 Mortality, part A. Washington, D.C.: U.S. Govo Print Off, 1967:1-8. (?O) MANTEL N, HALNSZEL W. Statistical aspects~ ofthr analysis of data from retmspecuve studies of disease. J Natl Cancer Inst 1959; 22:719-748. (11) Public Health Service. Involuntary smoking. In: The health consequentn of' smoking., 1975. Atlanta, Ga.: Centen for Dii,ense Control. 1975:83-112: (12) WHrrE JR. FttoEt HF. Small-airnvays dysfurtction in nonsmokers chronically, exposed to tobacco smoke. N Engl JI Med 1980: 302:720-723. (13) HtRArAstA T. Non-smoking wives of heavy smokers have a higher risk of lung cancer A study from Japan. Br Med J' 1981; 282:183-185. (14); TRICHOt'OULOS D. KALANDIDIA, SPAEROSL, MACIMAHON B„ Lung cancer and' passive smoking. Int J Cuscer 1981: 27:1-4. (13)i HAMMOND EC. GARitNREL L, SLtDMANH,.. LEwF1t,.. Tar and nicotine content of cigarette smoke in relation to death rates. Environ Res 1976: 12:263-274. (16) HAMMOND EC, GARFINKEL L Aspirin and coronary heart disease: Findings of a prospective stud'y. Br Med J 1975: 2:269-274. (17) HAENStLL W. LoveL,uvD DB. SIRIUN MG. Lungc.ncer mortality as related to residence and smoking histories. 1. White males.. J Natl CGnca, Inst 1962; 28,947-1001. (18) HALNszLL W;, TAEtnER KE. Lung-cancer mortality as related to residence and smoking histortes., Il. White females. J Natl Cancer Inst 1964: 32:803-838'. /19) E.NSrtoM JE. Cancer, and; total mortality among active Mormons. Cancer 1978; 42:1943-1951.. (20) AULR{ACw D.,GARFINCEL L. HAMMOND EC.Qlangesin,bronchlal epithelium in relation to cigarette smoking., 1955-1960 vs. 1970-1977. N Engl J Med 1979: 300:381-386. (Z1) HAMMOND EC. SELIROii lJ.. SEIDMAN' f?l.. Asbestos expOSure.,clg- arctte smoking and death rates. Ann NY Acad; Sci 1979; 330: 473-490.