Tobacco Documents Online

A n AN ESTIMATE OF NONSMOKERS" LUNG CANCER RISK FROM PASSIVE SMOKING A quantitative assessment is made of nonsmokers' lung cancer risk from breathiing ambient tobacco smoke. This assessment 'is based on an exposure model incorporating average concentrations of tobacco smoke expected to be encountered in the two microenvironments, at home and at work, in which the average adult appears to spend about 900. of the time, weighted by the estimated probabillities of co-occupation by smokers and nonsmokers, ar by typical' respiration rates. It is al'.so based on a response estimated from epidemioli studies of lung cancer in nonsmokers with certain lifestyle characteristics. James L. Repace, MS., and Alfred H. Lowrey, PhD.t KEY WORDS: Risk Assessment; Indoor Air Pollution; Tobacco Smoke; Lung Cancer Approx. 5000 words in text 1500 words in appendices 6 tables ACKNOWLEDGEMENTS: The authors are grateful to RL Phillips for unpublished data from his published studies of mortality in members of the Seventh Day Adventist Church. We also thank B Fischoff, J Horowitz, D Patrick, G Sugiyama, W Ott, and J Wells for useful discussions. tJames L. Repace is a physicist and policy analyst in the Office of Air and Radiation, U'.S. Environmentali Protection Agency, Washington, DC 20460. Aifred~H. Lowrey is a researc- chemist inithe Laboratory for the Structure of Matter, Navall Research Laboratory, Washing. DC 20375, The views presented in this article are those of the authors, and do not necess reflect the policies of the agencies named. • t

, , 0 a ABSTRACT We have performed'a quantitative assessment of nonsmokers' risk-of lung cancer fr- passive smoking. Our estimates should be viewed as preliminary and subject to chan5e as improved research becomes available. We estimate that nonsmokers are exposed to from 0 to 14 mil.lilgrams of tobacco tar per day, and that the typical passive smoker is exposed'to 1.5 milligrams per day. We derive a phenomenological exposure-response relationship yielding 5 lung cancer deaths per year per 100,000 persons exposed, per milligram daily tar exposure. Aggregate exposure to ambient tobacco smoke is estimated to produce about 5000 lung cancer deaths per year (range 3000 to 14000) in U nonsmokers aged > 35 years, with an average loss of life expectancy of 17 + 9 years pe fatality. The modeled loss of life expectancy for the most-exposed passive smokers app to be about 2/3 of that reported for pipe smokers and 1/2 of that for cigar smokers. Mortality from passive smoking is estimatedto be from one to three orders of magnituc higher than that estimated~for carcinogens currently regulated~as hazardous air pollutants under the federal Clean Air Act. 4r_

I INtROOUCT~ION' Exposure of nonsmokers to indoor air poLlutton from tobacco smoke (also known, as involuntary or passive smoking) has recently become a public health concernl for several reasons: o Such exposure is widespreadZ+3 a Studies of the effects of tobacco smoke on smokers worldwide have implicated'it as the most_important cause of lung cancerl,4 o Existence of a threshold for carcinogenesis is doubtfu11,5- o There is suggestive new evidence of lung cancerl' (and other serious health effects) in nonsmokers exposed to ambient concentrations of tobacco smoke.2,9.10 In the 1982 report on cancer and smoking,1 the Surgeon General asserted that despite the incompleteness of the evidence, nonsmokers should avoid exposure to second-hand smoke to the extent possible, a judgement supported by the World Wealth Organization and the National Academy of Sciences1'1. Nonsmokers are commonly exposed~to tobacco combustioniproducts in diluted side- stream and~exhaled'mainstream tobacco smoke from cigarettes, cigars, and pipes.2 Tobacco smoke contains 60 known or suspect carcinogens, i~ncluding 51 in the particula: phase; the carcinogenic activity of tobacco smoke appears to require this phase.1 Biloassays indicate that sidestream tobacco tar is more carcinogenic per unit weight than mainstream tar.1 This raises the question of whether the quantity of tobacco tar to which the average nonsmoker is exposed creates a significant risk of lung,caneer. In order to answer this question, we first justify, and thenperform, a quantitative risk assessment (QRA). QR2, deals with the question of how much morbid'ity and mortality an agent is likely to produce given specified lievel!s of exposure; typicalily utilized' in the regulation of carci'nogens, it is important because control efforts cannot proceed without assurance that the health gains are worth the costs.14 QRA involves knowing: the health effects from exposure, the distribution of exposure to the polTutant, the population, at risk, the dose response function, and exposure to and effects of confounding, substances.5,6>7,14 On the basis of such assessnients,

t . -2- infbrmed risk management judgements can be made; in this manner, five carcinogens have been regulated as hazardous air pollutants.14' In this work, we draw upon the epidemiology of lung cancer1,8,15,16 and on indoor air pollUution physics2+11'',1'7 to produce a risk analysis5,6,14,18,19 in whi'ch we correlate nonsmokers' lifestyles, exposure to airborne tobacco tar, and incidence of lung cancer. In our analysis, we first review estimates of the average exposure of the general population to ambient tobacco smoke. Second, we elucidate studies linking tobacco-related disease in nonsmokers to exposure-related variations in lifestyle. Third, we couple these two factors to develop a phenomenological estimate for the aggregate lung cancer risk to the U.S. nonsmoking population, and to develop an exposure-response relationship for the estimation of the risk to the most-exposed. Fourth, to check the reasonableness of our estimate, we compare our estimated level of lung cancer mortality and resul~tant loss of life expectancy from passive smoking to those from cigarette, pipe, and cigar smoking, employ an alternative method of calculating an aggregate risk based on the lung cancer risks of active smoking, and make a crude estimate of the range of risk. Finally, we compare our estimated risk from ambient tobacco smoke to that from various ai'rborne carcinogens currently being regulated as hazardous air pollutants, to determine the significance of the estimated risk. 0 YARIATION'0F EXPOSURE WITH LIFESTYLE In earl!ier work2,9,20L27 we studtied factors affecting non-mokers" exposures to tobacco smoke, and conducted! fi'eld~ surveys of the levels of respi'rable particles indoors and out, in both smoke-free and smoky environments. This work established that ambient tobacco smoke imposed significant air potlution burdens on nonsmokers, and', using controlled experiments, we devel'oped a model' to estimate those exposures. This model predicts that the exposure of U.S. nonsmokers ranges from 0 to 14 milligra:-.. of cigarette tar per day ('mg/day), depending upon the nonsmcker''s lifestyle,2 and that the average population exposure for'adults of working age is about 1.5 mg/day.25 A W dh-m

3. 0 . Table 1, derived from the model,26 estimates probabilrity-weighted exposure to the particulate phase of ambient tobacco smoke for a typical M. adult nonsmoker. We omit exposures received'in other21 Indoor microenvironments, outdoors, and'lin transit,, which account for the remaini'ng 121. of people's time. The table is derived from considerations that ambient concentrations of ambient tobacco tar have been found to be directly proportional to the smoker density and~inversely proportional to the effec tive ventilation rate.2 The ventillatton rate tables given by ASHRAE29 can be used to estimate both the range in effective ventilation rate (from the design mechanical rates) and smoker density (from the design occupancies), and thus upper and lower bounds and average concentrations for model workplace and home microenvi'lronments can be estimated.2,20,26 Table 1 suggests that individuals receiving exposure both, at home and at work constitute a high exposure group, with the workplace appearing four times as strong a source of exposure as the home; the reason for thi'ls differentia is the generally higher occupancy (i.e., smoker density),encountered'in the workplace. This estimate of exposures represents a mod'eled'weighted average taken over the entir=_ populati~on, including those who are not exposed. A limited comparison of these estimates can be made with the results of a study of the prevalence of perceived passive smokiing in metropolitan San Francisco. Friedman et al.3'questiloned nearly 38,000 adult nonsmokers and ex-smokers who receive_ mulittphasic healthicheckups in 1979 and 1980. In general, sex and race were found to __--be correl'ated to passive smoking only to a small degree, and self-reported exposure of at least one hour per week during the working years ranged frcm a high of 78;. during ages 20 to 29, to a low of 600. during ages 50 to 59. Friledman et al.3 conclude that passi!ve smoking is a hiighly prevalent phencmenon, in qual'itative agreement witti = findings.2,26 Quantitati'.vely, Table 1 suggests that employed persons receive exposure_ with an 85: probability, greater than that reported by Friiedman, et ali. for their sub_ Our estimates are for the general U.S. popullation. Friedman et a13 caution tha_ tne "health~conscious" subpopulation may be 'atypical. Also, differences between our es-.i-

exposure probabilities an&those reported by Friedman, et al.'s3 subjects might be due to such poorly understood'factors as differences between people's perceptions and actual exposures,21 given the persistence of tobacco smoke ilm indoor spaces long after smoking has ceased.2,20,27 E.g., Jarvts and Russell* in a study of _ urinary cotinine in a sample of 121 self-reported nonsmokers, state that only 12: of subjects had undetectable cotinine levels, despite nearly 50% reporting no passive smoke exposure. VARIATION OF RISK WITH LIFESTYLE White and Froeb3U evaluated the effect of various degrees of long-term j>20 yrs workplace exposure to tobacco smoke on 2100 healthy middle-aged workers. Of the workers, 83: held~professional, managerial, or technical positions, whi'1e the remaini: 17» were blue collar workers. Passive smokers of both sexes suffered stati'stically significant decl!ines in mid- and end-expiratory flow rates which averaged about 113,5 percent and 22 percent respectively, and did not differ significantly from the values measured in noninhaling or light smokers of cigarettes, pipes, and cigars. They concluded that chronic exposure to tobacco smoke inithe work environmen- is deleterilous to the nonsmoker and significantly reduces small airways function to the same extent as smoking 1 to 10 cilgarettes per day. Kauffmann et al.,31 compared pulmonary functioniin about 3800 people in France: 849 male "true" nonsmokers (defined as those not exposed at home)' 165 male passive smokers (defined as those exposed at home), 826 female "true" nonsmokers, and 1941 female passive smokers./ The authors restricted the analysis to~subjects aged 40 years or more (i.e., to~those who had been exposed for 15 or more years to:smoking by their spouses) and who were 1!iving in households with no persons over the age of 118 years except their spouses. They found that nonsmoking subjlects of either sex whose spouses were current smokers of at least 10!grams of tobacco a day had~mi'd'-expiratory flow rates averaging 11.500 lower than those married to nonsmokers. For women in soc;: classes withithu highest percentage of paid work, the effect of workplace smoking 'J'arvis MJ1, r7ussei'1 MAM, measure:nent ana estlmat7on or smoxe cosage to nonsmoKers f c- environmental tobacco smoke Brit. Med. J!, iln, press. w.1 Ci+7 -r

n o. apQeared to confound'the effect of passi've smoki'ng at home. However, in the large subgroup of women without paid work (i.e., not exposed to workplace smoki'ng)i, a clear dose-response relationship to amount of husbands' smoking was observed. They concluded that women living with heavy smokers appeared to have the same _ reductions in mid-expiratory flow rates as light smokers, and that after 15 years exposure in the home environment, passive smoking Is deleterious to pulmonary function. A third study by Kasuga* of urinary hydroxyproliine levels as a function of passive smoking status showed that urinary hydroxyproline levels in nonsmoking wives and chilidren varied in a dose-response relationship with husbands and parental smoking habits, when adjusted for pre-existing respiratory disease. Elevated urinary hydroxyproline levels have been correlated with degradationlof lung,tissue.* These three epidemiologic studies provid'e evidence that variations in the exposur of adult nonsmokers to ambient tobacco smoke at home and'at work camproduce obser- vable vable pulmonary effects. Like effects have been observed in ~hildren exposed at home.= LIFESTYLES WITH INCREASED LUNG CANCER RISK Nine epidemiologic studies have examined the lung cancer risk incurred by the nonsmoking spouses of cigarette smokers. In each study, the only exposure variable was the strength,of the spouse's smoking habit. The studies were conducted in Greece33, Japan34, the U.S,35,59,70,71, Germany60, Scotland72, and' Hong Kongt,73, In the Greek study, Trichopoulos et. al.33 used the case-controli technique: - involuntary exposure to cigarette smoke as measured by the husbands' daily consumptic7 was foun6to increase the average risk of lung cancer by a factor of 2.4 (p<.01) when lung cancer patients were compared to 225 controls,64 and a dose-response relationsn- was observed. Divorce, remarriage, husband's death, or change in smoking habits t Chan WC, Fung SC, Lung Cancer in Non-Smoker In Hong Kong, Unpublished. *Kasuga H, Hydroxyproline and Passive Smoking, presented at 'New Etioliogies in Lung Cancer, Honr_ Hawaii, March 21-23, 1983.)

0 6. wer,e considered. Although the sample was small'., Trichopoulos et. al.33 suggest that the conservative social setting,rendered the study less susceptible to bias due to smoke exposures outside the home. In the Japanese study (1966-1981) of lung cancer in 91,540 nonsmoking women, Hirayama34 used the prospective technique: relative to those women not exposed at hoMe (controls), involuntary exposure of wives of smokers was found to increase the average risk of lung cancer by a factor of 1.8 (p<.0U1), where the exposure was also estimatec from husbands' daily consumption. The annual LCD rate in the controls was 8.7 per 100,000. Hirayama34 found that the exposed wives experienced an average annual increase in lung cancer mortality rate of 6.8 per 100,000, with a range of from 5.3 to 9.4 per 100,000, in a dose-response relationship depending upon the degree of the husband's smoking. Hirayama34 found further that the risk of lung cancer death in nonsmoking women increased both with the time of exposure and number of cigarettes smoked daily by the husband. Hirayama34 also reported a factor of 2.9 (+ .3, at the 95~~ conf. Level!), for increased risk of lung cancer in 1010 nonsmoking husbands with smoking wives. More recently, Hi'irayama extend'ed his earlier work to suggest74 increased risk of nasal sinus cancer, emphysema, chronic bronchitis, and ischemic heart disease in passive smokers, and evidence of decreased risk ilninonsmoking wives _ exsmokers. In one U.S. study, Garfinkeli35 reported results from an analysis of data collected from the American Cancer Soci'ety"s (ACS) prospective study of lung cancer risk in 176,- nonsmoking womeni(1960 to 1972), as a function of involuntary exposure as indicated by their husbands' cigarette consumption. 72: of the nonsmoking women were married' to smokers. Three smoking categories were identified: none, less than a pack per day, or greater than a pack per day. The U'.S, study reported~nonsienificant risk ratios of 1.001, 1.27, and 1.10 respectivel!y for the three categories (averace risk ratio Is 1.1~9 for wives whose husbands smoke). .. m . ~ ~ . U± L ~

b A great deali of correspondence, pro and con, on the relative merits of these three studies was generated, and is summarized in the Surgeon General's Report.l Hirayama34 suggests that the disparity between the findings of hiis study and that of Garfinkel'35 may be due to the smaliler room size in Japanese houses and the closer proximity between Japanese spouses compared with American spouses. In attempting to explain the different results between the ACS and the Greek and Japanese studi!es, Garfinkel35 and Hammond and Selikoff37 have suggested that husbands' smoking habits are not good surrogates for the total tobacco smoke exposure of nonsmoking wives. Friedman et ali.3 have suggested that although traditional Greek and Japanese wives' passive smoking may have depended almost entirely on theiir husbands' smoking habits, contemporary (1981)A.S. spouses' smoking habits appear to be an inaccurate index of passive smoking. In 1965, 3810 of U.S. women were in the civilian labor force, up only 3 percentage points from 1955.38 8ased'on Table 1, we estimate that a person exposed on the job but not at home, would receiive an averae_ exposure 4 times as high as one exposed only'at home. Presumably then, 38: of the ACS"control"group ha&unaccounted-for workplace exposures whichimay have been four times higher than 62: of his "exposed" group. This may explain,the different results of the Hirayama34 and Garfinkel35 studies.75 In the second U.S. study, Correa, et al!.59, studied &male and 22 female non- snoki'ng lung cancer cases and 180 male and 131female controls as part of a larger study including smokers, with 13381ung cancer cases and 1393 controls, in Louisiana, andreported that nonsmokers married to heavy smokers had an increased risk of lung c_ cer, as did smokers whose mothers smoked. Men with smoking wives had a nonsigniifiicanc risR ratio of 2.0 compared to~their counterparts with nonsmoking wives, and women with smoking husbands had an average rtsk ratio of 2.07 (,p<.05) compared~to women wi•- nonsmoking husbands. A dose-response rellationshi!p was observed, with the peak rilsk reaching 3.52 (p<.05)1. The combined data for men and women passive smokers was signtficant (p<.05) for the heavier smoking category (141 pack-years).

~ Preliminary -u- /' dataffrom a third U.S. study, by Kabat and Wynder70, a case-control study of passive smoking in nonsmokers in 25 male cases and controls, and 53 female cases and controls, where the majori'ty, of the patients were from New York City. The controls consi~sted of patients hospitalized for non smoking-related diseases, roughly two-thirds being cancer patients. No differences on exposure to passive smoking at home or at work were found in the women. However, the male passive smokers displayed a statistically significant (p=0.05) dtfference in lung cancer (odds ratio 1.6)) relative to the non-exposed group. Interestingly, when the data are further broken down and~the male cases and controls are reclassified into exposure categories based upon home and workplace, the odd5 ratios are: 1.00 (neither at work nor at home); 2.4 (at home only); 3.4 (at work only); 9.6 (both at work and at home), although the number of cases is extremely smal!l and the confidence intervals very wide, and the breakdown ratios do not attain statistical significance.* A fourth U.S. study by Miiller71 of mortality from all forns of, cancer in 123 nonsmoking women (only 5 lung cancer cases) as a function of husband's smoking_histor; reported!a non-significant odds ratio of 1.4 for all women (p=.15) for wCmen whose husbands smoked relative to those who did not, and when empl!oyed women were excluded the odds ratio increased to 1.94 and was statistically significant (p<.02). Knoth et al.60'reported on a stud'y of 39 nonsmoking German females with lung can-- 61.5Z were found to have smoking spouses. The authors state that this was threefol'd that expected on the basis of smoking habits of German malles. Chan and Fung* studied lung cancer cases in 397 persons iiniHong Kong, 2 nonsimoke- out of 208 male cases, and 84 nonsmokers out of 189 female cases. Among nonsmoking women, the proportion of cases reporting passive smoking was 40.5: compared to 47'.51. amongithe controls (a risk ratio of 0.35) and it is stated that more non-smoking patients had nonsmoking,spouses; however, the proportion of married women among the cases is not given. Wynder and Goodman61 in reviewing this manuscript, stated that the nature of the survey question'~egarding,exposure was unclear. *Kabat G, private communication. ~ .

-9= Gillis et a1.72 reporte~preliminary results of a study of passive smoking and lung cancer in 91 male controls without domestic passive smoking and in 901subjects expose at home, and in 40 female controls and 58 subjects. No effects of lung cancer were noted'in the females, but elevated rates of myocardial infarction were reported (risk ratio 3.0). In the males, elevated rates of both lung cancer (risk ratio 3.25) and myocardial infarction (risk ratio 1.45) were reported. When smokers were included, (156 smokers plus 156 smokers withipassive smoke exposure) a clear dose-res relationship was shown,. The statistical significance i's not given. RISKS IN SMOKERS WHO 00 NOT INHALE Wynder and Goodman61'and Jarvis and Russell!62 assert that pipe and ci'gar smoking i~nvol~ve heavy passive smoke exposure. Epidemiologic evidence suggests that pipe and cigar smokers tend~not to directly inhale the smoke, and pathologic findings show lung abnormalities in such smokers which are intermediate between those of nonsmokers and cigarette smokers.1,15 Simflarly, the lung,cancer risk. for pipe and cigar smokers is less than for cigarette smokers, but greater than that for nonsmokers, and• dose-response relationships are observed.1,15 Most importantly, lung cancer risks in very, light pipe and ci'gar smokers (less than five cigars or pipesful per day) are nea / the same as those of "nonsmokers;"1,15 yet, cigar and~pape tobacco tars appear to have a carcinogenic potential comparable to that of cigarette tars.1,1'5 This suggests that pipe and'cigar smokers may experience tobacco smoke exposure similar to that experienced by nonsmokers who are subjected' to very heavy passive smoking, a supposition supported by modeling the exposure of a non-inhaling cigar smoker (see Appendix A). Thus, pipe and ciigar smoking are also lifestyl'es with both increased exposure to ambient tobacco smoke and increased risk of lung cancer.

. -10- U M LI'FESTYLES WITH DECREASED LUNG CANCER RISK. It might be expected that subgroups of the population which proscribe smoking among their membership woul&have a lower probability of passiive smoking, and therefore shoulid'also have a lower incidence of smoking-related disease than the general nonsmoking population. One such subgroup is the Church of Jesus Christ of the Latter Day Saints, popularly known as the Mormon Church, which advises against the use of tobacco. Enstrom40 found that active Mormons who were nonsmokers had standardi'lzed mortality rates for lung cancer which were 21%, compared to 19: for a sample of the U.S. general!population "who had never smoked cigarettes." Interestingly, however, this result occurred despite the fact that 31% of the active Mormon cohort were former smokers. This confounding factor was not present for certain subgroups in the following study. Phillips et. al!.41,42 have studied~mortality (1960'-1976) in Seventh Day Adventists (SDAs), a conservative religious group who also follow rigorous proscriptions against the use of tobacco. As with with the Mormons, SDAs have rates of mortality from lung cancer and'other smoking related cancers that are fractions, respectively 21: and 661., of the rates for a demographically comparable group inithe general U'.S. population among whom smoking is epidemic.41 A sizable subgroup (35".) of SDAs report prior cigarette use, especially among men.42' SDAs appear to be less likely than the general, population to be involuntarily exposed to tobacco smoke, as children or as adults, at home or in the workplace, because neither SDA homes nor SDA businessE are likely to be places wnere smoking is permitted, and because the great majority of SDA family and'socfal contacts are among other SDAs who do not smoke (,See Appendix C)~.

Phillips et. a1!.41,42 compared mortalilty in two demographiically similar groups of Southern Californians: SDAs (from 1960 to 1976) and non-SDAs (from 1960 to 1971). In particular, for two select subgroups of each group, 25,264 SDAs and 50,21'6 non-SDAs who were self-reported nonsmokers whoinever smoked, age- adjusted mortality rates were compared for smoking-related and nonsmoking-re- lated diseases.42 Table 2 compares age-adjusted'lung cancer mortality ratios for two SDA cohorts relative to nonsmokers in the general population who never smoked. The first cohort consists of all SOA, and includes those who never smoked, ex-smokers, and smokers. The first row of Table 2 gives the mortality ratios relative to the never-smoked non-SDAs in the general population. The second row compares the second SDA cohort (those who never smoked),to the non-SDA who never smoked. The val~ues given are averaged over both sexes. From Table 2 the results show that the non-SDA group of nonsmokers who never smoke& (but who were more likely to suffer involuntary exposure to tobacco smoke)ihad an average lung cancer mortallity rate of 2.4 times that of the never-smoked-SDAs (the group less likely to have suffered such exposure by virtue of their liifestyle). This ratio is consistent with the mortality ratio of 1.8 reported by Hirayama34, the value of 2.4 foun6by Trichopoulos et al.33, and'the value of 2.0 foun6 by Correa, et a159. Furthermore, the difference in the annual age-adjusted lung cancer mortality rates between non-SDA and SDA men is 6.3 per 100,000 persons, and between non-SDA women and SDA women is 8.6 per 100',000!(Tablie 3). These differences are consi'lstent with the value of 6.8 per 100,000 which Hirayama34 found for the average risk of lung cancer in passive smoking Japanese women.* Phillips, et a1.42, who did not have the benefit of comparison of their study with that of the passive wI !

r-. -12- smoking studies, nevertheless cortcnented that the difference in SDA/non-SDA lung cancer risk strongly suggests that factors other than direct cigarette smoking may be etiologically related to lung cancer, and observed that SDAs are likely to have much less passive smoke exposure than non-SDAs.42' _ DOES AMBIENT TOBACGO SMOKE POSE A CARCINOGENIC HAZARO? The Internati'onal' Agency For Research on Cancer (IARC) criteria for causaliity to be inferred between exposure and human cancer state that confidence in causality increases when o Independent studies agree o Associations are strong o Dose-respcnse relationships exist o Reduction in exposure is followed by red'uction iin cancer incidence.7 We now wish to interpret the evidence we have discussed. We first summarize some arguments against an effect of passive smoking: two epildemiological studies, one large one in the U.S., and a small one in Hbng Kong, find little or no effect. The absence of a threshold for carciinogenesi's has not been proven,. Sidestream smoke has not been demonstrated'to cause cancer in humans. And, as Wynder and' Goodmanbl have observed: lung cancer in smokers is predominantly associated with Kreyberg16 type I, carcinomas, whereas Kreyberg type II predominates in nonsmokers, especially females; moreover, these twoltypes of cancers tend to occur in different parts of the lung; the histologic changes observed in the de- velopment of lung cancer in smokers are rarely seen in nonsmokers;,if measurements of blood levels of nicotine, cotinine, and car5oxyhe:noglobin are truly representativ of uptake of particles and volatiles, it remains doubtful whether ambient tobacco smoke could lead to a pathologic response in otherwise healthy lung,tiissues; pipe * The crude LCD rate in SDA women is also consistent with Hirayema's controls to Y within 80,. ~ Gb O O ~ C11 ~--r CJ

r-• and cigar smokers who claim to be noninhalers may be underreporting inhalation dept On the other hand', mainstream tobacco smoke is a potent human carcilnogen, which is associated~with a wide variety of lung cancer histopathology.16 Evidence of a threshold for cancer is doubtful. Bioassays indicate that sidestream tobacco smoke is an experimental carcinogen. Tobacco smoke is a common indoor contaminant in microenvironments where most persons spend the majority of their time. Of eight epidemiologiic studies of passive smoking,and'lung cancer, five, im, the U.S., Japan, Greece, Germany, and Scotland, suggest that nonsmokers are at elevated risk of lung cancer from exposure to spouses' smoking. Each of the first three studies fin&a doubl'ing of risk, on the average, and displays a dose-response relationshi'p. Moreover, the cross-cultural nature of the studies suggests that the same confoundil factors are not likely to be present.* Lung,cancer risks and histopathology in pipe and cigar smokers (who~appear to have mostly sidestream smoke exposure) are far closer to "nonsmokers" than they are to smokers who inhale routinely (i.e. cigarette smokers). Lung cancer risks in nonsmckers who never smoked are half as high among a religious group which proscriibes smoking than in a comparable subgroup in the general population,. Because society is ni'sk-aversive, public health agencies assess and controlI carcinogenic risks despite incomplete evidence. For example, under section 112 of the U.S. Clean Air Act, enacted to control emissions of carcinogenic and other especially harmful airborne contaminants, the basic criterion is not whether absoi,. confidence in causality between exposure and human disease has beemestablishedi, but simply whether the pollutant "may be reasonably anticipated to result in an i'ncrease in mortality or an increase i'n serious irreversible, or i'.ncapacitating reversible il'lness."13 Pn d'etermining! "reasonablie anticipation" in practice, this criterioniamounts to a determination of the probablility that the pollutant * For a discussion of confounding factors, see Appendix D. 4

-14- Is a human carcinogen, the extent of human exposure, and the use of quantitative risk assessment.12 Even though numbers generated in such risk assessments are often held~to be preliminary and subject to change, nevertheless, such numbers are consi'dered as evidence of the order of magnitude of the effects, and are used in policy-making and risk management.5,6'+12 On the basis of the LARC criteria, we believe the evidence is sufficient for "reasonable anticipation" of an increase im lung cancer mortality from~passive smoking, meeting the test for a hazardous air pollutant risk assessment. We now estimate the significance of the public health risk. ESTIMATION OF TOTAL LCD RISK AND AN EXPOSURE-RESPONSE RELATIONSHIP We now estimate a phenomenological exposure-response relationship based on consi!stency65 of evidence provided by studies of lung cancer i'n nonsmokers and from our exposure assessment. In both the Japanese34 and SOA41,42 studies, which we take respectively as consistent with arguments for increased risk with increased exposure, and decreased~risk with decreased'exposure, the magnitude of the risk increase was about 8'LCDs per year per 100,000 at risk. The population at risk1'5 is nonsmokers over the age of 35. (Iin,1979, there were about 62,424,000 men andiwomeniin this category.38 Applying the risk factor of 8 LCOs/l!00,000 to this popullation, we estimate that the contribution of passive smoking to lung canser in nonsmokers is 5000 LCDs per yeart (Thi';s is about 5: of 1980 LCD rate). We have estimated that nonsmokers in the U.S. population of working age_ '~+ i ~ , 4.t,t~~. J 0..4 : are exposed on the average to about 1.5 mg of tobacco tar per d'ay,~ncluding, the estimated 15: of the population who receive no exposure at home or work. t By comparison, in 1982, an estimated~17000 U.S. nonsmokers died of lung cancer.1

-15- « We shall assume that this is the exposure of physiological relevance, even to~retired persons, whose exposures appear to be less than the employed,3 since there is a long latency for the induction of lung cancer. Using the statistical risk of 8 LCDs per 100,000, we estimate a phenomenological exposure-response relation appropriate for the general U.S. population at risk, of about 5 LCDs per 100,000 person-years at risk per 1 mg/day nominal, exposure. We have previously estimate6the range in nominal exposure as 0 to 14 mg/day.2 Three cross cultural studies of lung cancer and passive smoking showed'an exposure- response relationship. Assuming!a linear exposure-response function4,5,6,63 (this assumption has been shown to be valid under almost any model of carcinogenisis with respect to low-dose k~netics)63, and zero excess risk from tobacco smoke for zero exposure, we calculate a maximum risk of about 70 LCDs per 100,000 person-years for the most-exposed'lifestyle. We have previously modeled this lifestyle as typified by that of a nonsmoking musician who performs regularly in a smoky nightclub and who resides in a small apartment with a chaiinsmoker; many other scenarios may be drawn.2 We now wish toldetermine the reasonableness of this phenomenologic exposure- response rel!ationshilp by comparing the estimated risk for the most-exposed lifestyle with those of pipe and cigar smokers; by comparing its predictions with those from an exposure-response relationship extrapolated from smokers who do inhale, and finalily by estimatTng a crude range from two ULS. studies of passive smoking anc lung cancer.

-16- ISTIMATED LOSS OF LIFE EXPECTANCY. One way of testing the reasonableness of our phenomenological exposure-response relationship is by using it to predict the loss of life expectancy for the most- C exposed 1!ifestyle, and comparing it to the loss of life expectancy in various types of smokers, particularly those who do not inhale. ~ Rei'f50 argues that there exists a genetically-determined di!stribution in, natura1susceptibility to lung cancer in people; the effect of exposure to tobacco smoke is to shift this distribution toward death at earlier ages. In other words, exposure to tobacco smoke produces a loss of life expectancy. One method of presenting risk data involves calcul'ation of the loss of life expectancy, ilniunits of days of life lost per individual, averaged over the entire population at risk. When the average 1!ife-loss is multiplied by the number of individuals at risk, the impact of the hazard on society in person-years of life lost can be assessed. More importantly, we can dasplay the age-specific probabilities of death from the hazard, as well as the average number of years of life lost by the average victim. Appendix C gives the method of calculation. Averaged over all of the population at risk, (i.e., including those who die of other causes), the average loss of life expectancy from passive smoking is calculated to be 16 days, which is equivalent to an ultimata loss of 2.T mill1ion person-years of life for the totad' at-risk U. S. populiation in 1979 over 35 years of age (62.7 million persons). The estimated worst-case loss of life expectancy is 149 days, again averaged over al!l of the popul'ation at risk. The estimated numoer of lung cancer deaths per year age standardized to the 1979 population at risk - Clvw to .... p-e...:.b w-.K rf.-Jrl', Z is about 4700 nonsmokers. The estimated mean life expectancy lost by a passive-s~,c• lung cancer victim is 17 + 9 years. In order to test the reasonableness of our estimates, we compare the estimatec loss of life expectancy from our worst-case estilmate to the loss of li~fe expectanc;• found in pipt and cigar smokers. As we have argue~l earlier, pipe and cigar smokers ~ ~ ~ ~ ~. ~

-17- . canibe viewed as very heavy passive smokers. Thus our model!ed worst-case life- style might be reasonably expected to have exposure comparable to, but probably less than, suchismokers, with commensurate risks. Table 4, adapted'from Cohen and Lee49 gives this comparison. The estimated most-exposed lifestyle has about 2/3 the loss of li'fe expectancy of the average pipe smoker, and about 1/2 the loss of the average cigar smoker. ESTIMATE OF AGGREGATE RISK BASED ON RISKS IN SMOKERS We now derive an alternative estimated exposure-response relationship frcm evidence provided by studies of lung cancer in cigarette smokers (see Appendix 8'). Using the Surgeon General's estimate that 850. of all lung cancers are due to smoki- we estimate a current annual LCD rate to smokers at risk of about 316 per 100,00U. Assuming a one-hit mod'e1 for extrapolation of the risk (which in this range is functionally equivalent to the assumption that that a mi'llIgram of tobacco tar inhaled by a nonsmoker produces a response equivalent to that i'nia smoker) we produce an estimate of about 0.87 LCDs/100,000,person-years, and a corresponding annual aggregate risk estimate of aoout 555 LCDs per year, an order of magnitude lower than our phenomenological, estimate. We now speculate an why these two different methods produce such disparate estimates of risk. One possibility is that nonsmokers may have a reduced tol- erance to the effects of tobacco smoke. Another possibiIi'ty is a"large dose" effect62, whereby incremental amounts of tobacco tar at the large doses experienca- by smokers do not produce proportilonal incremental damage to lung tissue already heaviily damaged by active smoking, causing a single-hit model to underestimate ttie risk when extrapolated48 over two orders of magnitude to low doses. A third possi- bility is generated by modeling the dose, as opposed to the exposure, of nonsmcker_ to tobacco smoke. We have translated the nonsmokers' exposure into dose by means - a simple single-ccmpartment model for•lung deposition and clearance.22 This mocel

. -18'- su95ests that tar may accumulate on the surface of nonsmokers' lungs to an equilibri_ dose an order of magnitude higher than the nominal exposure, to a level of about 16 mg per day, due to the long,pulmonary residence times for respirable aerosols. - If this 16 mg dose, rather than the 1.5 mg1exposure, is the operative factor, then the typical passive smoker would have a risk, accordi,ngito thds model, of about 9' per 100,000, in agreement with the phenomenological estimate. In our earlier work2 we discussed anecdotal evidence that aryl'hyd'rocarbon hydroxylase levels and pigmented alveolar macrophages were increased in two passive smokers, consistent with the existence of such an effect. It has also been found that serum thiocyznate~ benzpynene69 and urinary hydroxyproli'net levels tn some passive smokers have been found to be be comparable to the elevated levels typically found'in smokers. These observations lend support to the notion that the dose in equilibrium may in- deed be larger than the dailly exposure. Moreover, the simple model we have proposed ignores the effect of cancer la~~tency. The long latency period for lung cancer indicates that childhood passive smoking may be an important factor affecting risk iniadulit life: Dolliand Peto4 have suggested that the effect of passive smoking may be surprisingly large because lifelong exposure may produce a lung-cancer effect four times as great as that which is limited to adult life (recall the observation of Correa et a159 childhood passive smoking appeared to elevate the LQO1risk of future smokers). As Bonham and'Ylilson55 have shown frcm a national' study o,` 40,000;children,in 197'0, 62°. came from homes with one or more smokers. If the exposure-response relationship based upon LCDs in cigarette smokers is multiplied by the estimated~exposure for very heavy cigar smokers (Appendix A), we would~expect a mortality rate of (45 mg/day x .5 x 10-5 LCDs/yr/mg/day) about 23/100,000ILCDs/yr for cigar smokers. In fact, Enstrom and Godley53 in a study of mortality rates (11966-1968) in, 1'D'million men and 24 million women nonsmc+ce^ who had never smoked cigarettes --inclludina„ however, oilce and ciaar smckers -- feur.z: tKasuga, op. cit.

-19- anpual LCD rates of 31 per 100,000 person-years iin the men and:13 per 100,000 in the women. Clearl!y the LCD rate per 1100,000 itn cigar and'pilpe smokers must be far larger than 31 because of the presence in the study population of large numbers of nonsmokers who never smoked. This inference is supported by the results of Garfinkel39, who in a study of lung cancer mortality in 94,000 male and 375,000 female nonsmokers aged > 35 years, reported an annual average of about 16 per 1100,000 inimen and 13 per 100,00CC in women, averaged over the period 1960 to 1972, age-standardized to the 1965 populati Moreover, Hilrayama34 and Trichopoulos33 found that lung cancer risks to passive smokers were significant fractions of the risks to active smokers. Both White and Froeb30 and Kauffmann et al.31 found that the degree of pulmonary impairment in passive smokers was comparable to that in light smokers. The lung cancer risks to light pipe and cilgar smokers d'o not appear to be very different from those of nonsmoke (ii.e., passive smokers). These facts liead~us to conclude that an exposure-response relationship based upon lung cancer risk to cigarette smokers who inhale must under- estimate the risk to passive smokers, and that therefore the risk to nonsmokers from passive smoking extrapolated from the risks to smokers must underestimate the true risks. RISK RANGE ESTIMATION The two U.S. studies of passive smoking and lung cancer in women as a function of husband's smoking can aid in estimating a crude range of mortality: the ACS epidemiological study by Garfinke135, which displayed a non-significant risk ratio of about 1.2, and the study by Correa, e*t al.62, which showed a signifi- cant risk ratio of 2.0. Let us assume that these two studies represent respectively a lower and upper bound estimate based on domestic passive smoking. Garfinke135 has reported an annuali LCD risk level of 13.3 per 100,OCO averaged over all exposure groups in the ACS study, from 1960 to 1972. using Garfinkel's risk ratio; of 1.2, this yiel~d's a presumed~ annual, risk of 2.7 LCDs per 1U0,0001 from domestic passive smoki'.ng. Using!our estimate of .45 mg/day for average domestic

-20- F exposure, this rate yi'elds 6 LCOs per 1M,000!per mg/day, 20% higher than our phenomenol~ogical11y estimated relationship. If as we have argued', the true backgrounc_ rate is closer to the 6.2 per 100,000 person-years of the SOA women, then 20w, of 6.2 divided by .45 yields 2.8 LCOs per 100,000 per mg/day, yielding-a crude lower bound of about 3000 LCDs per year. A crude estimate of an upper bound may be obtained by assuming that the riskk ratio of 2.01 found by Correa et al.62 is correct, then the effect of passive smoki'na would be 100". of background, or 6.2 per 100,000 divided by .45 mgJday, or 13.8 per 100,000 per mg/day, or about 14,000 LCDs per year. The study of HSrayama34 reported a risk ratio of 1.8 on a background LCD rate of 8.7 per 100,000, so that this infer- ence appears plausible. Also, the prediction of the exposure-response relationshi'p extrapolated from smokers is inconsistent with the lower bound estinlated by assuming that the Garfinkel35 risk ratio is correct. Finally, a reanalysiI570 of the Garfinke135 study suggests that Garfinkel's risk ratio is incorrect due to confounding by exposure to passive smoking at work by 384.1 of Garfinkel's control group. When these tainted controls are eliminated in computing the risk ratio, Garfi!nkel's risk ratio becomes 1.7, nearly identical' to tha of Hirayama's 1.8.75 This analysis75 which is dependent upon an exposure-response relationship of 5 LC0's per 100,000 person-years, also appears to explain the observE LCD rate and observed~risk-ratio of the ACS Cohort, lending further credence to our estimates.

-21- IS AMBIENT TOBACCO SMOKE A HAZARDOUS AIR POLLUTANT? Although our quantitative estimates should be regarded as pretiminary and'subjed to confirmation by further research, we believe that the evidence we have marshadlel~ is consistent with the judgement that ambient tobacco smoke "may reasonably be antic7 . pated to result in an increase in mortality". We further submit that this evidence consistent with a phenomenologicaY estimate of an effect of passive smoking that is : the order of 5O- of the annual lung cancer caseload in the U.S. To place these estiT in perspective, table 5 gives a comparison of our estimated risk of passive smoking - risks estimated by the U.S. Environmental ProtectioniAgency for the carcinogenic hazardous air pollutants currently'regulated under section 112 of the CleaniAir Act. As table 5 demonstrates, passive smoking appears to pose a publ!ic health risk larger than the hazardous ai'r pollutants from all industrial emissions combined. We conclude that societal measures taken to eliminate passive smoking,in the workpliace, and to, establish separate facilities for nonsmokers and'smokers in public places, are justified on public health grounds by the apparent magnitude of the carc genilc risk, and for the same reason, educational programs aimed at reestablishiing tr, dcmestic smoking parlor are warranted to prevent passive smoking at home.

. APPEIQD?X A: PASI4.E Sh!CtCIbIG BY CIGaR S-MCK-7R5 Fpideniolrgical evidence indicates t'L3t most cigar s.zsekers do r.ot inhale the woke while most cigarette sznkers do.1'5 Witnout inhalation, tobacco s=ke reaches mainly the oral cavity an&upper digestive and respiratory tracts, but ' does not reach the lungs.15 This suggests that the bulk of cigar trckers' exposure to tobacco smoke comes through indirect inhalation fron polluted room air, thus placing the lung exposure of ciyar s-nokers closer to the categorf o~ passive s;^c than it does to cigarette smokers. Based~ upon measurements of indoor air pollution frocn cigar Smoke, we ccnser,7a- tively estimate that smoking a large cigar (prefarred by most cigar s;;nkers){ liberat= about 2-1/2 times as much tar as smoking an average tar cigarette.20 we now esti;.ate the order of magnitude of the exposure that cigar 5{sokers get by incirectly breathin,, roon air polluted with their own~smoke. The approxi.m,ate upper limit of a ciyar smokers' exposure may be calculated by adapting, a model which we have developed for a.mbient cigarPtte aeresol'.2~20 This model must be adjusted for the fact that a cigar smoker is breat:7irg very near to `W`:e source as well' as for the difference in e:r,issions.20 tre conservatively estimate that the gradient in aerosol concentratiomfrcm near to far fran the ciyar is a factor of abcut 2-1/2.20 Eq. Fl gives the concentraticn ics, estimated for passive s;xking by cigar s:nokers (derived fran our model2 multiplied by t.ao factors of 2-1/2): Pts = 4063' Ds/Ca (ug/m3) , (Fl:) where DS is the active s,^soker density, and Ca is the effective air excnange rate.2,20 Assume the cigar hoker chain-s-rokes 32 cigars per day (CPD) (so that one cigar is burning constantlyy in a 40 m3 rcem at 1 air change F>er hour (ach)11,24,26, Then Ds = 2.5 ciyars/i00 m3, Ca = 1 ach, Rcs = 10.1 3, and tte exposure wculd be < 161, r,g/day, ass,=isg a respiraticn rate of 11 m3/hr for 16 hrs. An est:.nated 95% of cigar srrukers scke less than 9 CPD, and the average cigar s7oker is rex---e,-4 to # of=ice on St--kirr and Health, LF2-5, pers~cnal =znunicat:cn.

t -T r sioke 2.2 CBD.# Thus, we estimate that most ncninhaling cigar smokers wou1d be expose& to < 45 mg/day, and the average cigar sToker to about 11 r,r,/day, ccmpara5le t our worst-case estimate for passive s=kIny (about 14 m,/day)2'. By comparison, the average cigarette sasoker who inhales is exposed to about 544 Rr,/day,- and a chains-nke to about 1600 mg/day. tne conclude that the cigar s;ioker who does not inhale receives an exposure which is much closer to the exposur: of a passive-smoking nonszruker than to the exoosure of the cigarette smoker wto inhales. Our measurements20 shcw t"at a pipe appears to produce emissions about the same order oi magnitude as a cigar. Several prospective epideniological studies have demonstrated higher lung cancer mortality for pipe and' cigar smokers than for ncnsmokers.17 Dcse-respcnse relation=_ were observed with increased'use of pipes and cigars as measured by both amount wick_ and reported depth of i'nhaLation.15 Lung cancer mortality ratios ior pipe and cigar s7cking rarged fran,l to 10 times that of nonsmokers, with the average risk running abcut triple that of nons=kers, averaged over 24 different studies. Lung cancer mortality st7.:dies of cigar smmkers wr.n do inhale show risks cornparable to those in inhaLing, cigarette snckers.l' By canparison, risks in the average cigar or pipe snoker (less than five cigars or pipesfsl per dayY are nearly the same as those of "nons;nckers". Similarly, pipe and cigar smokers experience mortality rates frcm chr: obstructive pulr.icnary disease which in most instances are intenediat_ between ciCare smckers and ncnsmokers. In all of these studies, there does not ap_2ar to be any evidence of a t.'Les:old for ex,csure to tebacco saicke and tne risk oz luny, cancer.l"- ~

If I APPEVDIX B: FXTRAPOLATED =Iw.ATE OF RISK F9C.1t PPSSIVc: S.Y.CiC?Nn An alternative method of estimation of risk frcm passive s;mking is calculated as follows. In 1980, 108,504 individuals in the U.S. were reported to have died lung cancer.51 The 1982 Surgeon General's report on Smking and Cancer esti::ated' that 858 of IZLs are due to cigarette smckin3;1 this yields 92,228 LCDs/yr.1 We will assume these lung cancers occur primarily in smokers over the age of 35; in 1980, there were an estimated 29,228,0001snckers of all races and bothisexes in this age bracket." It follows that in 1980, there were 3.156 x 10-3 LCI's per smoker of lung cancer age. In 1973 the average cigarette was 17 mg tar, and ttte average sroker smoked 32 per day,2 for an estimated tar intake of 544 r.~/d'ay-snker. (a 1980 lung cancer death reflects a 20 to 40 year srnoking history, during which snoking rates increased by, and tar levels decreased by, about 50%,15) Thus, 3.156 x 10-3'LCts/wnoker divided by 544 rr,/day-s:roker yields a rate of about .5.8 x 10-b ' LCIDs/yr per ar,/day per smoker of lung cancer age. We now assu.:~e a one-hit6'( model for extraoolation of the estirated' risk in s-cokers down to estimate tne risk in nonsrokers at the modeled average exposure. This model has the for-1 P(D) = 1 - exp (bD), where P(D) is the estimated risk, b is the exposur=-respense function, and D is the ex:)osure. Ero:n above, b- • 5.9 x 10-6 LCDs per year per M/day. We take D=-1.5 mc;/day, fraa our estimate of the average expcsure for the typical U.S. ncnSZOker,26 assu.'ning that per mi~ll~isr_m, tobacco tar produces the sa.Te carcinoGenic response in nens;okers as it does in sckers. This calculation yields an esti.-sated annual =1ri~sk of abcut 0.97 x 10'S from passive saiokirrg, or about an ord'er of magnitude lower than the phzncmenologica_ estimate made earlier. We have ass•.r^ed that nons-ickers > 35 yrs are at risk of lung cancer.52 In 1980, there were about 63,831,000 r,ons-nekars aged > 35.` Thus, we estimate that 0.87 LCDS/yr per 100,000 passive srckers times 63,831,000 passive smckers at risk ~ ~ ~ 'US Public Health Service, Division of Fieal'th Intervi'ew Statistics, Washington, DC, _ ~ ~ ~ dbhm;

B-2 equals 555 LCCs per year in U.S. r,cnsrokers Ertxn passive smokiN, usir.g the or.e-hit modei of carcinogenesis for extrapolaticn. This model, because of its functicr,al, for.n, can be oonsidered as the first sta7e of the more canp,iex multista5e model. This fact, together with the downward curvature of the one-hit modei, means that it will always yield low-level risk estimates at least as larye as those of the multistage model; in addition, whenever the data can be fitted adequately by the one-hit model, estimates of both modelis will be conpara5le." hloreover, in practice, for extrapol'aticn over tso orders of magnitude in exposure, all the different models one-hit, multistage, 1ec,-arobit, weibull, mui!tihit, yield results wit.hirn an order of magnitude. The single hit model usually gives hiyher results unless the data being fitted~ are convex upward. EJcamination oc' the oriyinal F3aTzmnd data discleses that the exposure categories are too broad to determine the shape of the curve. '(oayard S, "Qt:.antitative Est?.n~atien of Tetrachlerebibenzcdioxin (TCCD)": U.S. Envirrrnental Protection Fi;ency, Carcinogen Assessent Greup, Unpublished.

115:~. APPEBTDI?C'C: AGE-ST:~.tiT~RDIiFD GkLZU:~;TiCN CF F'CD A.\',r3L U.S. MCR'iALT:Y ?VD -LLSS OF OF LIFE EYPECraNCY F1CM I;~'~iOLUMI-nRY EX.=0.5URE T0 MBI''_Yi TCEACCO E:`^.QCE v ApproxiMateiy 50% of SCAs in the cancer aye range (:>35 yrs old), are adult converts to the church; others were either born into an SII4 ti=e or 3oined the church prior to age 20, typically with other iiss-ediate fanil'y meswers. A ~ large proportion of SCAs tend to be heavily involved in church activities. Cn1y a very sa11 procortion of SL1As report current use of cigarettes (males, 1.71; fe:nales 0.58).42 (By contrast, in 1970, 43.5% of adult males and 31.1% of adult fe:nales in the general poYul'ation aged > 17 years reported soking).43 D1oreover, a substanti'aL portion of SMAs kork for "an organisaticn owned and operated by the SCA Church" (nearly 45% of SDA fe:nales and 40% of SDA males in the study group, (aged > 25 years), reported workiry for the SG+, Church.)*.41,42 Clearly, SDAS are less likely than the general population to be involuntarily exoosed to tobacoo smke, as children or as adults, at hcme or in the wvrkplace, because neither Sa?n cores nor SL'2, businesses are likely to be places where saokir.g is permitted, and because the great majority of SMA family and social contacts are among other SD?s who do not sAke.42 Table Cl shcws the age-standardized calcslation of esti.sated loss of life expectancy and annual lung cancer mortality frcrn passive smoking. The calculation is based on the lung cancer mortality difference between tf,o Southern California cohorts of seLf-reported nensmokers who never soked. Based on lifestyle differences, they appear to have different average levels of involuntary scke exposure. The :nore-expcsed group are designated noz-Si ls, ~rd =e-less-exxsed group SI`.~s (see text). _ Columns 11, 2, 5, and 6 are unpuolis"ed~ data* frcr. which age-ad3usted mortality rates were calculated in the stndy of mortality in the Seventt-Cay Adventist (SCA) by Phillips et a1.411,42 Colur.ns 1 and 2 and 5 and 6 give the age-specific lung!cancer deat'is and:person-years at risk respectively for the SrA and t.`ie ncn-ScA. * UnpubLis3:ed data f^---n reference 42, Phillies RL: personal c=.:3unication.

C-2 Colu:nns 3, 7, 10, and 11 s4cw the average rnsnbf:rs of individual~s at risk annually during the study, allowing for those who died during the study. Cols. 4 and 8 show the annual average luny-cancer death rate ('LCD)~ per 100,000 persons, and Col. 9 gives the differences between the ncn-Si?s and SCAs in those rates. Col. 12 gives average LCD rates weighted to reflect the fact that there were three times as many wanen as men in the study, and that the female data attained statistical signiticance whereas the male did not - altrcugh the ccmbined data were significant.41,42,* We assur.e a coamn LCD rate for bot:z sexes in the calculation that follows. Col. 13 gives the mean age of the individuals in the 5-year age group, and Col. 14 gives the number of persons alive at that mean age per 100,000 born alive. CoL. 15 gives the total number of persons in the 5-year age group (5 x Col. 14) per 100,000 born alive (whites only) fran the 1974 U.S. Life Ta51es-= Co1. 16 gives the age-specific LCD:rates attributedito passive smking, standardized to (i.e., weighted'by) the age specilfic population distributicn in 1974 for U. S. whites, and Col. 17 gives the rates adjusted such that the r,:ean LC:) rate averaged over all age groups, is held fixed at 8 per 100,000. Col. 18 gives the average life expectancy54 correspondiny to t:ze mean age given in Col. 11, which is taken to represent that of the entire five-year age group. Col. 19, the product of Cols. 17 and 18, gives the esti.-nated age-specific age-standardized person-years of life lost due to lung-cancer tr:m passive snicking, deter-lin.ed-by t.".e product of the values of Co'_. 15 times t`:e 1979 U. S. po;ulation divided by the sL~n of the val~ues of Col. 15. The sL^n of the values of Col. 19 gives an estirrated 4227 gerscn-years of llife lost due to passive sTcking per 100,000 persons alive at age 35 in the U. S. populaticn in 1979. 4227 person-years, when divided by the 94',724' perscns54 at risk at age 40 (LTs were not cCserved at earlier ages in the SDa st::dy; _ hc~,ever, they are cbserred in the SenersL nensmoking U. S. Fopulaticn at age 35)'-= yields 16.5 days, the mean ni..^rber of days of life lost, and multipljing by t`~4

C-3 peak-to-mean exposure ratio, 1i12`days for the maxi-~~n number of days lost (where the risks of the non-white population are taken to be the same as for the white population.) The s1..rn of Co1. 20 gives an estimated age-standardized rortalilty total' cf 6,999 LCDs per year, whichintay be canpared with the 7523 LCDs per year derived~ by assu<ning the risk of 8LCDs per 100,000 was distributed unifor,nly over all age groups. Usin3 the nurnber of males and females in the 1979 pe_uLation38 and the percentages of those who are non-smokers43 to for.n a weighted average, we estimate that about twv-thirds of these LCCs are in nons~okers. F~carnining Col. 20, shows that of those individuals assumed to contract lung cancer frcm passive s„nking, that approximately 1-1/21 do so at each year of age fran 40 to 69, and that over age 70, approximately 3% d'o so each year. Of those who actually contract fatal lung cancer tron passive sa-icking, the mean life expectancy iost is about 17 + 9 years, and about 8% lose as much as 33 years.

AP°ENOIx d I AGE-STAIIOAR0I7E0. ESTIMATION OF LUHG CAIICER OEATIIS FR0I1 PASSIVE SNOKING Females ~ I I SOA Never Smokers Non-SOA Never Smokers 1. 2. 3. 4. 5. 6. 7. 0. Total Average LCOs Tbtal Average LCOs ' LCOs Person Annual per 100,009 LCDs Annual No. per 100,000 (17 yr yrs at No. of Per- Person- (12.50 yr. Person yrs. of Persons Person 5-yr. A e Group period Risk sons at 111sk Years perlod) At R1sk at Risk Yrs. 35-39 0 3791 223.0 0 0 5766 450.3 0 40-44 0 11494 616.1 0 1 16466 1300.9 6.0731 45-49 0 10157.5 1103.4 0 2 30319 3046.0 5.2193 50-54 1.119 24000.5 1459.3 4.5106 4 6 1630 4099.0 • 6.4909 55-59 1.000 24102 1453.1 4.0403 8 71209 5666.9 11.222 60-64 1.101 24(151.5 1414.0 4.5777 7 65054 5171.2 10,760 65-69 1,140 23326.5 1372.1 4.9214 4 55614 4420,0 7.1!1lli 70-74 0 21009 1202.9 0 9 44240 3517.3 20.340 15-79 1.000 101122 1107.2 5.3219 10 29250 2325.1 34.100 l10-04 1.775 13435.5 790.3 51.069 6 15301 1216.3 39.213 95t 2.250 1(1011,5 509.3 22.541 10 7091 627.3 126.73 Total 15.40T 195,015 -T.1,412' 10r:1UU9 GT- 410,1120 32,657 267-A 201 Males i-yr, Age Group SOA Never-Smokers Non-SDA Never Smokers 15-39 0' • 1926.5 113.0 0 0 1581 119.3 0 10-44 0 -5732.5 337.2 1) 0 3479 276.6 0 ,5-49 0 9111 539,0 0 0 9662 160.0 0 .0-54 0 11400 675.3 Il 1 19313 1535.2 5.1779 .5-59 1.119 10359.5 609.4 10.0017 2 23040 1095.6 0.3065 0-64 1.000 0163.5 515.5 11.440 4 19535 1552.9 20.4161 5-69 3.401 7306.5 434.5 46.0435 0 14105 1121.2 56,1115 0-74 1.115 6360.5 319.1 11.5301 0 9106 777.9 0 5-79 0 52111,5 310,5 0 2 , 6541 520,0 30.5764 0-n4 1.143 3951.0 232..0 20.0055 4 3511 279.6 113.733 5f 2.235 3160.0 105.9 70.7270 2 1671 '132,0 119.609 Total 111 --.0 - 13 73 JO1.J -;- ~~ . - -- 2 ~ -- ~ r J 4.7 r ~JlT- 1115.3996 T12,9Jlj JT,r . ~~'~~ObOe , I

V I 1 Females Itale/Femalc • . .. • . . ........... .... . ~..v..e .~ 12. 13. 14. 15. Mcan Ilo. of Persons At Risk In Entire 5 yr Age Group 16, . Age Speeiflc Age Stand. (1974 U.S. Ilhlte Population) LCOs 476,005 0 410,610 21.04 461,645 20.11 441,950 17-.59 427,305 19.96 396,900 21.27 355,085 14.41 302,275 39.30 233,445 60.11 156,045 3'.61 591565 55.16 3,701,790 i 203.14 9. 10. I1. Annual Mean Ilo. Annual Average Average Nel9hted Mean of Persons A I_f•OS A9e-Speclflc Aee-Sl+eciflc Ilean A I_CI) age of At Risk at yr. Age per 100,000 i1o. of SIIA No. of SOA per 100 000_ 5 yr. cach yr of oup (Ilon-SllA- , b Non-SOA A Ilon-SOA (unlsex~ Group 5 yr. Group SOA) at Risk at 01sk ;-39 0 681 , 913 0 37.5 95,201 i-44 6.0731 1905 2599 4.64 42.5 94.122 -49 5.2193 4149 5451 4.50 47.5 92,339 I-54 1.9003 6350 0569 2.01 52.5 09,590 -59 7.1131 7120 9625 4.61 57.5 • 05,417 1-64 6.1023 6506 11654 6.07 62.5 79,396 -69 2.2110 5191. 7349 4.05 67.5 71,111 1-74 20.340 4000 5952 13.0 72.5 60,455 -79 20.015 , 3432 4263 29.2 77.5 46,609 1-04 -.10.656 2006 2510 2.35 02.5 31,209 ~ 1' 104.109 1211 1536 92.6 01.5 11,913 , Total 163.6473 44,120 51,435 164.69 .'(1914 Census (M/II I tes per 100,000 at Olrth) Males yr. Age oup J -3n 0 232 : • • -44 0 614 -49 0 13110 -54 5.1179 2211 -59 -2.4152 2505 4 -64 9.0651 2060 -69 10.6140 1556 -14 -17.5301 1162 .79 • 30.5164 031 • •.04 04.0415 512 . I 40.9612 319 ` Trr/ n1 1F41 1",F11 li "I11/1 • ZLC9Ot?03 t

I ! 19. Person 20. l0 , Yrs• of 1.lfe LCOs per 11. Average Life Lost Due to LCOs ' Year 1n Age Group 1n 5 Yr. Age f, ro u~ AdjusteJ LCOs Expectancy for the 5- yr Age Grour from Passlve 5moklnq Entlre 1979 U. S.+ Population I ^ 35-39 0 0 0 40-44 23.47 33.1 777 540 45-49 22.32 20.7 641 514 50-54 13.53 24.6 333 311 55-.59 21.45 20.0 446 493 60-64 29.31 11.2 504 674 65-69 15.49 14.0 211 356 70-74 42.24 11.1 469 971 75-19 11.27 0.6 630 1605 00-f14 3.94 6.6 26 91 . 051 59.20 3.1 • 1114 . 1364` 1749 . 4,221 r,999 - . c.. . ~ ~. . ~ .... ~ rn . ' . ' v . fN 10 ;, i . . z~~so~oe G -V I

•APPt?JDIX D: DISCUSSICN OF CCDIECUNDI. FAC:CPS r The IARC criteria for causality and hunan cancer specify that pussible sources of bias and confounding error should be ccnsidered.7 What factors other than passive s:ncking could account for a Lung cancer difference between t•.,;o cotorts The most obvious one is misclassification. Saae of the individuals classificy' as nonsmokers could have been smokers or exsnokers, giving rise to a spurious effr-- hbrkplace or residential exposure to lung carcinoGens or dietary differences betweer , the cohorts might also give rise to spurious differences. However, this is not likely to be an effect constant over three positive studies in three different countries, all of which report about a doubling of risk when the exposure variable is spouses' s;noking. Arsenic, asbestos, berylLiun, chloroethers, chrani4:,-a, coke oven e:nissions, nickel, radon, and vinyl chloride, as we1L as tobacco s~oke, have been i.molicater in the etiology of lung cancer.16 Possible differences due to industrial exposures s"culd be expected primariliy in blue-collar wvrkers. Phillips et a141,42 have stated that the S&A/ncn-SMA subgroups were demographically and educationally similar, suggesting similar occucational distributions, al thcugh1 there is no infor^sation on this point. There is no reason to bel!ieve that dcrtestic radon levels, which are a property of the soil, wvuld be any dilfferent in SrA homes than Non-SCA hcmes. It is also possible that dietary differences between the two groups might have contributed to the SCA/nonsCr, lung cancer differer.ce. 54% of SrPs follow a lact-_--~varian diet and 41% rarely use caffeine beverages. However, Hirayama cbse_^jed a dose-respcnse relationship between expcsure to passive s;=king and lung.cancer even in those with an apparently cancer-inhibiting diet. Also SiA/non-SDA cancer differences are not significant for other szcki.r,-rel~ated cancer sites; this runs counter to a protective effect of diet as a confLundin:, factor. Finally, Hiraymma cbser.,ed that the :nacnitude of this effect varied f~ mortality ratio of 1 for passive smeki.r, .+cr,en wno did not follow a protective die= ~ `--r

r-r D-2 to 0.82 for wcren wt9o used green-yellow vegetables cnly occasionally, to 0.72 for wxxnen who ate tr.e'n daily. Thus the magnitude of the effect does not ap, ear to te sufficient to account for the observed S:A/NOnSPA lung cancer difference. Finally, it should be considered'lthat co-exposures to other lung carcino- gens (e.g. radon) may increase the effect of passive srAkir*.g.66

TABLE 1. ESTIMATED PROBABILITIES OF NONSMOKERS EX?OSURE TO TOBACCO SMOKE . AT HOME AND1AT WORK (after Repace and Lowrey26) Non~exclusive probability of being exposed at work: 630, Probabi.lity of not being exposed at work: 370. Non-excl!usive probability of being exposed at home: 62. Probabili'ty of not being exposed at home: 38» Lifestyle: Daily Average Probability of being exposed (Rounded Values) Modeled Daily ~ Average Exposure Daily Proba- biliity-Weighted Exposure At work and at home: 63% x 62% = 39% 2.27 mg .89'mg Neither at work nor at home: 371. x 38: = 14: 0.00 mg .00 mg At home but not at work: 62". x 370. 3 230. 0.45 mg .101mg At work but not at home: 63% x 380. = 24: 1.82 mg .44 mg Total: 1AOS 1.43 mg/day Table 1. The estimated exposure to the particulate phase of ambient tobacco smoke for U.S. adults of working age, at work and at home (these two microenvironments account for an estimated 88'. of the average person,"s -- both smokers and nonsmokers-- time),26 determined from average concentrations of tobacco smoke calculated for model workplace and'home microenvironments, weighted for average occupancy.2,20,21 Our confid'ence in the modeled exposure is based upon concentration estimates for the workplace which yield an average value which differs by only 241, from the value calculatediby averaging ouer field measurements in 23 commercial.buildings in the metropollitan Washington, D.C. area2; our concentration estimates for the d'omestic environment are consistent with values obtained iim the Harvard Six-City Study25. These concentrations, when multiplied by average respi'rati'on rates taken frcm Respiration and Circullation,, Altman PL and Ditmer DS, Federation of Amerilcan Societies for Experimental! Biology, Bethesda MO 1971, yiield typical microenvircn- mental exposures, which are then weighted by probabil'ities as shown in table 1.26 The non-exclusive probabilities are estimated from statisticall surveys of the smoking poli'ciles of a sample of 1000 U.S. corporations and from the percentage of 40000 chiildren in the National Health Interview survey who were raised in homes with one or more smokers.'~"

TABLE 2: Ar...c-AaJ[.5TED SCA-14-NCtSCA RATIO CF LG' G CPNCEIR MCRTAl.ITY (after Phillios, et al.*,42) By Health i?abit Index Best Average 'Worst Averace Third Third Third I. A11 SCAs42 0.54 0.54 0.40 0.96 II. SrAs who Never Smoked* 0.41 0.41 0.32 0.78 Values shcwn are adjusted by Mantel-Haenzel procedure (p 10.01). Lung cancer mortality ratios taken frem a prespective study of two d2srncrae;-- icalLy siTilar cohorts.41,42 The non-SlA care fran the general scuth Cali_°ornia pcpulation, and were seif-reported ncns;aokers who never s:rcked. The SDA ca-e frW a southern California subgroup less likely to engage in passive smoking by virtue of lifestyle differences. The health habit index is a measure of how faithfully individuals adhered to the Church's teachings; the wvrst third were also more likely to have a non-SDA spcuse)*,42 (Values quoted in text are the reciprocals of nuabers given here.) *Urrubliahed data frcn ref. 42; Phillips, RL, perscnal =munication.

TABLE 3. CALC'uC.ATIOV OF ?,GE-Si?NC'nMI1ED MOATALITY Rr1TES FaCs+t D?LL? SU2PLIED BY R. L. PHILLIPS (141,42,*) FOR CkLIFOR.'V'LA S:?, & NGNSIA wH0 NEVER SMCFCED ATEWRY h'UDIBER LCIS (41) : PEE:SON'-YEARS AT RISK'J PGE ADJUSTED MOHI'eaLITY RAT IO ( 41) : CRUDE MOFa?.LITY P"..,E-ADJUbiE'D'[ MOfL"AI.ITY RA: E Male SDA 10.013 73581.5 13.6 x 10-5 13.6 x 10-5 0.67 - Male (+)onSDA 23.000 112958.0 201.4 x 10-5 20.4' x 10-5 Fe:naIe SDA 15.401 195015.5 7.9 x 10-5 6.2 x 10-5 0.42tt ' Female NtnSDA 61.000 4'10828.U 14.8 x 10-5 14.8 x 1075 ~ Age-Ad3ustzent Procedure: The age- ad)usted mcrtality ratio is applied to~the NonSCA mortality rate to generate the SDA mortality rate. ttSigniBicant at p<0.01; ccrbined male and fe:ral!e ratio is 0.54' and is also signif- icant at this level. A calculation of differences in the annual aCe-ad3usted c:ancer mortality rates for two groups of demographicaliy caepara'b1e self-reportad r,ons,;:okers wto never smoked, the SDA, wMio: a_pear to have the lesser exposure to, passive smkinc~,4'2' and' tt'.e NonSi.A. Data on LCDs and person-years at risk cs.ie fran a prospective study of lung cancer mor- tality in southern California.4'1,42,x The non-S,:.A - SCA differences are respectively (20.4 - 13.6) = 6.8'per 100,000 for the men, and ('14.8 - 6.2) = 8.6 per 100,000 for the wcmen. These differences are consistent with the average value of 6.8 per 100,000 which~ was found in~ t"e Japar:ese study34 of lung cancer in the nonsxking spcuses of sxicers. Because of lifestyle differences, the SCA are less likely to engage in passive wcki:.yg than the ncn-SuA.42 The consistency of both the lung cancer mortality ratios and rates between the Japanese and SDA cohorts with presu•ned'lesser exxsure and the consistency in lung cancer mortality ratios between Greek33 and Su=+ corcrts wi th presL.;,,e.~4 lesser exposure strengt::ens the assunpticn that this is evidence of tre cnagnitud'e of tl^e eFfec_ of passive s;cking, since the studies are cross-cultural, with the sa:re set of xte.^.tia'_'_. confeunding factors unlikely to be coerative7',16.

Te'1BI1. 4. ES'DLu.ATED LCSS OF LIFE E~~ECL'A:tiCY etLk1 ALTIUE SMCKItiG (ALL GAUSrS) ' 'D PASSIVE SMCKI*1G (:LUhG CA;tiCER ONLY) - adaoted frm Cvnen and'~ Lee~- w Cause pavs Cigarette smnking - male 2250 Cigarette smoking - fe:nale 800 Cigar smoking 330 Pipe saoking 220 Passive Snokingt' ('Est. most exposed llifestyle) 149 Passive srcki:x}t (Est. averac;e lilfestyle) 16 tEstimated this work; averaged over all nor,smokers at risk, i.e.,those who are presL.. to die fran passive smoking-induced lung cancer, and those who do not. Estimates giv- for passive s:noking are pCencmenological estimates. The consistency of the ca1'culatad loss of life expectancy Letaeen a nonsrokinc lifestyle involvirr,passive smoking~at the modeled extre:ne and the loss of life expectancy in smokers unlikely to inhale lends credence to the numtbers prcduced by our quantitative assessment of the risk of passive sr.oking. The calcuiation:for passive-ssokirg-induced lung cancer is predicated upon mortality rate differences between the non-S:.A and Si'.A in the California pcpulaticn, which are consistent with those found ine passive s¢-mkers in Japan. We assL^e that the results of this calcu'ia= approximate the effect of passive s„9okin3~ in the c;er.eral' oc_ulation of nens;ckers in th.e U.S.

. Table 5. CCMPaR2SCN OF FSTI:"ATED RISKS FRCM VPRIOUS KP:ZARLCL'S AIR FOLLLTPPN:S Risks have been assessed for ncn-occupaticnal exposures of the general pcgu?a- tion to several hazardous air pollutants. All are airborne carcincgens; all 'aut passive s=king are being regulated by society. The statistical mortality given is before controL. POLLLTP_*lr FSTI,NATED: ANhV?L MCRTAr I'ITY Reference Passive Saoking 5000 LCL's per year this work Coke Oven Emissions <150 LCDs per year (44) Vinyl Chloride <27 CDs per year (56) Rad ionucl ides (world-wide impact from CCepartrent of 17 CDs per year (57) Energy facilities) Benzene <8 CDs per year (58'). Arsenic <5 LGTs per year (45) CD = Cancer Death; LCD = Lung Cancer Death Risks for passive srnking and radicnuclides are test estiirates, and risks for other pollutants are upper bcund.

REFERE*:CES 1. The Health Consez ences of Smoking: Srroking and Cancer, A Report of the Surgeon Ger.eral, U. S. Dept. of Health & Hu.-ran Services, 4aashingtcn, D.C., DFII•iS Pub. No. 82-50179, Govt Printing Office, 1982. ` 2. Repace JL, Lc.rrey AH: Indoor air pollution, tcbacco s;roke, and public health. Science 1980; 208: 464-472. 3. Friedman GD, Petitti DB, Bawo1 RD: Prevalence and correlates of passive smoking. Am J Public Health 1983; 73: 401-405. 4. Do11 R, Peto R: The Causes of Cancer. New York: Oxford University Press, 11981. 5. Scientific Bases for Identification of Potential Carcinogens and Fs tiration of Risks, Report of the Ingeragency Regulatory Liaison Grcup, F1ork Grouo on Risk Assessmnt. J. Natl. Cancer Inst. 1979; 63: 24'1-268. 6. National, Emissicn Standards for ldentifying, Assessing, and Reyulating Ai'rborne Substances Pcsir.g, A Risk of Cancer, Envirorr,enta1 Prote.tion Agency. Federal Register 44: 58642-58661, 1979. 7. Internaticnai Agency For Research on Cancer (I=RC), Monographs V 1-20, Supn1. ,"Chemicals and Industrial, Processes Associated With Cancer in Hunans, LARC, Lyon (1979). 8. Pitot HC: Fundamentalis of Oncology, 2nd Ed. New York, Marcel Dekker, 1981. i9. Repace JL: The problem of passive sxking. Bull N. Y. Acad. of Med 1981; 57: 936-946. 10. c:ld Health Organization: Health As7:ec`s Relate'c to Indoor Air Guality. Euro Reports and Studies 21, Regional Office for Eurcoe, Ccoenhagen, Denna z'.t , 19 79 . 111. National Research, Council: Indoor Pollutants. National AcadeTry Press, Washington, D. C., 1981. 12. Ruckelshaus WD: Science, risk, and public policy. Science 1983 221: 102E-1029. 13. U.S. Clean Air Act, As Amended Au_c.:st 1977, 95th Congress, 1st Sessicn, No. 95-11, Senate Carm.ittee on Envircr^ent and Pe:b1i'c 'ekor'cs, U.S. Q.-ve_:me::t Printing Office, Washington, D6C. Noce:rcer 1977.

14. Lave L: puantitative Risk.Assessment in Re711aticn. Washingten, Brcekings ~ Institution, 1981. See also•Albcert RE: Discussion of APCA critical review paper on Toxic Substances In The AtrosYheric Envircr^ent. J Air Polluticn Contzol Assoc. 1983; 33': 836-837. _ 15. SZroking and Health: A Report of the Sur7een General. U.S. Dept. of Health,. Education, and'F/elfare, CHEW Pub. NojPHSf 79-5006F, 1979. ~ 16. Ives JC, Buffler PA, Greenberg SD: Envirronmental Associations and Histcpathc- logic Patterns of Carcinccra of the Lung: The G°hallege and Dilensna in Epidimial'cgic Studies. Fm. Rev. Respiratery Disease 1983; 128: 195-209. 17. Meyer B: Indoor Air Cuality. Reading: Addison Wesley, 1983. 18. Caranittee on Science and Technology, U.S. Hcuse of Representatives, 98th Co.ngrs A Review of Risk Assessment Methodologies. U.S. Govt. Printing Office, 1983. 19. Fischoff B, Lichtenstein S, Slovic F, Derby S, Keeney R: Acceptable Risk. Cambridge: Cambridge University Press,l!981. 20. Repace JL, Lowrey AH:: Tctacco s:roke, ventilatien, and indoor air quallity. ASF?R5E Transactions 1982; 88: 894-914. 21. Repace JL, Ott WR, Wallace LA: Total hLcnan exposure to air pollution. PrccQed- ings of The '73rd Annual Meeting of the Air Pollution Control Assoc., June 22-27, Mcntreal, 1980. 22. Repace JL: Tobacco srnke an&the ncr.s;reker. Preceedings of the 5th World Conference on Smoking and Health, Winnipeg, July 10-15, 1983. To be publ,ished. 23. Beck FG, Repace JL, LoGrrey AH: Ncns"ckers and cigarette s,cke: a icdified Fercepticn of risk. Science 1982; 215: 197. 24. Repace JL: Indcor air polluticn. Envircrmrt Internaticnal 1992; 8:21-36. 25. Repace JL, Seba CB, Lcwrey AH, Gregory T'r1: Effect cf ne;ative icn generators cr. atnb'lent tobacco stnrke. J. Clinical Ecology, in press. 26. Repace JL, Lo•4rey AH: Mcdeling exposure of ncns:-,ckers to ar,bient toM.acco _roke. Proceedings of the 76th Annual Meeting of the Air Polluticn Control Asscciaticn, Atlanta, June•19-24, 1,983. 27. Repa~-- JL: Risks of passive ---,ckirr. Ir.: Risk, Consent, ar:c Air. College Park: Center For Ptiilesc;~hy and a;blic^Poiicc.. -F ,- ---- "E`~

W 28. State Legislaticn on Smoking and Health~1980. U.S. Cept. of Health and Hisnan Services, PubLic Health Service, Atlanta, Ga. 1981. 29. ASHRAE Standard 62-1981, Ventilation for acceptable indoor air quality. Atlanta: American Soc. of Heating, Refrig. and'Ventil!. Engineers, 1981. 30. khite JR, Frceb HF: snall airaays dysfunction in ncnsrokers chronically expesed to toEaccols•noke. New Engl J Med 1980; 302: 720-723. 31. Kauff±<sann F, Tessier JF, Orioll P: Adult passive sraking in, the hcme environment: a risk factor for chronic airf1cw limitation. American J. Epidemiol. 1983; 117: 269-280. 32. Tager I, Weiss ST, Munoz A, Rosner B, Speizer FE: Longitudinal!study of the effects of maternal smoking on pu]rtinary function in children. N Engl J Med 1983; 309: 699-703. 33. Trichcpculcs D, Kalardidi A, Sparros L, MaCmahcn B: Lung cancer and passive snoking. International J Cancer 1981; 27: 1-4. 34. Hiraya-na T: ncnvroki'r,g wives of heavy sssv--kers have a higher risk of lung cancer: a study fran Japan. Brit Med J 1981; 183-185, Hirayama T: Passive smoking and lung cancer. Brit Med J 1981; 282: 1393-1394. 35. Garfinkel L: Tirr+e trends in lung cancer mortality arrong nonsr,okers and a note on passive sr,cking. J. National Cancer Inst. 1981; 66: 1061-1L066. 36. Hirayama T: Passive smcking and lung cancer. Proc. 5th P:orl'd Conf. on Smoking and Health, Winnipeg, July 10-15, 1983. In press. 37. Harm.-cnd EC, Selikoff IJ: Passive srcking and lung cancer with co^rents on two new papers. Envirormental Res. 1981; 24: 444-452. 38. Statistical Abstracts of the United States, 19801, U. S. Dep,art-nent of Carmerce, Bureau,of the Census,Washirgtcn, D. C., 1980. 39. Garfinkel L: Cancer rncrtality in ncnsmokers: prospective study by the americar cancer society. J. National Cancer Inst. 1980: 65: 1169-1173. 40. Erst_m JE: Cancer and total mortality amcng active =^-cr.s. Cancer 1978; 42: 1943-1~951. _. t•:, .~~_:.• . . . . .. . : ~~ ..

41. Phillips RL, Garfinkel L, Kuzra 74, Beescn WL, Lotz T, Arin~B: Mortality ammg california seventh-day adventists for selected cancer sites. J Natl Cancer Inst. 1980; 65: 1097-1107. ~ 42. Fhillips RL, Kuz-+a JW, Beeson WL, Lotz T: Influence of selection versus lifestyle on risk of fatal cancer and cardiovascular disease a.^;-cng seventh day adventists. American J Epideniol. 1980; 112: 296-314. 43. Pdvance Data No.52, Changes in Cigarette Smking and Current Sarckirg Practices Armng Adults: United States, 1978. U'.S. Dept. Health and Huran Services Washir.gtcn, D.C., 1979. 44. Coke Oven Emissions, Background Infor;raticn Lbcunment, U.S. Envircnmental Protection Agency, Washington, DC, 1982. 45. Final Risk Assessrrent on Arsenic, Envircnr,ental Protecticn Agency, Washicgtcn, oC.,1980: 46. Sax NI,: Dangerous Properties of Industrial Materials, New York: Littcn Educaticr. Publishing, 1979. 47. Selikeff IJ: Household risks with inorganic fibers. Bull N York Acad Med 1'9~C1; 57: 947-961. 48. Hoel DG, Kaolan NL, Anderson MW, L-np1'icatien of nonl!inear kinetics cn risk estimaticn in carcincgenesis. Science 1983; 219: 1032-1037. 49', Cchen BL, Lee I: A Catalog of risks. HealthiPhysics 1979; 36: 707-722. 50. Reif A: Effect of cigarette s-rruking cn susceptibility to lung cancer. Cr,colec/ 1981; 38: 76-85. 51. Advance Report of Final Mcrtality Statistics 1980. In Mcnt.'7].iy Vital Statist:cs Report Vol. 32, 1983. 52. Reif A: The Causes of Cancer. Pmerican Scientist 1981; 69: 437-447. 53. Enstrcm JE, Gcdley, Fv: Cancer Mortality Amng a Representative Sa,~Le of Ncnsankers in the United States Curins 1966-68. J.vatl~.Cancer Inst. 1980; 65: 1175-1183. 5... . . _ _ ~ "

~ 54. Vital Statistics of the United States, Life Tables, 1974. U.S. Dept. of Health, and Welfare, National Center for Health Statistics. Washington, DC, 1975. 55. Sonham GS, Wilson RW: Children's Health in Fanilies with Cigarette Srckers. American J Public Health 19811; 71: 290-293. 56. National Emission Standards For Hazardous Air Pollutants: Praoosed Star.dard For Vinyl,Chloride, Federal Register 1975; 40: 59532. 57. RadicnucLides, Background Infor:nation Docu^ent, EPA/520/1-83-001, Envircrrental Protection Agencv, Washington, D:., 1983. 58. Final Risk Assess7ent on Benzene, Environ-,ental Protection Agency, Washington, LC., 1979. 59. Correa P, Pickle LW, Fonhan E, Lin Y, and Haenszel W: Passive Srcking and Lung Cancer. Lancet 1983; II: 595-597. 6(1. Knoth A, Bohn H, Schmidt F: Passive Smoking as a Causal Factor of Bronchial Carcinara in Female Non-s:mkers. Med. Klin. 1983; 78: 66-69. 61. Wynder EL, Gooc':ram MT: Smokirg and Lung Cancer: Sone Unresolved Issues. Epidemiologic Revs 1983; 5: 177-206. 62. Jarvis iaJ, Russell M.P13s Measurement and Estiiration of Smcke LCs..~-7e to 63. Non-s:nckers frem Environmental Tohacco Smoke. Brit. Med. J. in press. Cr,utp KS, Hoel DG, Langley CH, Peto R: Fundamental Carcinogenic Prccesses and Their Ir,nlicaticns for L¢w Dcse Risk Pssessr.ent. Cancer Research 1976; 36: 2973-2979. 64. Trict^.opculcs D, Kalandidi A, Sparros L: Lung Cancer and Passive Smking: Conclusion of Greek Study. Lancet 1983; II: 677-678. 65. Hirayama T: Passive Scscking and Lung Cancer: Consistency of Association. Lancet 1983; II: 1425-1426. 66. Bergran H:, Axelscn 0: Passive Srtcking and Indcer Radon Daughter Ccncentratiens. Lancet 1983; II: 1308-1309. 67. Hell!er +vD: Lung Cancer and Passive Srckirg. Lancet 1983; II: 1309.

.. 68. Cohen JD, Bartsch GE: A comparison between Carboxyhe^:oglobiniand Serum Thiocyan as Indicators of Cigarette Smoking. American J Publiic Health 1980; 70: 284-286. 69. Repetto M, Martinez, M, JiEuropeen de Toxicologie 11974; 7: 234. 70. Kabat GC, Wynder EL: Lung cancer in nonsmokers. Cancer 1984 53: 1214-1221. 71. Mi,ller GH!: Cancer, passive smoking and nonempioyed and employed'wives. West J Me 1984; 140: 632-635. 72. Gililis CR, Hole oJ, Hawthorne VM, Boyle P: The effect of environmental tobacco smoke initwo urban conrnunities iln the west of Scotland. ETS -- Envi'ronmental tobacco smoke; Report from a Workshop on effects and'ileveJs. University of Geneva, March,15-117 1983. 73. Chan and Fung, Lung Cancer in Hong Kong. American J. of Epidemilology 1979. 74. Hirayama T: Passive Smoking and Lung Cancer: Consistency of Associ'ation. The Lancet 1983; ii: 1425-1426. 75. Repace JL: Consistency of Research Data on Passive Smoking and Lung Cancer. The Lancet 1984; ii: 506.

TABLE 1. ESTIMATED PROBABILITIES OF NONSMOKERS EXPOSURE TO TOBACCO SMOKE AT HOME AND AT WORK (after Repace and Lowrey26) v` Non-exclusive probabil!ity of being exposed at work: 631, Probability of not being exposed at work: 37: Non-exclusive probability of being exposed at home: 621. Probability of not being~exposed~at home: 38: Lifestyle: Daily Average Probability of being exposed (Rounded Values) Modeled',Daily Average Exposure Daily Proba- billity-Weighted Exposure - At work and at home: 63% x 62: = 39'. 2.27 mg .$9 mg. Neither at work nor at home: 37% x 38: = 14'. 0.00 mg .00img At home but not at work: 62: x 37% = 23: 0.45 mg .10 mg At work but not at home : 63 : x 38: = 24. 1.82 mg, .44 mg Total: 100% 1.43 mg/day Table 1. The estimated exposure to the particulate phase of ambient tobacco smoke for U.S. adults of working!age, at work and at home (these two microenvironments account for aniestimated 88: of the average person's -- both smokers and nonsmokers-- time),26 determined from average concentrations of tobacco smoke calculatedifor model workplace and home microenvi'ronments, weighted for average occupancy.2,20',21 Our confidence in the modeled expQsure is based upon Concentration estimates for the workplace which yiel'd an average value which differs by only 24: fromithe value calculated by averaging over field measurements in 23 commercial. buildings in, the metropolitan Washiington, D.C. area2; our concentration estimates for the demestic environment are consistent withivalues obtained in the Harvard Six-City Study25. These concentrations, when multiplied by average respi'ration rates taken frem Resoiration and Circulation, Altman PL and Ditmer OS, Federation of American Societies for Experimental Biology, B'ethesda MO 19711, yield typical microenvircn- mentall exposures, which are then weighted by probabillities as shown in table 1.26 The non-exclusive probabilities are estimated~from statistical surveys of the smoking,po'.icies of a sample of 1000 U.S. corporations and from the percentage of 40000 children in the Nati'onal, Health Interview survey who were raised in homes with one or more smokers.,~,`

TABLE 2: PGE-ADJL'STED SDA-T0--NCNSDA RATIO OF LG'[vG C?NMR MC42T?1.ITY (aAer Phill!ies, et al.*r42) Sv Health Habit Index Best Average worst Averace Third Third 'I2sird' I. A11 SDAs42 0.54 0.54 0.40 0.96 IIl. SL'As who Never Srrcked* 0.41 0.41! 0.32 0.78 Values shown are adjust'ed'by Mantel-Haenzel procedure (p < 0.01). Lung cancer mertali~ty ratios taken frccn a prespective study of two denngraph- ically si.^+ilar cchorts.41.42 The ncn-SDA care fran the general south California pcpulaticn, and'were sel'~:-reported ncnsrckers who never smcked. The SDA azre fr.- a scuthern California subgr¢up less likely to engage in passive szCking by virtue of lifestyle differences. The health habit index is a measure of hcw faithfully individuals adhered to the Church's teachings; the worst third :ere also mcre likely to have a ncn-SDA spcuse)*p42 (Values quoted in text are the reciprocals of nusbers given here. )l *L'r.k,ublished data frce ref. 42; VI-Uli:s, RL, pa-acnal crrrn:nicaticn.

TABLE 3. CALCJLATSOV OE e3GE-SLANG~iR^JI1ED~ NfORTALI'1'Y R4TES FRCd~l CaT? SUF?LIED BY R. L. PHILLIPS (41,42,*) FOR Gkr-I1FORyIA S:A & yGtiSii, wY.O NEVER SrCcCED kTEGORY NUMBE.T2 LCIS (41) PERSG.1F-YEAR5 AT RISFfi'3 AGE ADJUST ED MUFuALITY RATIO (41) CRUDE MOfZI'?,L?TY AGE-AL1TUSiEo'r MOK.*A.LIT_ Y RA2°_ Male SDA 10.013 73581.5 13.6 x 10-5 111.6 x 10-5 0.67 Male NonSCA 23.000 112958.0 20.4 x 10'5 201.4 x 10-5 Female SDA 15.401 195015.5 7.9 x 10-5 6.2 x 10-5 0.42tt Female NonSDA 61.000 410828.0 14.8 x 1!0-5 14.3 x 10-5 U Pge-Adjust•r,ent Procedure: The age- adjusted mortality ratio is applied to the NonSMA mortal~ity rate to generate the SPA :lortality rate. ttSignificant at p<0.01; canbined~ male and female ratio is 01.54 and is also signif- icant at this level. A caLcalation of differences in the annual age-adjusted a:ancer mortality rates for two groups of dencfiraphically caaparable se1F-reportac nensmokers whio never =cked, the SDA, who appear to have the lesser exposure to passive s:aokimg42 and the NonSCA. Data on LQs and person-years at risk cane from a prospective study of lung cancer sicr- tality in southern California.41,42,* The ncn-S^.a - SDa differences are resr,ectively (201.4 - 13.6) = 6.8 per 100,000 for the men, and (14.8 - 6.2) = 8.6 per 100!,000 for the women. These differences are consistent with the average value of 6.8'per 100,000 whica was found in the Japanese stud..,34 of lung cancer in the nens-nekir,g si.xuses of s~okers. Because of li~festyle differences, the SDA are less likely to engage in passive s-icking than the ncn-Si.A.42 The consistency of both the lung: cancer mortality ratios ar.d rates between the Japanese and SCA cohorts with oresumed lesser exposure and the consistency in lung cancer mortality ratios betweemGreek33 and S:a conerts with presL,°1ed lesser exposure strengthens the assL.°nption, that this is evid'ence of t`* ;na7ni=de of the eF:ect of passive r;oking, since the studies are crcss-cultural, with the szme set of pcte.^.tial_.. confounding factors unlikely to be operative7,16, . _~ ,,- -::

i V .' . TABLE 4,. ESTL^"_aTED L~S OF LLFE _{.~?C??ItiC: F."~M ACrIVE S~lC4CItiG (ALL C=IJSc~..) „. P,ND PASSIVe SMC(fiI% ( LUNG CMCER CNLY)' - adaoted :ran Coten and Lee,y ~ Cause Davs Cigarette --=king - male 2250, Cigarette s-neking - female 800 Cigar snoking 3301 Pipe smoking 220 Passive Sznokingt! (Est. most expose&lifest;cle) 149 Passive smokingti (Est. average lifestyle) 16 t'Estimated this hvrk; averaged over all nonsznckers at risk, i.e.,those who are presL- to die fran, passive s;,ioking-induced lung, cancer, and~ those who do not. for passive smokiny are phenamer.ologicall estimates. Estimates ci•: The consistency of the calculated loss of life expectancy between a nor.s;oKing lifestyle involving passive s-cking at the crqdeled extreme and the loss of life expectancy in smokers unlikely to inhale lends credence to the numbers produced~ by our quantitative assessment of the risk of passive sr,.ckiny. The calculation for passi~ve-sneking-induced lung cancer is predicated upemmortality rate differences between the nen-SL?, and SIA in the California population, which are consistent wit."l tt.cse found in passive sackers in Ja^an. We ass---ne that the results of this ca1!cu'_a_ approximate the effect of passive s-aokiny in the general population of nons-ckers in the U.S. ~ ~

t 0 Table 5. CCr1PARISCN OF ES-Th.ATED RISiCS FRCM VARICLS FLAZ:--RCCC'S AZR POLLITrA'r:5 Risks have been assessed for non-occuFational ex..~osures of the general pepula tion to several, hazardcus air pollutants. All are airborne carcinogens; all b:t passive smoking are being regu7ated by society. The statistical m7rtality given is before control. PCLLL'I'?^.T ESTI:^ATED ANh'U?- MCR'TP:L.ITY Reference Passive srroking 5000 LCCs per year this work Coke Oven Elissions <150 LCDs per year (44). Vinyl Chloride <27 CDs per year (56). Radicnuclides (world-wide imcact fro:n Derart.^ent of 17 CDs per year (57) Enercyy facilities) Benzer.e <8 CDs per year (58) Arsenic <5 LCDs per year C45) CD = Cancer Death; LCD = Lung Cancer Death Risks for passive sr.ioking and radienuclides are test esti.mates, and risks fer other pDllutants are upper bound. ..,..,:;