Tobacco Documents Online

AUG 03 '94 10:33AM GUMC PHARMACOLOGY 546 REVIEW ARTICLES Passive Smoking as a Cause of Heart Disease A. JUDSON WELLS, PHD Kennett Square, Pennsylvania The effects of passive smoking on ischemic heart disease are reviewed. Short-term exposures of 20 min to 8 h result In incre9sed platelet sensitivity and decreased ability of the heart to receive and process oxygen. Longer term exposure results in plaque buildup and adverse e$ects on blood cholesterol. The available epidemiology fs reviewed, and it is concluded that passive smoking increases the coronary death rate among U.S. never smokers by 20% to 70%. The newest Environmental P.2i10 JACC vol. 24, No. 2 AuQust 1994:346-54 Protection Agency procedures for estimating deaths ffrom passive smoking, when applied to the epidemiologic results on heart disease and passive smoking, Indicate that !n 1985 an estimated 62,000 ischemic heart dLqease deaths in the United States were associated with exposure to environmental tobacco smoke. CIlni• clans are advised to counsel their patients to avoid tobacco smoke at home, at work and In transportation settings. (J Ans CoR Cardlol 1994;24:546 S4) In August 1992 the Council on Cardiopulmonary and Critical environmental tobacco smoke, although they are diluted Care of the American Heart Association published its posi- considerably by the ambient air. Logically, therefore, one tion statement (1) on environmental tobacco smoke and would expect environmental tobacco smoke exposure to cardiovascular disease and "concluded that [environmental result in heart disease but at a level lower than that from tobacco smoke] is a major preventable cause of cardiovas- active smoking. The same argument is used by the U.S. cular disease and death." Their conclusion was based on Environmental Protection Agency in their recent report (7) their own investigation and on two earlier reports by Glantz on lung cancer from passive smoking, that is, that the and Parmley (2) and Steenland (3). The report by Glantz and chemical similarity between mainstream tobacco smoke and Parntley presents an excellent overview of the epidemiology, environmental tobacco smoke means that lung cancer from physiology and biochemistry of the link between passive environmental tobacco smoke is biologically plausible. smoking (the exposure of nonsmokers to environmental However, an analysis of the epidemiology is required before tobacco smoke) and heart disease. The Steenland report the level of risk can be estimated. presents a risk assessment based on epidemiology and There is one important difference between lung cancer concludes that environmental tobacco smoke causes an and heart disease insofar as passive smoking is concerned. estimated 35,000 to 40,000 iscbemic heart deaths/year in the With lung cancer, the potentially fatal health etY'ect results United States. This confirms an earlier risk assessment by only from long-term exposure, perhaps ;!!20 years, whereas Wells (4). However, as noted later, these estimates may be with heart disease, the potentially fatal effects are not only low, long-term and chronic but short-term and acute as well. The present report reviews briefly the principal evidence Acute heart effects from passive smoking. The acute ef- connecting passive smoking with ischemic heart disease, fects of passive smoking on the cardiovascular system can including the most recent data. In addition, 'steps will be occur from exposures of 20 min to 8 h. They are covered in outlined that practicing physicians can take to protect their detail by Glantz and Patmley (2). Suffice it to say here that patients from environmental tobacco smoke exposure. these effects consist of 1) a decrease in platelet sensitivity that leads to greater platelet aggregation and increased risk Biologic Plausibility of coronary thrombosis, and 2) an increase in oxygen de- mand by the heart at a time when oxygen supply is decreased Active smoking is a well known cause of heart disease (5). and the heart's ability to process oxygen is also decreased. There is clear evidence of dose response, and the chemicals The evidence for increased platelet sensitivity comes in mainstream smoke thought to be mo3t important in from the laboratories of J. W. Davis in Kansas City and causing heart disease, namely, carbon monoxide, nicotine H. Sinzineer in Vienna. The blood of smokers has a greater and polyaromatic hydrocarbons (6), are also present in tendency to coagulate than the blood of non-smoke-exposed nonsmokers. This is expressed as a lower platelet sensitiv- ity. Davis et al. (8,1) shodved that exposure of nonsmokers Manuscript received Apri126,1993; revised manuscript received blarch 8, for 20 min to environmental tobacco smoke in an elevator 1994, eccepted Meroh 11, 1994. : Dr. A. Judson Wells, 5 Ingleton Cuole, lobby of a hospital drove their platelet sensitivity more than Kennett 3quare, Pennsylvania 19348. half way H0%) toward the active smoker level. Burghuber 01Wy,yy C{ie Ame,Ycan Coffege ofCardioto:y 073J-1o97/94/57.00

JACC voAUG"03"'94 10:33AM GUMC PHARMACOLOGY AuQust.1994:346-54 et al. (10), in Sinzinger's laboratory, found that nonsmokers, when exposed 20 min to a more intense level of environmen. tal tobacco smoke (30 cigarettes smoked in an 18-m3 room) lost -80% of their platelet sensitivity advantage over smok- ers. In both cases the platelet sensitivity returned to normal shortly after the exposure to environmental tobacco smoke had been terminated. However, Sinzinger and Virgolini (11) found that after repeated exposures of nonsmokers to envi- ronmental tobacco smoke at the same intensity, their base- line platelet sensitivity moved much closer to that of the smokers. In addition to their work on platelets, both labora- tories found an increase in endothelial cell carcasses in the blood of nonsmokers who had experienced short-term expo- sure to environmental tobacco smoke, an indication of arterial wall damage. The experiments by Davis et a1. (12) with sham cigarettes indicate that most of this platelet and cell count effect arises because of.the nicotine in tobacco smoke. The other acute effect that environmental tobacco smoke has on the heart is a complex of effects driven largely by the carbon monoxide in the smoke. Again, the reader is referred to Glantz and Parmley (2) for a detailed discussion and a complete list of references. Another good discussion of the effects of low levels of carbon monoxide on cardiovascular function is the editorial by Dwyer and Turino (13). As noted in the 1986 Surgeon General's report on passive smoking (14 [Table 10, page 151]), carbon monoxide levels in typical environmental tobacco smoke atmospheres range from I to -50 ppm, with many values in the 3- to 25-ppm range. The concentrations of carbon monoxide when environmental tobacco smoke was present were often two to three times the concentration when environmental tobacco smoke was not present. Exposure to 10 ppm of carbon monoxide over 8 h results in 1.4% of the blood hemoglobin being tied up as carboxyhemoglobin (14), and 100 ppm leads to an equilib- rium value of 14% (12). In most human experiments with carbon monoxide or environmental tobacco smoke, the exposure results in nonequilibrium carboxyhemoglobin lev- els of 2% to 4%, which fall within a range that would be attained by equilibrium exposure to many environmental tobacco smoke atmospheres. The reaction of carbon monoxide from environmental tobacco smoke with blood hemoglobin results directly in a reduction of the ability of the blood to transport oxygen. Thus, the heart rate for a given level of activity must increase to maintain the same oxygen supply. There is also evidence (2,13) that carbon monoxide attacks and binds to some of the proteins and enzymes, such as myoglobin cytoehromes (including cytochrome P-450), catalase and peroxidase, some of which are essential to myocardial mitochondrial respiration. The overall effect of these carbon monoxide- or environmental tobacco smoke-induced changes is reduced exercise ability both in patients with coronary artery disease and in young healthy subjects (2). This conclusion is confirmed by recent work of Leone et al. (15) who conducted exercise tests in healthy and postmyo- WSLLP.3/10547 PASSIVE 3MOKiNG cardial infarction subjects in environmental tobacco smoke atmospheres controlled to 30 to 35 ppm of carbon monoxide. There is also evidence from animal models that both short- and long-term environmental tobacco smoke exposure re- sults in reduced cardiac mitochondrial respiration (2,16). Long-term effects from passive smoking. In the previous discussion on platelets it was stated that short-term exposure to environmental tobacco smoke results in increased endo- thelial cell carcasses in the blood. This indicates damage to the arterial endothelium, which is thought to be the initiating step in the development of atherosclerotic plaques (17). Giantz and Parniley (2) present evidence from animal stud- ies, as of 1990, indicating that both increased platelet aggre- gation and the presence of carcinogenic agents can increase plaque formation. These experiments, mostly in chickens, pigeons or mice, indicate that carcinogenic polyaromatic hydrocarbons, such as benzo-o-pyrene, tend to concentrate in the lung and heart, and they accelerate the growth of arterial plaques. A recent report by Zhu et al. (18) states that exposure of New Zealand male rabbits for 10 weeks for 6 h/day to environmental tobacco smoke that contained either 18 ppm of carbon monoxide (definitely in the range of many environmental tobacco smoke exposures) or 60 ppm of carbon monoxide (accelerated exposure) resulted in larger lipid lesions in the aortas and pulmonary arteries than was the case for unexposed rabbits. The mean percent athero- sclerotic involvement from the aortas was 30% for the control animals, 36% for the low dose group and 52% for the high dose group. For the pulmonary arteries, the corre- sponding percents were 22%, 29% and 45%. There was also evidence of greater platelet aggregation in the exposed groups than in the control group. Penn and Snyder (19) studied plaque formation in the abdominal aortas of cocker- els exposed to environmental tobacco smoke that contained 35 ppm of carbon monoxide. Although they found no signif- icant increase in the number of plaques formed in exposed compared with nonexposed cockerels, the size of the aver- age plaque was increased -50%. In humans, Howard et al. (20), using ultrasound measure- ments, found that carotid artery wall thickness was progres- sively increased on going from nonexposed never smokers to environmental tobacco smoke-exposed never smokers to former smokers and finally to current smokers. The differ- ence in wall thickness between the environmentall tobacco smoke-exposed group and the nonexposed group was sta- tistically significant and increased as the hours of reported exposure to environmental tobacco smoke increased. Envi ronmental tobacco smoke also appears to have an effect on cholesterol in nonsmokers. Several investigators (21-23) have found in studies on children that exposure to environ- mental tobacco smoke from their parents resulted in higher total cholesterol levels in the blood and lower levels of high density cholesterol. More recently White et al. (24) found sinuflar results in adults exposed to environmental tobacco smoke in the workplace. This is despite the finding by Le Marchand et al. (25) that nonsmoking wives and presumably

AUG 03 '94 10:34AM GUMC PHARMACOLOGY 548 WELLS PASSIVE 3MOKING P.4i10 JACC Vol. 24, No. 2 August 1994:546-54 Table 1. Epidemiologic Studies of Passive Smoking and Ischemic Heart Disease Among Never Smokers Population No. of or Control Time Study Csee Subjects Fatal or Tier Study Locale Frame Type P M F M End Point Nonfatal Adjustment Factor Asslgned' Butler (27) California 1976-1982 Pros, 1J0# 75* 12,866 1,489 IFID f A 4 Dobson et al. Australia 1988-1989 C/C 160 183 532 293 MI or IHD f+nf A, PIHD 3 (28) Garland et al. California 1974-1983 Pros. 19 (29) 695 IHD f A, B. C. PIHD, W, MS 1 He at aL (30) China 1980s C/C 34 - 68 - MI or CA of A, B, C. D, X. R. RE, 0, 1 FH, FIIID, AL He at al. (31) China 1989-1992 C/C 67 - 135 - MI or CA nf A, B, C, X, E, 0, MS, P, W 1 Hirayama (32) Japan 1966-1981 Proe. 495 - 91,540 - IHD f A 4 Hole at al. (33) Scotland 1973-1988 Pros. 551 651 1,784 671 IHD f A, B, C. SS, W 2 92 46 1,784 671 Angina, ECG nf A. B. C, SS, W 2 Humble at al. Georgia 1960-1980 Pros. 16 - 513 - CVD f A, B, C, SS, W 2 (34) Jackson (33) New 1986-1988 GCd 9 21 62 61 CHD f A, SS, PIHD 3 T.ealand 11 28 112 123 MI nf A, SS, P1FID 3 Lee et al. (36) England 1979-1982 C1C 77 41 318 133 IHD nf A, MS 3 Sandler at al. Maryland 1963-1975 Pros. 988 370 14,873 4,162 AHD f A, MS. E. ris 3 (37)T Svendsen at al. USA 1973-1982 Pros. - 13 - 1,245 MI or IHD f A, B. C, PIHD. W. AL. E t (39) - 69# - 1,245 MI or ECG f+nP A. B. C. PIHD, W, AL, E I Total 2,233 898 •1 w highest tier assignment. ttog mean of Adventist Health Smog cohort and spouse pairs (27). #Adventist Health Smog cohort only, no spouse paly data (27), irUpdated through December 1988 by Hole DJ, private communication, January 1990. JINumbers contain some people later found not to quallfy for inclusion in the analysis. 9Update of earlier report by Helsing et al. (38). #tIncludes the 13 fatal events. A a age; AHD - arteriosclerotic disease: AL ffi alcohol: Ba blood pressure or hypertension; C- cholesterol; CA = coronary aneriography; C/C - easo eonttvl; CHD - coronary heart dlscase; CVD - cardiovascular disease; D s diabetes; E- education; ECG - abnormal electmcardiogram: f= fatal; Fw female; FH = family history of hypertension: FIHD = family history of ischemic heart disease; HS - housing:lHD - ischemic heart dlsease; M= male; M1= myoeardiat infarction; MS m marital status; nf = nonfatal: 0 A occupation; P= personality type; PIHD - personal history of ischetnic heart disease; Pros. = prospective; R- tvice; RE = residence; SS a social status; W- weight or body mass index; X = exercise. children of smokers have less cholesterol in their diets than wives of nonsmokers. Another indication of the striking effect of environmental tobacco smoke on the blood is found in the work of Tribble and Fortmann (26) who measured plasma ascorbic acid levels. Passive smokers exposed to at least 20 h/week of environmental tobacco smoke had a statistically significant reduction of ascorbic acid levels that was 65% of that experienced by active smokers. In summary, as with active smoking, there is ample biologic evidence, including human evidence, that exposure to typical levels of environmental tobacco smoke can cause a buildup of arterial plaque and thus lead to heart attacks. Also there are acute effects related to the nicotine and carbon monoxide in environmental tobacco smoke that may be important mechanisms through which environmental to- bacco smoke causes adverse heart effects. Epidemiology The avaiMlT. epidemiology ar*ocisting passsive Wlnoling with heart disease is displayed in Table I(27-39). With 12 studies and 3,131 cases, it is almost as extensive as that associating passive smoking with lung cancer (30 studies and 3,083 cases [7]). There is no heart study of the quality of the large Fontham et al. (40) study on lung cancer, which was designed specifically for passive smoking and where data from five U.S. centers were combined. In that study both patients and control subjects provided serum samples for cotinine assay to eliminate smoker misclassification effects. However, correction for smoker misclassification bias is much less for passive smoking and heart disease than for passive smoking and lung cancer because the relative risk for heart disease from active smoking is so much less than for lung cancer. Another aspect that some investigators would regard as a strength is that the proportion of studies that are prospective (vs. case control) is much higher here (7 of 12) than in the lung cancer studies (4 of 30) (7). We have no epidemiology that deals specifically with acute effects, in which the subjects were asked whether or not they were exposed to environmental tobacco smoke during the hours immediately before their heart attacks. Whatever, acute effect there is is included with the chronic effects in the recorded fatal and llonfatal coronary events. Each study in Table 1 is assigned a quality tier ranking of 1 to 4 based on the number and importance of the heart risk factors for which corrections have been made. The impor-

AUG 03 '94 10:35AM GUMC PHARMACOLOGY TACC Vol. 24, No. 2 August 1994s546 34 Table 2. Relative Risks Associating Ischemic Heart Disease Morbidity With Passive Smoking w~P.~5i10 549 PASSIVE SMOKING Study Adjusted RRp (95% Cl) Adjusted RR, Corrected for Smoker Misctassifieation Statistical Weight Quality Tier Dose Respoose'" Women He et al. (30) 1.50 (0.63-3.6) 1.50 5.1 1 1.54, 2.30, 5.07, 12.67 He et al. (31) 2.99 (1.13-7.34) 2.90 4.4 1 Significant trend 9 5 I l Combined 2.03 (1.08-3.83) 2.04 . on y Hole et al. (33) 1.13 (0.69-1.86) 1.10 15.6 2 1.07, 1.69 Combined 1.41(0.95-2.09) 1.39 25.1 1+2 Dobson e[ 31. (28)t 2.46 (1.47-4.13) 2,48 14.4 3 Jackson (35) 2.7 (0.37-12.3) 2.71 1.6 3 Lee at al. (36) 0.93 (0.54-1.62) 0.91 12.6 3 Combined 1.51 (1.16-1.97) 1.50 53.7 1+2+3 Men Svendsen at al. (39) 1.61 (0.96-2.71) 1.43 14.3 1 1.20, 1.75 Dobson at ai. (28)1' 0,97 (0.50-1.86) 0.90 8.9 3 Jackson (35) 1.03 (0.27-3.9) 0.92 2.2 3 Lea 1.24 (0.59-2.59) 1.19 7.1 3 Combined 1,28 (0,91-1.81) 1,18 32.5 1+2+3 Women and men Combined 1.77 (1.18 Z,65) 1.65 23.8 1 only Combined 1.48 (1.08-2.02) 1.40 39.4 1 + 2 Combined 1.42 (1.15-1.75) 1,37 86.2 1+2+3 'For He et al. (30). crude RRp for exposures in cigarette-years, 1-199, 200-399, 400-599, 600+: for Hole et al. (33), cigarettes/day, <1S, 15+; for svendsen at al. (39), ciearettes/day,1-19, 20+. tSome fatal cases are included, but the proportion is believed to be small. CI a confidence interval; RR, - passive smoking relative risk or odds ratio adjusted for the factors listed in the next to last column of Table 1. Nare added In proof A further morbidity study of men and womea in Italy has come to the author's attention (La Vecchia C, et al. Lancet 1993;341:505-61(ktter)). Their adjusted Rt~ for highest exposure is 1.30 (9596 CI 0.50-3.40). Inclusion of their results lowers somewhat the combined morbidity RRp but would not change the mortality risk assessment (see Table 4) or the conclusions. tant potential confounders were considered to be age, hy- pertension, cholesterol, weight, social class, marital status, personal history of heart disease, exercise and history of diabetes. Studies that corrected for six of these nine or five plus two others not listed were assigned to tier 1; those that corrected for four of the nine or three plus two not listed were assigned to tier 2; for two of the nine, or one plus two others, tier 3; and for fewer than tier 3, tier 4. The size of the studies was accounted for in the statistical analysis in Tables 2 and 3. Other quality features of the studies were thought to be too subjective to use in the quality ranking. Because it comprises almost half of the total cases, the Sandler et al. (37) study deserves special attention. It is a prospective mortality study in which 98% of the households in a western Maryland county were enrolled. Information on cigarette, cigar and pipe smoking habits for each household member age 16.5 years or older was recorded. An adjust- ment was made to the base population for those leaving the county during the 12-year follow-up. Because there were • ohly 2% blacks, the analysis was restricted to whites. Death certificates were matched against-those enrolled for primary, contributing and underlying causes of death. To determine environmental tobacco smoke exposure, Sandier et al. con- structed a household smoking score based on the smoking contribution of all persons living in that household other than the subject. Information was also obtained on age, marital status, years of schooling and quality of housing, all oI'which were found to have important effects on death rates and smoke exposure status. Hence, relative risks were adjusted for these items, but, unfortunately, no data were available for the other six important heart risk factors in our list. Morbidity studies. The adjusted relative risks and 95% confidence intervals for studies of nonfatal coronary events are shown in Table 2. In two studies (28,39), some fatal events are included, but the proportion is or is thought to be small. For the case-control studies, odds ratios are assumed to be valid estimates of relative risks. The studies are arranged by the tier levels noted in Table 1. Each of the relative risks is corrected for misclassification of smokers as never smokers using the method that was used by the Environmental Protection Agency (EPA) in its lung cancer report (7 [Appendix B]). This method relies on observed gender-specific misclassification rates for never smokers obtained from cotinine measurements or repeated questionnaires. This bias arises because smokers tend to marry smokers. Thus, "never smokers" who are really misclassifled smokers are more likely to be found in the exposed never-smoker category than in the nonexposed category. The result is an overestimation of the relative risk unless the passive risk approaches or exceeds the active smoking risk. The overestimation can be important in lung cancer studies, where the relative risk of ever smokers (those who have smoked at some time in their lives) versus all never smokers is higher (-8.0) but is less important in heart disease studies, where the corresponding relative risk for ever smokers is 1.7. Corrections to the relative risks for women are very small; for men the corrections are larger because the proportion of ever smokers is much higher,

AUG 03 '94 10:36AM GUMC PHARMACOLOGY P.6i10 550 WBLL.S JACC vol. 2d, No. 2 PASSIVE SMOKING August 1994:546-54 Table 3. Relative Risks Associating Ischemic Heart Disease Mortality With Passive Smoking Study Adjusted RRn (95% Cl) Adjusted RRp Corrected for Smoker Misclassification Statistical Weight Quality Tier Dose Response' Trend (p value) Women Garland et al. (29) 2.7 (0.7-10.5) 2.73 2,1 1 3.012.25 Hole et al. (33) 1.65 (0.79 3.46) 1.63 7,0 2 2.09, 4.12 Humble et al. (34)t 1.59 (0.99-2.57) 1.59 16.9 2 1.02.2.11,2.55 Combined 1.68 (1.14 2.47) 1.67 26.0 1+2 Jackson (35) S.8 (0.95-35.2) 5.8 1.2 3 Sattdler et at. (37) 1.19 (1.04-1,36) 1.18 213.5 3 1.20, 1.27 < 0.005 Combined 1.25 (1.10-1141) 1.24 240.7 1+2+3 sutter (27) 1.29 (0.94-1.77) 1.29 38.3 4 1.12, 1.50 Hitayama (32) 1.15 (0.93-1.42) 1.15 85.8 4 1.08, 1.30 Combined 1.23 (1.11-1.36) 1.22 364.8 All 4 Men Svendsen et al. (32) 2.23 (0.72-6.92) 2.62 3.0 1 0.90, 3.21 0.04 Hole et al. (33) 1.73 (1.01-2.96) 1.75 13.3 2 Combined 1.81(1.11-2.96) 1.89 16.3 1+2 Jackson (35) 1.1 (0.23-5.2) 1.06 1.6 3 Sandler et al. (37) 1.31 (1.05-1.64) 1.21 77.3 3 1.38. 1.25 Combined 1.38 (1.13-1.69) 1.30 95.2 1+2+3 Butler (27) 0.54 (0,30-0.96) 0.54 11.4 4 0.41, 0.61 Combined 1.25 (1.03-1.51) 1.19 106.6 All 4 Women and men Combined 1.73 (1.28-2.34) 1.75 42.3 1 + 2 Combined 1.28 (1.15-1.42) 1.25 333.9 1+2+3 Combined 1.23 (1.12-1.35) 1.21 471.4 A114 U.S. women and men Combined 1.75 (1.13-2.66) 1.79 22.0 1 + 2 Combined 1.25 (1.12-1.40) 1.22 312.8 l + 2 + 3 Combined 1.22 (1.10-1.35) 1.20 362.5 A114 =For Garland et al. (29) exposure to ex• and current amokers; for Hole et al. (33). RR9 for 33 cases, adjusted to a8e oniy, exposure to <1S,1S+ ci8arettes/day; for Humble er al. (34), exposure to <10.10 to 20, 21+ ci8areueslday; forSandtcr at al. (37),1-5, 6+ on a smoke exposure score. where 6+ is approximately equal to 10+ cigarettes/day of current exposure (trend data are from Helsing et al. (381, where RRp is higher); for Butler (27), ez- or I to 10, current or 11+ cigarettes/day; for Hirayama (32), ex- plus I to 19, 20+ cigarettes/day; for Svendsen et al. (32), L to 19. 20+ cigarettes/day. tExposure to current smokers only. Abbreviations as in Table 2. leading to more misclassified smokers. The corrected risks for each study are shown because many investigators still worry about the impact of this bias. The statistical weight (wt) for each study is the inverse of that study's variance and can be calculated from chi values and the passive smoking relative risk (RRp) as wt =(chi/ln RRP)2, or from the 95%a confidence limits (CL) as wt -1.962/(in CL - in RRp)~. The combined relative risks are the weighted log means of the appropriate individual values (7). To get some measure of dose response for those studies where data are, available, the relative risk estimates for successive exposure levels are shown in the last column along with p values for trend, if available. Three of the six studies on women show some evidence of dose-response versus one of four among the studies on men. As the EPA report (7) and earlier studies have found, pooled data on passive smoking health effects on men are not as consistent as data on women because male never smokers are subject to more miscella neous environmental tobacco smoke exposure, referred to as "background," relative to spouse exposure than is the case for female never smokers (41). Pooled results for men and women combined are also shown. In general, the pooled relative risks decrease as lower tier studies are added to the analysis. MortaHty studies. The results of the mortality studies are shown in Table 3, where we have more cases, 2,336, than we had for the morbidity studies, and the pooled risks show good statistical significance throughout the table. Five of the seven studies in women show evidence of dose response, but only two of the five studies in men show such a trend. As with the morbidity studies, the relative risks decline as the lower tier studies are added to the analysis. This strongly suggests that a number of heart risk factors affect the results and need to be adjusted for if meaningful passive risks are to be obtained. Omitting the large Sandler et al. (37) study from the analysis has only a minimal effect on the combined point estimates. The combined relative risks for the U.S. mortality studies are quite similar to those for the worldwide studies, and the risks for the mortality studies are consistent with those for the morbidity studies. From the epidemiology, one can conclude that there appears to be a 20% to 70% inerease in ischemic heart disease risk that is associated with exposure to spousal or

AUG 03 '94 10:37AM GUMC PHARMACOLOGY JACC Vol. 24, No. 2 Au¢ut1 1994:546-54 household environmental tobacco smoke and that this in- crease in risk is not explained by misclassification of smok ers as nonsmokers or by the other heart risk factors. Mortallty Risk Assessment Three risk assessments of ischemic heart deaths in the United States from passive smoking have so far been made. They are Wells (4), 32,000 deathslyear; Glantz and Parmiey (2), 37,000 deaths/year (this is the Wells estimate [4] when fully corrected for background environmental tobacco smoke); and Steenland (3) who used a somewhat different procedure to arrive at 35,000 to 40,000 deaths/year. All three of these assessments were based on estimates of never- smoker heart death rates and the estimated population of never and ex-smokers. Never-smoker death rates that are representative of the whole United States are difficult to estimate. Previous risk assessments have depended on rates from the two large American Cancer Society studies (3,4), the U. S. veterans (3), Seventh Day Adventists (3) and the Nurses Health Study (3). The never-smoker death rates calculated from these specialized cohorts may be lower than the U.S. average. In the recent EPA report on lung cancer and other respiratory diseases (7), a different methodology was used based on total lung cancer deaths, including those of smokers. This methodology will be applied here. The EPA method (7), developed by K. G. Brown, starts with the total number of lung cancer deaths, which for women was 38,000. These deaths are then allocated, using the passive and active smoking relative risks, among four categories: 1) deaths from spousal or household environmen- tal tobacco smoke among never smokers and y5-year ex- smokers; 2) deaths from background environmental tobacco smoke, including workplace exposure, if any; 3) deaths from ever smoking less those from environmental tobacco smoke among _5-year ex-smokers; and 4) deaths from non- tobacco-related causes. (See pages 6-8 to 6 21 in ref. 7 for a description of the method.) The EPA made an estimate of environmental tobacco smoke-attributable lung cancer deaths for never-smoking women and then applied the incremental female environmental tobacco smoke death rates to never-smoking males and 5-year ex-smokers of both genders because good passive smoking data were not avail- able for the latter categories. In our heart case, we will start with the known total rischemic heart disease deaths for 1985, namely, 251,000 for women and 285,800 for men (44). As noted in the previous section, the rclative risks appear to increase as the number and importance of heart risk factor adjustments increase. Accordingly, we will calculate deaths for risks from tiers 1 and 2 and from tiers 1, 2 and 3, but the latter will be designated the preferred case because of more deaths and a tighier conddcnce interval. A(,o, i:nlike the problem for lung cancer, the male ischemic heart disease relative risk data are similar to the female data. Therefore, we will assume that never-smoking men and women both have corrected passive P. 7i10 WELLS 551 PASSIVE SMOKINC3 smoking ischemic heart disease relative risks (RR,,)of 1.22 for the preferred case (tiers 1 to 3) and 1.79 for tiers I and 2 based on the combined corrected U.S. data in Table 3, For 2~5•year ex-smokers we have followed the EPA procedure (7) and have assumed that their incremental death rate fromm passive smoking and ischemic heart disease is the same as for never smokers. It is known that an cx-smoker's cardiac risk drops more rapidly after smoking cessation than does his or her risk for lung cancer (45), but this has no influence on the passive smoking risk analysis because we are treating the passive smoking exposure as an add-on effect for ex- smokers. There are two other key variables that affect the risk assessment: 1) the ischemic heart disease relative risk for ever smokers relative to all never smokers (RR9.), which we assume to be 1.7 for both genders (5); and 2) Z, the ratio of environmental tobacco smoke exposure from spousal smok- ing plus other environmental tobacco smoke exposures to exposure from other sources alone. The Z factor is a measure of the background environmental tobacco smoke to which most people in the United States are exposed. It is estimated by comparing current cotinine concentrations in the body fluids of never smokers exposed to spousal smok- ing to the same cotinine measurement for those who are not so exposed (cotinine is a metabolite of nicotine with a longer half-life). It is further assumed that the increased ischemic heart disease mortality risk from environmental tobacco smoke among never smokers is directly proportional to their cotinine assay. In general, a lower Z means relatively more background environmental tobacco smoke exposure, a higher relative risk estimate when corrected for background, and a higher number of calculated deaths. The EPA, in their lung cancer report (7), used Zt (for women) = 1.75 based on five U.S. studies. We have chosen a more conservative value of Zr = 2.6 based on the median value for cotinine for the never-smoking women in the Fontham et al. (40) study, which is the largest U.S. passive smoking/lung cancer study based on five population centers largely in the southern part of the United States, where most of the U.S, studies on heart disease and passive smoking are based. Men are thought to be exposed to more background environmental tobacco smoke than women. Cummings (41) found, for a northern United States locatfon, Z g 1.55 for women and 1.27 for men. Using the ratio of Cummings, we have assumed Zm (for men) - 2.6 x (1.27/1.55) - 2.1. Other important variables, which are the same for heart disease estimates and lung cancer estimates, are taken directly from the EPA report (7), namely, U.S. population z35 years old, 58 million women and 48 million men; ever-smoking prevalence (pl) of 44.3% for women and 72.8% for men; never-smoker spousal passive smoking exposure (p2), 60% for women and 24% for men; -5-year ex-smoker population 8.7 million women and 15.0 million men; and the fraction of these ex-smokers exposed to spousal smoking, 77% for women and 41% for men. The calculated numbers of deaths are shown in Table 4.

AUG 03 '94 10:38AM GUMC PHARMACOLOGY Ss2 PASSIVE SMOKINO August 199k546-J4 P.8i10 ..z..:,.a 7ACC Vol. 24. No. 2 Table 4. Estimated Annual ischemic Heart Disease Deaths in U.S. Nonsmokers Attributable to Passive Smoking, 1985: Effects of Selected Variables Esdmated Deaths Variation RR,C RR„n 7.r 7. Women Men Both Preferred case (tlers 1+ 2 + 3) 1,22 1.7 2.6 2.1 33,198 28,714 61,912 Tiers 1+ 2 only 1,79 1.7 2.6 2.1 91,551 91,253 182,804 Preferred case with Lower 4 1.22 1.7 2.1 2.1 40,677 28,714 69,391 Higher RR,, 1.22 2.3 2.6 2.1 27,599 22,270 49,869 Higher Z. 1.22 1.7 2.6 2.6 33,198 22,145 55,343 Lower RR, 1.18 1.7 2.6 2.1 27,750 23,710 51,460 RRw - passive smoking relative [isk from spousal exposure, unadjusted for background exposure but corrected for smoker misclassification; RR,,, = ever smoker relative risk versus all never smokers; Zf (7,,,) = ratio of environmental tobacco smoke exposure from spousal smoking plus other sources to exposure from other sources alone, for women (men). The preferred case, where the passive relative risk, RRPC, is 1.22, the active smoking risk, RR., is 1.7, the Z ratio for women, Zr, is 2.6, and that for men, Zm, is 2.1, yields an estimated total number of passive smoking ischemic heart disease deaths in the United States of 62,000 for 1985. The calculated deaths when data from tiers t and 2 only are included are 183,000. The estimated numbers of deaths are very sensitive to RR,,, RRSm and Z. Table 4 shows the effect of varying these variables. However, none of the estimates is <49,000. The estimated annual deaths shown in Table 4 are con- siderably higher than the 32,000 to 40,000 noted in earlier studies. One reason is that the never-smoker death rates that we derive from the known total ischenlic heart disease deaths in 1985 are higher than the never-smoker death rates assumed in the earlier risk assessments. Another reason is their use of Z= 3.0 based on early work in England (45), whereas we have used the lower U.S. values. An estimated allocation of ischelnic heart disease deaths among the four categories for the preferred case (top row, Table 4), broken down into never smokers, 5-year ex- smokers and other ever smokers, is shown in Table 5. The percents for each category of the 536,800 total U.S. ischemic heart disease deaths are 4.5% for exposure to spousal tobacco smoke, 7.0% for exposure to background tobacco smoke, 34.5% for ever smoking and 54.0% for nontobacco causes. Thus, environmental tobacco smoke is estimated to cause (4.5 + 7.0)/34.5 = 33% as many cardiac deaths as those caused by active smoking. The relative cardiac risks for passive and active smoking for women, 1.22 and 1.7, become 1.41 and 2.23 when the comparison group for both is never smokers with no environmental tobacco smoke expo- Table S. Passive Smoking Mortality for lschemic Hean Disease by Attributable Sources for United States, 1985 Smoking and Exposure Status Spousal Environmental Tobacco Smoke Background Environmental Tobacco Smoke Ever Smoking Nontobacco Causes All Causes Never smokers Women 12,444 12,962 , 81,309 106,714 Men 2,583 9,783 39,130 51,496 Both 15,207 22,745 120,439 158,211 k5-year ex-smokers Women 4,301 3,491 21,896 29,60 Men 5,110 11,239 44,957 61,306 Both 9,411 14,730 66,853 90,994 Other ever smokers" Women 71,826 42,771 114,597 Men 113,224 59,774 172,998 0 .~ Both 185,050 102,545 287,595 ca Total Women 16,745 16.453 71,826 145,976 251,000 La r,] Men 7,693 21.022 113,224 143,861 285,800 0 ~ Both 24,438 37,475 18S,050 289,837 536,800 ts 'Includes ever smoking e$ects on k5-year ex-smokers, thought to be small.

AUG 03 '54 10:38AM GUMC PHARMACOLOGY JACC Vol. 24, No. 2 AuBaet 1994:546 5e sure of any kind (RR0, and RRaI in the notation of the report [7} from the EPA). For men, RR. = 1.53 and RRo1 = 2.24. Thus, the excess risks, 0.A1/1.23 and 0.53/1.24, exhibit ratios of 33% and 43%, respectively, for women and men. That such high ratios between the mortality effects of passive and active smoking might be biologically plausible is supported by the dramatic platelet sensitivity effects noted earlier (passive effects at 60% of smoker level) and the 20% to 50% increase in lipid lesion area in animals after environmental tobacco smoke exposure (18,19). To summarize, on the basis of the rather thorough tneth- ods of the EPA to calculate deaths from passive smoking, the cardiac deaths so calculated are 40% to 100% higher than those calculated by earlier methods. What is thought to be the best estimate in Table 4 indicates 62,000 ischemic heart disease deaths in the United States in 1985 caused by passive smoking. It is also possible, if more information about potential confounders can be obtained, that the toll is even higher. Discussion The risk assessment used here assumes a "no-threshold" model. With respect to heart disease and active smoking, the 1983 Surgeon General's report (5 [p.119]) notes that there is no evidence to suggest a threshold for the effect, that is, no safe level of exposure. Likewise, when the relative risks for ischemic heart disease and passive smoking are plotted against exposure variables, they trend toward zero risk at zero exposure with no evidence of a threshold. The "attrib- utable risk" approach was used in the risk assessment primarily because the EPA had used it for lung cancer, and it has been extensively peer reviewed by their Science Advisory Board. The method is appealing because it does not rely on estimates of never-smoker death rates. A base year of 1985 was used to allow direct comparison with the EPA results (7) on lung cancer and because good ischemic heart disease death data were available for that year. Since then many smokers have quit, resulting in less spousal exposure, and many workplaces and public places have restricted smoking. A projection to 1994 assuming (arbitrarily) one-half as much background exposure, a pro- jected 9% lower heart death rate and spousal exposure reduced in proportion to the reduction in spousal active smoking indicates ^-47,000 environmental tobacco smoke deaths instead of the 62,000 estimated for 1985. A 14% increase in population :05 years old offsets to some extent ,the reductions in heart death rate and cigarette smoke exposure. The spousal exposure rates that were used in the analysis (60% for women, 24% for men) are for exposure to ex-smokers as well as current smokers and came from the exposure of control subjects in the U.S. lung cancer studies in the EPA report (7). They appear to be appropriate for 1985, They are thought to be more representative than the somewhat higher exposures among the control subjects in the U.S. heart studies (65% and 29%), which are dominated by the large Sandler et al. study (37). yygLL3P.9/1053 PASSIVE SMOKING The major uncertainty in the ischetnic heart disease death estimates and in whether or not a passive smoking effect exists arises because of the many other factors that might cause cardiac disease and death and might confound the passive smoking effect. Supporting evidence for an effect is the strong evidence of biologic plausibility, and, as adjust- ments are made for more cardiac risk factors, the passive smoking relative risks tend to increase. One factor not accounted for in our analysis is nicotine from diet sources that, if present, would result in a falsely high estimate of the background effect. Domino et al. (47) have published nico- tine contents of various vegetables, but Repace (48) has presented an analysis indicating that these amounts would constitute only 0.7% of the nicotine absorbed by typical passive smokers. Deaths calculated using the extremes of the 95% confidence interval values for the relative risk of 1.22 for the preferred case in Table 4 indicate a range of 26,700 to 98,400 deaths/year. Another uncertainty is publication bias. The author is aware of two epidemiologic studies •of passive smoking and heart disease that did not get beyond the abstract stage. Hunt et al. ([49J and private communication, August, 1987) in a morbidity study in Utah found an adjusted relative risk of 2.7 for spouse-exposed, never-smoking women, but there were only 23 cases. Palmer et al. (50) found a relative risk for myocardial infarction of 1.2 for spouse-exposed, never- smoking women, but they decided that the power was too low and did not follow up with a published report. Inclusion of these studies would have little effect on the results. There is also an early study showing that exposure to cigarette smoke increases myocardial infarct size in dogs (Prentice et al. [51]) that apparently did not mature into a published report. Conclusions Ischemic heart disease appears to be by far the major mortality risk from passive smoking. There are both short- and long-term cardiovascular effects from passive smoking. Practicing physicians would do well to warn their at-risk heart patients to avoid smoky rooms. Such patients and the public in general should seek work where no-smoking rules apply; they should avoid riding in automobiles when others are smoking; and, if their spouses smoke, it should be suggested that they smoke outside the home. Although the case for ischemic heart disease from passive smoking is well established, a large study, like the Fontham et al. (40) lung cancer study, where all of the other important risk factors are controlled for, would be helpful. Also, it would be useful to have some autopsy research in humans who have been exposed long term to environmental tobacco smoke, but who died from causes other than heart disease, to see whether higher levels of plaque had built up in their coronary arteries. Some epidemioiog°,~ on mortality effects, if any, from very short term environmental tobacco smoke exposure would also be helpful.

RUG 03 '94 10:39AM GUMC PHARMACOLOGY 554 WELLS PASSIVE SMOKING References 1. Taylor AE, Johnson DC, Kazemi H. Environmental tobacco smoke and cardiovascular disease; a position paper from the Council on Cardiopul- monary and Critical Care, American Heart Association. Circulation 1992:86:699 702. 2. Glantz SA, Parntiey WW. Passive smoking and heatt disease: epidemiol- ogy, physiology, and biochemistry. Circulation I991;83:I-12. 3. Steenland K. Passive smoking and risk of heatt disease. JAMA 1992;267: 94 9. 4. Wells AJ, An estimate of adult mortality in the United States from passive smoking. Environ lnt 1988:14:249-65. 5. US Surgeon General. The health consequences of smoking: cardiovascu- lar disease. A report of the Surgeon General. Rockville (MD): US Depatt- meat of Health and Humaa SetvIces, Office on Smoidng and Health,1983:1. 6. Albert RE, Vanderlaan M, Burns FJ, Nishizumi M. Effect of carcinogens on chicken etherosclerosis. Cancer Res 1977;37:2232-5. 7. Respiratory health effects of passive smoking: Iung cancer and other disorders. Washington (DC): US Environmental Protection Agency, EPM600/6-90d006F, 1992:4-28,5-4-7,5-30,6•1 20.B•1-28. 8. Davis JW, Shelton L. Watanabe 13, Arnold J. Passive smoking affects endothelium and platelets. Arch Intem Med 1989;149:386-9. 9. Davis JW, Shelton L, Zucker ML. A comparison of some acute effects of smoking and smokeless tobacco on platelets and endotbelium. I Vasc Med Biol 1990;2:289-93. 10. Burghuber OC, Punzengruber C, Sinzinger H, Haber P, Silberbauer K. Platelet sensitivity to prostacyclin in smokers and non-smokers. Chest 1986;90:34-8. 11. Slozinger H, Virgolint I. Besitzen Paesivraucher ein erhohtes Thrombo- serisiko? Wien Klin Wochenschr 1989;20:694-8. 12. Davis JW, Shelton L, EigenbergDA, Hignite CE, Watanabe IS. Effects of tobacco and non-tobacco cigarette smoking on endothelium and platelets. Clin Pharmacol Ther 1985:37:529 33. 13. Dywer EM, Turino GM. Carbon monoxide and cardiovascular disease. N Engl J Med 1989;321:1474-5. 14. US Surgeon General. The health consequences of involuntary smoking: a Report of the Surgeon General. RockvOie (MD): US Public Health Service, DHHS (CDC) 87-8398, 1987:151,202. 15. Leone A, Moti L, Bertanelli F, Fabiano P, Filippelli M. Indoor passive smoking: Its effect on cardiac performance. lnt Cardloi 1991;33:247-d2. 16. van laarsveld L, Kuyl JM, Alberts DW. Exposure of rats to low concentration of cigarette smoke Increases myocardial sensitivity to ischaemia/repettusion. Basic Res Cardlol 1992;87:393-9. 17. Ross R. The pathogenesis of atherosclerasis-an update. N Engi I Med 1986;314:488-500. 18. Zhu B-Q, Sun Y-P, Sievers RE, Isenberg WM, Glantz SA, Psrniley WW. Passive smoking increases experimental atherosclerosis in cholesteral-fed rabbits. J Am Coll Cardiol 1993;21:225-32. 19. Penn A, Snyder CA. Inhalation of sidestream cigarette smoke accelerates development of arteriosclerotic plaques. Circulation 1993:88: Pt 1:1820-5. 20. Howard G, Szklo M, Evans G, Tell G, Eckfeldt J, Heiss G. Passive smoking and carotid artery wall thickness: the ARIC study [abstract]. Circulation 19920:862. 21. Moskowitz WB, Mosteller M, Schieken RM, at a1. Lipoprotein and oxygen transport alterations in passive smoking ptradolesccnt children: the MCV twin study. Circulation 1990;81:586-92. 22. Feldman 1, Shenker IR. Etzel RA, et al. Passive smoking aiters lipid profiles in adolescents. Pediatrics 1991;88:259-64. 23. Pomrchn P, Hollarbush J, Clarke W, Lauer R. Children's HDL-chol: the -effects of tobacco smoking, smokeless and parental smoking. Presented at 30th Annual Conference on Cardiovascular Disease Epidemiology, March 29-31,1990: San Diego (CA). 24. White JR, Criqui M, Kulik JA, Froeb HF, Sinsheimer PJ. Serum lipoproteins in nonsmokers chronically exposed to tobacco smoke in the workplace. Presented at 8th World Conference on Tobecco or Health, Buenos Aires, March 30-Apti13, 1992. 25. Le Marchand L. Wtlkens LR, Hankin JH, Haley NJ. Dietary patterns of &msle rt;-inROiokert vrit#s and wJthnut exposure to en4innmentcl toir::rcao smoke. Cancer Causes Control 1991;2:11-6. 26. Tribble DL, Fortmann 3P. Reduced plesma ascorbic acid concentrations P.10i10 JACC Vol. ?A, No. 2 August 1994:546-54 in women regularly exposed to environmental tobacco smoke (ETS) [abstract]. Circulation 1992;86: Suppl I:1-675, 27. Butler TL. The relationship of passive smoking to various health out- comes among Seventh Day Adventists in California (dissettation]. Los Angeles (CA): Univ of California. 1988:160,173. 28. Dobson Al, Alexander liM, Heller RF. Lloyd DM. Passive smoking and the risk of heart attack or coronary death. Med J Aust 1991;154:793-7. 29. Garland C, Barrett-Connor B, Suarez L, Criqui MH, Wingard DL. Effects of passive smoking on ischemic heart disease mortality of nonsmokers. Am J Epidemiol 1985t121:645-50. 30. He Y, Li L•3, Wan Z, et al. Women's passive smoking and coronary hean disease. Chin J Prev Mad 1989;23:19-22. 31. He Y, Lam TH, Lti LS, et al. Passive smoking at home sad at work as risk factors of coronary heart disease In nonsmoking women !n Xian, China. Abstract no. 169, Third International Conference on Protective Catdinl- ogy, Oslo, June 27-July 1, 1993. 32. Hlrayama T. Passive smoking. NZ Med J 1990;103:54. 33. Hole DJ, Gillis CR, Chopra C, Hawthorne VM. Passive smoking and cardiorespiratory health in a general population in the west of Scotland. Br Med J 1989;299:423-7. 34. Humble C., Croh J, Gerber A, Casper M, Hames CG, Tyroler HA. Passive smoking and 20.year cardiovascular disease mottaGty among nonsmoking wives, Evans County, Georgia. Am I Public Health 1990;80:599-601. 35. Jackson RT. The Auckland Heart Study (dissertation]. Auckland, Now Zealand: Univ of Auckland, 1989:157-172. 36. Lee PN, Chamberlain J, Alderson MR. Relationship of passive smoking to risk of lung cancer and other smoking related diseases. Er I Cancer 1986;54:97-105. 37. Sandler DP, Comstock OW, Helsing KJ, Shore DL. Deaths from all causes in nonsmokers who lived with smokers. Am J Public Health 1989:79:163-7. 38. Helsiug KJ, Sandler DP, Comstock GW, Chee E. Heatt disease mortality in nonsntokers living with smokers. Am J Epidemiol 1988:127:915-22. 39. Svendsen KH. Kuller LH. Martin MJ. Ockene JK. Effects of passive smoking in the Multiple Risk Factor Intervention Trial. Am J Epidemiol 1987:126:783-95. 40. Fontham ETH, Correa P, Wu-Williams A, et al. Lung cancer in nonsmok- ing women: a multicenter casacontrol study. Cancer Epidemiol Blomark Prev 1991;1:35-43. 41. Cummings KM. Statement before the US Environmental Protection Agency Science Advisory Board Indoor Air Quality and Total Human Exposure Committee. Environmental Tobacco Smoke Review, Arlington (VA), December 4, 1990. 42. Hammond C. Smoking in relation to the death rates of one million men and women. In: Haenszel W, editor. Epidemiological Approaches to the Study of Cancer and Other Chronic Diseases. Bethesda (MD): US Public Health Service, 1966:127-Z04. 43. Stollman S, Garfinkel L. Smoking habits and tar levels in a new American Cancer Society prospective study of 1.2 million men and womea. J Natl Cancer Inst 1986;76:1057-63. 44, Health United States 1989. Hyattsville (MD): US Department of Health and Human Services DHHS(PHS) 90-1232, 1990:123-4. 45. US Surgeon General. The health benefits of smoking eessadon. A report of the Surgeon General. Roekville (MD): US Department of Health and Human Services, Office on Smoking and Health, 1990:v,vl. 46. Wald NJ, Nanchanal K, Thompson SG. Cuckle HS. Does breathing other people's tobacco smoke cause lung cancer? Br Med J 1986;293:1217-22. 47. Domino EF, Hornbach E, Demana T. The nicotine content of common vegetables [letter]. N Engl I Med 1993;329:437. 48. Repace JL. Dietary nicotine won't mislead on passive smoking. petter]. 8r Med J 1994;308:61. 49. Hunt SC, Martin MJ, Williams RR. Passive smoking by nonsmoking wives is associated with an increased incidence of heart disease in Utah. Presented at Ametican Public Health Association meeting, Las Vegas (NV), October 1, 1986. 50. Palmer JR, Rosenberg L, Shapiro S. Paseive smoking and myocaedial infarction in women [abstract]. CVD Epi Newsletter No. 43, winter,198g. 51. Prentice RC, Csrroll R, ScOion PJ. Thomas JK. Itecent expoeure to cigarette smoke increases myocardlad infarct size [abstract]. I Am Coll Cardio11989;13:124A.