Tobacco Documents Online

/Review Passive Smoking and the Risk of Heart Disease Kyle Steenland, PhD Ob)ective..-This paper reviews the evidence that exposure to environmental tobacco smoke (ETS) increases the risk of heart disease death among persons who have never smoked (never-smokers). The annual number of heart disease deaths in the United States attributable to ETS is estimated, as is the indi'vidual risk of heart disease death for exposed never-smokers. Dete Sources.-Nine epidemiologic studies and numerous experimental'stud- ies are available to evaluate the association of ETS and heart disease. Date Synthesfs.-The relative risk for never-smokers iiving with current or former smokers, compared with never-smokers Iiving, with nonsmokers, has ranged from 0.9 to 3.0in nine studies. Seven studies were positive, one was pos- itive for women but not men, and'one was negative. Several studies have shown~ a dose-response relationship and: have controlled for other risk factors. Evidence from experimental studies suggests that'ETS can damage the cardiovascular sys- tem, via both short-term and long-term mechanisms. Assuming that the observed heartdisease nskfor those exposecf to ETS is not an artifact of misclassification or ..oonfounding, approximately 35 000 to 40 000 deaths from ischemic heart disease among never-smokers and long-term former smokers are estimated to have oc- curred annually in the United States as a result of ETS exposure in the early 1980s. An individual male never-smoker living with a cun-entor former smoker is estimated to have an approximately 9.6% chance of dying of ischemic heart disease by the age of 74 years,,compared with a 7.4% chance for a male never-smoker living with a nonsmoker. The corresponding lifetime risks for women are 6.1% and 4.9%. Conelusibns.-The public heatth~burden due to ETS exposure is likely to be much greater for heart disease than for lung cancer, which, has been the focus of most debate to date. Individual lifetime excess risks of heart disease death due to ETS of one to three per 1 DO can be oompared ;wifh much lower excess risks of one death per, 100000, which are often used ini determining environmental limits for other toxins. Exposure to ETS is not currently regulated at the federal level, except for domestic air traffic. (JAMA. ]99ZZS79499). ENVIRONMENTAL tobacco smoke (ETS) has been associated with a vari- ety of diseases, particularly lung can• cer. In 1986, the National' Research Council' estimated that about 3000 lung cancer cases per year among,persons in the United States who had never smoked (never-smokers) were attributable to ETS. In 1990, the Environmental Pro- tection Agency published a draft report reaching similar conclusions.= VPhi.le the hung cancer risk among never-smokers exposed ~ to ETS is well established, a possible risk of heart disease due to ETS is more controversial. Yet the epidemi- FromtneNWUOna3 Waa,ne Ia.C?ccupaarW isrt.ry .na rwartn. Caiurwu; pva ri.unrn r.a+eas to me raWOW titeet,ne ta Oeeu pnqryi Satety.nohManrt. MMttopR1l 4676 Co urwia wkwy. Crcm,.t,.,oH 4an6 ()k Suw"nd) ) ologic evidence for a heart disease effect has been increasing in the last severali years. This article discusses the avail- able data on ETS and heart disease. Based on the assumption that the epi- demiologic studies are reasonably accu- rate, the annual 'number of deaths in the United States due to ischemic heart dis- ease (IHD) attributable to ETS is esti• mat.edi as well as the individual lifetime risk of IHD death due to ETS. DATA ON ETS EXPOSURE Environmental tobacco smoke is dif. ficult to measure directly. Indirect mea- sures that have been used are airborne respirable suspended particulate (de- Sned as particles of less than 2.5-µm diameter) 1 and urinary cotinine (a me- tabolite of nicotine). Passive monitorss of vapor-phase nicotine are a promising new direct method to measure ETS.' Repace' has shown that the back- ground level of respirable suspended particulate (approximately 20, µg/mf) doubles in homes in which a smoker lives. ETS exposure also occurs outside the home. Approximately 28% of the US adult population smokes, and ETS ex- posure occurs in most indoor environ- ments. Cummings et a]5 have show-n that 919E of 663 nonsmokers had cotinine in theii• urine, including 81% of the 162 subjects who reported' no exposure to ETS in the previous 4 days (the rele- vant period for cotinine measurement), The average level of cotinine in the urine of nonsmokers was about 8 ng/mL, com, pared with about 1200 ng/tnL in~smok- ers. Other investigators" have shown that nonsmokers living with, smokers have 2.5 to 3 times the leve] of urinary cotinine compared with that of nonsmok- ers living with nonsmokers. The relative contribution of ETS ex- posure at work to total'exposure is not well known. Nonsmoking restaurant workers (perhaps a worst case for oc- cupational ETS exposu,re)! averaged 566 ng/mL of urinary cotinine in one study.F Conversely, Haley et ah have shown in. a limited sample that urinary cotinine for those exposed at home and at work increased only slightly compared with those exposedonly ati home. The National Research Council' con- e]uded that nonsmokers exposed to pa=- si've smoke are absorbing, the equiva- lent of 0.1 to 1.0 cigarettes a day, based on urinary cotinine levels. However, the constituents of sidestream smoke are different from those of inhaled~ tnain- stre8m amoke. Sidestre.am smoke is gen- erated at a lower temperature than mainstream smoke, the partic]e size is smaller, less of the generated smoke is particulate, and the pH is higher.' There are more carbon monoxide and ~nicotine breakdown products in dilute sidestream smoke than in mainstream smoke. These differences imply that it is difficult~ to determine the relative toxicity of side- stre,am smoke vs mainstream smoke. Consequently, arguments inferring ETS 04 JAAAIl Janusry 1, 1D92-Vol''26T, No. 1 Passne S+mking and F<bart D+sease-SZeentand i I p I

Tadr 1.-Hi.rt oo.a.e M1onp Ns..nSno+wrs Due ro ETS• ng 3 rk- ed n') ?S he ~S °x• )n- :at in _&2 to le- ~ ). .ne m- sol.w Ty" u+e sEe. Hdw st 4t•, 12-y bb.-up 3960 m.L 4037 .Onrr, lianbM 01801 20-y b6omyt S13>,onrn apsd 40+ y.. MYY4qNN 12-y'bbrUq4162mns.14:673 wame~ 6po 25* y.n 1963 11.er RsNO (96% C7) (lio. d.Op..rv.d Erpo... o..tlu ) con.n.on llynp wnh .moker or a-wrck.r,n 2.01 (1213.35) AdWbO la 3 CV rwck tacbn .nC. ..r1y 1970s (sK4.Oor1)', 14551 .0C1.+ ti+Y. do»4 response L1.4fQ Mt+ amokr n 1960 (sv& 1.50 (o.9a2.57) (76) AOtisuC br 3 CV rrk tactors, tl0ae 1.por1)) w.porw n 4onr.trns UKp.Mnvnch.rorbcamokw n 1.31:en.n6mddrs4r~fpo~wb.onrnony„ 1963 (oHoi.porl); amowrn b (492) mMsW K+ sawcojo^ en7rtrok4fs p/r.n Yws r.wQIM114 ronrn (1.1-1.4) (1539) 6wwb..n 4l 7-y M.Wap4lo/o.up. br 1245 m4n ~ M.mW b arwh,*r (e4porMd'by 1.61 (0.96-2.71) (90) Porm.+ dorerrspon+e ( P. .2). r`K1rtl.3S37y.1Y73-1a62.MRFT7 A+rG.rd): a[1rus1e0br3CVrrkMOrs4rd rWy (hCe.NUi men) erxc.fto^ 2A1 (P<.1) (19; ady. Ap.-~ pc>wl`m da.a. 2 n namPw.0 w.porw. rna 4.+ps ff.ks p.rfor+4):r.udts usst.Db I >x- lot :nt oc- 5G; ti. e .in ine ;rk it'h, )n- as- 1-a- red -he am ,in- en- ian ?re ine am ese ke:" TS and (iNlWd s1 10-y lnbr-Jp 05 Wamrn 4p40 M.rn.d to Wrtoto' Or w«++d.r 4P' 60-79 y n ID72-1974 (08wport) . 1*~ynf~` 16y 10kwW !!q, 91 840. MuAW b ~ACk4t or w-unOk'M 1.10 (bw)..1.31 $qWhcAnl Oo44 rwporN, bw.bwonrn apsd 40+ y (Miyl) ~(NO Cn (494) hph expo.un Lw fl W M ca.n. 162 aorcvK 26 m>tlbFist mvn.p. b r+wkrt a a-124 (1rn),:0.93 No Ooreroaposr la. Oui ETS ors, 22/.m.r es.c .rokr (w6ropor1) 1townen)(NS) exaosura (ros .oa.e none). lbspt.~-t+4.ao FN~ 34/.nd. ta.s, 66 oontno/e M.m.d b 4nnka Ior S or maty 3.003 Pcortrve I - - respons, hDODOAP casee: L.au ertaes on n.ewos Dob.on s1 N"• 16o Ivn.N wes. 163 mM A4s.. S.A-npono0 hortr cooYrte (9rtr 0.97 (msn~ . Ea-aokers had.4iateC ~rnA (t.76 715.oomeuh p.nod nao tlsnsd) 2.46 (wan.n)S man. 1.46 .omen I.i ro soess rex br. Workp4Ycq .xposurs, . voPLArbwl,c.w 'ETS nQr.afes.nvrannwMa'nOaxo Yru1c4; Cl, eoMd.no./re.r-ali IHD: uCtis+rc Asaf1 d.wx; CV,.tardlwasalar. MRF(T,i MWppieRW Gactor nf4ne+mm Tiul. 4n0 MS, nol {*v4c.n1. IMD erornfly (dYWfyd Y MMf4DOna/ CIasWRCabon dDAMs»S )ICOJ COd4a 4104rRuqM 414) a M.4rd po'Y bv N' iOuda e[CSpO fMlniWe et &7.. ° wt.M 1sW aA ; CV. o4.er ( ICD d.dhambrti 3DO arcwpli 4B6a .nd L.. a:v. ° M., - Dop.on a.I." .rd S.ne..n .(,.g,' slacs~ u.w nvd.rrs.. tEs+rn.ud hmm data in arew. jSqwrant am u+0.esrnwrOwv.1d health effects based on known health effects of mainstream'smoke (cigarette "equivalency") are not sppropriate. HEART DtSEIISE AND ETS Epidemiologicavidence has been in- creasing that passive smoking at home is related to heart disease among never. smokers. Earlier reviews'•' concluded' that the hypothesis associating ETS and beart disease was biologically plausible but that epidemiologic and ezperimen- tal, data were inconclusive. A review of the recPnt literature shows that six new epidemiologic studies'ou have been published regarding ETS and heart disease. Table 1 sununarizes all nine ep- idemiologic studies;'°" Seven are posi- tive, while one is positive for women but not' men. The five best-designed and larg- eat studies'"Ll-1"'are cohortscudies; three of the five controlled for the principai' cs,rd'iovasenlar risk factors (cholesterol,: blood pressure, and obesity), and three showed a positive dose response, while the other two showed a positive doee re- sponse for eertain subgroups. A recent review of mo6t of these studies" concluded that "heart disease is an important con- sequence of exposure to ETS" and esti- mated that' the excess risk of heart dis- ease for nonsmokers living with smokers was 30%. The principal weaknesses in the epi- demiologic evidence to date have been the indirect methods of assessing expo- sure (via spousal smoking) and the lack of data on exposures to ET5 outside the JMMk Jarcsry 1. 1962-Vo1 267, ra. 1 home. If the effect of ETS on the cor- onary system is long-term; early expo- sures during childhood might also be important, butchildkiood exposures have not been considered in the epidemio- logic studies. Also; there are many risk factors for heart disease„ and it is dif- ficult to control well for all of them. Another problem with the epidemio- logic data is the seemingly large effect that ETS has on heart disease compared with the effect of mainstream smoking. Active smoking is associated with heart' disease, with a relative risk of smokers vs nonsmokers of about' 1. Z7D Most stud- ies of never-smokers living with smok- ers indicate relative risks on the order of 1.2 to 1.3, compared with those of never-smokers living with never-smok- ers. These relative risks seem high com- pared with the risk for mainstream smoking. There are several counterar- guments, to this objection. Studies of mainstream smokers have used refer- ent groups,of never-smokers composed of never-amokers eapoeed''to ET'S and never-smokers not exposed to ETS, so that relative risks from mainstream and passive smoking studies are not directly comparable. Another argument is that dosimetry based on cigarette equiva- lents is misleading, since sidestreamn smoke is qualitatively different than' mainstream smoke, and exposure to sidestrcam smoke may be proportion- ately more toxic to the heart than ex- posure to mainstream smoke. Due to the relatively slight inaessed risk of heart disease for passive smokers and the many factors known to affect heart disease, the possibility of uncon- trolled confounding as a cause for the in- eea9ed risk camTot be ruled out. Con- fwulding by dietary factors might bias disease risks for passive smokers up- ward. This suggestion is based on findings that never-smokers living with smokers have less nutritious diets,than never- amokers living with nonsmokers.1'a The argument has prinapelly been made for lung cancer risk: and' has focused' on food containing carotenoids or retinoids, which are protective against lung cancer. A similar argument might be made for heart disease. However, severali of the _ heart disease studies have adjusted for cholesterol, the most established heart disease risk factor relhted~to diet. Fur- thermore, in one stvdy,"'never-amokersliving with smokers ate significantly less cholesterol-containing food than never- smokers living with~ nonsmokers (while ~ also eating significantly less canote.noid- ~ oontaining vegetables). While earo- N tenoids and retinoids may be protective against heart disease as they are against CJ lung cancer, little published data support ~ this claim. ~.i An increasingly substantial body of an- I-A imal and human experimental evidence supports the hypothesis that ETS in- ~ creases the risk of heart disearae. The 1386 ~ National Ftesearrh Council report noted Cr1 that levels of carboxyhemoglobin (COHb), amongtlio6f higtilyexpo6ed to ETS,was reported to be close to 3% and'also noted Pass+ve Smdcioq anC Heart~Dsease-Secn(arxi fS

that animal and theoretical' models sug- gested such levels might have adverse effects on the heart. However, the Na- tional' Research Council found little evi- dence that ETS exposure in healthysub- jects was detrimental. Since 1986 a series of new experimentgl studies have been reviewed by Glantx and Parmley,3° who cite human ~evidenee that ET5 exposure (1) increases COHb and'adversely affects exercise performance in both heart pa- tients and healthyindividuals; (2) increases platelet aggregation (atJevels slightly less than those aeen in active smokers) and' adversely affects platelet function; and (3) damages the arterial! endothelium (again at levels slightly less than those seen in active smokers). They also cite animal evidence that components of~ETS (eg, polycyclic aromatic hydt^ocarbons)'csn increase the risk of atherosclerotic plaques. Hence, ETS might be thought! to have either short-term effecLQ on the heart (via COHb or thrombosis) or long- term effects (via endothelial damage and plaque development): Two new human experimental studies lend further support t.ot.he adverse effectt of ETS on the heart. Allred et aP studied 63 nonsmoking men aith heart disease tested on a treadmill, after exposure to room air or to air with carbon monoxide levels of 117 ppm or 253 ppm (resulting in COHb levels of 0.6%, 2.0%, and 3.9%, respectively). The time to angina onset decreased sigrrificantlyby 4.29E and 7.1'lE in~those exposed to low and high carbon monoxide level6, respectively, compared with those exposed to room air. Sheps et aN' studied 41 nonsmokers with coronary artery disease to assess the effects of car- bon monoxide on ventaieulararrhythmias. Patients performed a baseline bicycle ex- ercise test and were exposed to room air (1.5':6 COHb), 100 ppm csrbon monoxide (4% COHb), and 200 ppm carbon mon- oxide (6% COHb) over 3 days, followed by more exercise. Those with 6% COHb had significantly more arrhythmias dur- ing exercise than those exposed ~ to only room air. When arrhythmias were weighted by severity a dose response was observed. The aboveexperimental findings among heart patients exposed to carbon monox- ide fin4 epidemiologic supportt bom a study of men exposed to high levels of carbon monoxide while working in New York City tunnels.m These men, exposed to csrbon monoxide levels of approxi- -mately 50 ppm, suffered a 35% excess of IHD mortaliby oEanpared with the US pop- ulation, an excess that declined sharply after employment oeasation (indicating a short-term effect). They showed no ex- oess of lung cancer, and eross-seecf.ional smoking data revealed amoking habits similar to the US referent population. Hence, irxsessed'ogarette smoking was unlikely to explain the excess heart dis- ease risk. flnally, iecentevidence from two stud: ies shows thatexposure to ETS may lower level5 of high-density lipoprotein choles- terol' (HDL.C): and increase 5brinogen. Active smoking lowers HDLC.' A re cent study of never-smoking adults showed that those with ETS exposure (measured by urinary cotinine) lhad sig- niScanUy lower HDL-C levelk than ado- lescenta without ETS exposure (756lower) and signi5cant]yhigher ratios of total cho- lesterol to HDL.C.r''Another czvss-sec- tional~ study" found higher levels of 5- brinogen in nonsmokers exposed'tA ETS vs nonexposed nonsmokers. The authors suggested that higher 5brinogen levels might lead to increased thrombogenesis. While these studies were not'Jongitudinal and provided evidence of a correlation but not causation, the data suggest other mechatusms by which ETS may contrib- ute to cardiovascular disease. In summary, recently published ex- perimental and epidemiologic studies strengthen the case for a true associa• tion of ETS exposure and heart disease. RISK ASSESSMENT Risk assessmenti often means the de- velopment of a mathematical model to predict risks of disease based on a given quantified dose or exposure. Frequently, animal data are used, with doses well quantified. Assumptions are then re- quired regarding the application of an- imalidata to humans. Occasionally, ep- idemiologic (human) data are available with sufficient quantitative detail on ex- posure to permit such a risk assessment,, but more often the data on exposure are qualitative (eg, exposed vs nonexposed), so that no quantitative dose response can be estimated. This is the case for ETS. However, two other types of risk assessment remain possible for ETS," The type usedin this article is based on the epidemiologic literature and on the observed relative risk for never- smokers exposed to ETS va those non- exposed. This type of risk assessment can be used to estimate the annual num- ber of deaths due to ETS exposure among never-smokers in the United States. The excess lifetime risk for an individual never-smoker due to expo- sure to ETS' (at an unknown average dose) beyond the background risk of a never-amoker with no ETS exposure can also be calculated. The most important assumption in this type of risk assess- ment is that there indeed is a real in- crease of risk for never-sewkers exposed to ETS compared with those not ex- posed'and that this increase in risk can be estimated from the existing epide- miolbgic studies. The estimates of poP ulation and'individual risks are crude, but they provide a sense of the public health burden of heart disease due to ETS exposure. Another method of risk assessment relies on modele predicting the risk of mainstream smokers for heart disease by number of cigarettes smoked, and then estimates the equivalent number of cigarettes absorbed by nonsmokers exposed to ETS. However, this "dosi- metric" method depends on too many assumptions about what is a"cigarett.e- equivalent dose"' for those exposed to ETS, and so is not used here. Presented herein are estimates of2he number of IHD deaths due to ETS ex- posure among never-smokers, among former smokers who ~ have quit 15 or more years previously, and among fotrmer smokers who quit 5 or more years previously: Afterquitting, smokers have a sharp reduction in heart disease risk (an estimated 50% in the flrst year), followed by a long, decline in risk (a re- duction in the long-term and presum- ably atherogenic effect), until reaching approximately the same risk as never- smokers after 15 years." Hence, long- term former smokers (those who have not smoked for 15 or more years) cann reasonably be considered as never-smok- ers. Former smokers with fewer yeat< since quitting will have an increased risk from both ETS and their previous main- stream smoking, but the epidemiologic data to date do not permit a separation of these effects. Herein are calculated the ETS-attributable heart disease deaths for former smokers who quit 5 or more years previ,ously, , and it is assumed that the true number of ETS-attribut- able heart disease deaths for all former smokers lies somewhere between the attributable deaths for long-term former smokers (15 or more years)~an& short- t.erm former smokers (5 or more years). Current smokers are not considered here. Any additional IHD risk due to ETS' exposure for current smokers is likely to be small compared with the effect of their mainstream smoking. US DEATHS ATTRIBUTABLE TO IHD ANNUALLY Famulas for Attributabi. Dwths As shown in formula 1 below, the deaths attributable to IHD among never- smokers is (EF,XdJ - (EF,X I,X N;), where EF„ is the age-specific etiologic fraction, d, is the age-epecific number of IHD deaths among US never-smokers, 1; is the age-specific mortality rate from IHD among US never-smokers, and N, are the age-epecific person-years at risk I ( 1 96 JM1,0. Jarwery 1, 1992-vo1267; No 1 Fasrus Srtokmp and Heart asease-Steenlentl

e tu ient k of aw and iber ters losir Lany tte- 3~ to `the ex- ,ong 5 or long ears iave risk 'ar), ire- um- •'1ming ver- )ng- iave can nok- F laur ogic .tion ated ease 5 or med but~ -mer the -mer iort- ars). ered .e to rs is the 3- the ver- logic er of :ers. for US never-smokers. The etiologic fraction is ann epidemi- ologic measure to estimate the propor- tion of disease due to a specific expo- sure, based'on the proportion of the pop- ulation exposed~ and the relative risk due to the exposure.* In the context of passive smoking, it is defined70' in for- mula 2 below as EF= p(RR i -1)+(1-pXRRz- Il/ p(W 1)+(1-pXRR}-1)+L He.rP, p is the fraction of never-amokers exposed to ETS at' home (living with a smoker), RR, is the rate ratio for never- smokers exposed to ETS at home vs never-smokers not exposed to ETS (the triily nonexposed), RR=is the rate ratioo for never-smokers exposed to : ETS at work or in social settings but not living with a smoker vs never-smokers not ex- posed to ETS' (the truly nonexpoeed). Darfvation of. RR, and RR2 The RRs for IHD for never-smokers (1.31 for men and 1.24 for women) living with current or former smokers are from the study by Helaing etal,10'the choice of which is dictated by several reasons. The goal here is to estimate the impact of ETS in the United States (other coun- tries often have different types of to- bacco and consumption patterns). The Helsing study is the only US study of IHD deaths in a large general popula- tion of both men and women, The study results are similar to ~ the approximate results for all ETS-heart disease stud- ies combined.'" Choosing a point esti- mate of effect from a meta-analysis of all studies wou)d'yield about the same re- sult. The Helsing study considered as exposed those never-smokers living with current smokers or ex-smokers, and hence assumes that heart disease can result from~ current exposure (a short- term effect) or past exposure (a long- term effect) from ETS: This definition of exposure is the one used in most stud- ies of ETS and'heart disease. Two adjustments were made to these RRs prior to their use in formula 2above,, following the methods outlined by Wald et al." The first adjustment is for the possibility that some people (approxi- mately 796) :have been misclassified as never-smokers, but are currenU or former smokers. Such misclassification is potentially a serious problem. How- ever, for heart disease this adjustment has little effect, largely because the heart disease RR for smokers compared with that for nonsmokers is relatively low (about 1.7). As a result of the adjust- ment, the RR for never-smoking men exposed to ETS decreased from 1.31 to 1.29, while the RR of 1.24 for women decreased to 1.22. ,IMMA; Jen,ary 1, 1992-vo1287, No. 1 Tablb 2.-AftADUtabb Deatfis br NeNr.SRpkers.• ars arM Ay. arpp, r. Ui Populatlon MMn.arnoken, , 1s7s-1sao MD Ret. pe.100 000 MMwimolen Ltlobqk hacfbn PlenwiM. 2) MD D.etts Qu* bETat (FornrM. 1) Vko W 3o-4a 20 060 DD+. 1.6: 20a8 74 4.5-d4 11403101 , 32,e 2W9 7661 65. a 748 0O 221,3 . 10a6. 16 300 srn Jo-41 13 746 652 2.6 .1900 73 44N I bal 316 161 ~ 7 A900 . 1 402 s5+ 3086316 12464 .19oQt 737, totN 211026 •wiD kqrar ldrnr reM d....., o+d E7S..rw.ann.rd sw.orc ar... Th. re wrrr dw.ra d s,o tormAas nolsd. 1TA..e rvnG.q an Rr pRad +cfs af tr dm In fo a.rrrrd. h6,. ard laen mumr_ The second adjustment has a greater effect and adjuste for background ETS exposure outside the home. The refer- ent group of never-smokers living with nonsmokers was not truly nonexposed to ETS. It can be assumed that the ref- erent group had an unknown increased rate of disease (b) compared with those truly nonexposed to any ETS: Never- smokers living with smokers showed about three times the cotinine as those living with nonsmokers.' If the increased rate of disease for never-smokers living with~ smokers should be about three times (3b) thati of~ never-smokers living with nonsmokers but' exposed to ETS outside the home (b), then aecording to formula 3, Observed RR = 1+3b/l+b. Solving for b and using an observed RR of 1.29 for men, for male never-smokers living with smoking spouses, the RR for IHD death compared with that for never-smokers with no exposure to ETS (no exposure at home, work, or social settings) is 1.51, while the RR for male never-amokers not living with smoking spouses butexpOsed to background ETS at'work or in social settings is 1.17 (b -.17). The corresponding RRs for women are 1.37 and 1.12. These adjusted RRs are used in the calculation of the etiologic fraction (formula 2 above). Tha Fnactlon of Id.vsr.Smokara LMng With 8eiok.rs To calculate this fraction (the p in for- mula 2), data are taken from the never- smoking controls inn four US ease-con- tr+ril studies of lung cancer and ETS, conducted in the late 1970s and early 1980s.g'6 These studies involved 658 men and 878 women,,of whom 19% and' 5596 had spouses who were smokers or ex-smokers. Age-specific exposure prev- alence was not available from these stud- ies. Age-specific data from 778 female controls in a recent US lung cancer case- control study (Elirabeth Fontham, PhD, written communication+ July 1991) tend to confirm the overall estimate for women and show little difference in the percentage exposed after the age of 455 years. Approximately 75% of US adults are married.'6 The assumption made here is that the married and unmarried indi- viduals are alike regarding their poten- tial forr exposure to ETS. Some data justify this assumption, based on uri- nary cotinine levels of single women.= HD R.bs Attributabl9 to ETS Age-specific (at' 5-year intervals) and sex-specific IHD rates for US never- smokers were estimated using data from ~ four cohort studies: (1) the cohort study conducted by the American Cancer So- oety* (follow-up 1982' through ~ 1985) (Lawrence Garfinkel, MA, American CancerSociety; written communication„ June 1990), (2) the US veterans cohort study' (follow-up 1975 through 1980, men only) (Aaron Blair, PhD, National Cancer Institute, written communiea• tion, June 1991); (3) the Seventh-Day Adventist cohort study'* (follow up 197•i through 1982) (Paul Mills, PhD, Loma Linda University, written communica- tion, June 1991), and the Nurses Health Stud}* (follow-up 1976 through 1988, women onlyJ (Graham Colditz, MD, Nurses Health Study, written~commu- nication, July 1991). To combine these rates, unweighted averages (three stud- ies per sex) were taken of the age- and sex-specific rates, and direct standard- ization (with the 1980 US population as the standard) was used to create sumy mary rates for the three age categories in this review (<45, 45-64, 65+ years). Estimates of the age- and sex-specific number of never-smokers in the United States were obtained from the 1978 through 1980 National Health Interview Surveys' (and Robert Brackbill, PhD, National Institute for OecupaUonal Safety and Health; .vritteni communi- cation, April 1991). PassNe Smokmp ano liean a,ease-Sleena'no 97

TaDie 3-AttrlDidabk D.aens ta form.r Smahsrs' .O Rn* ftotopk. aa wid Ape ua rppt/IA/pn Of pt 106 000 Moqon qro~p Y Ibenrr L.et.n (+~) IY1+wwUnok.rs (Fan-4. 2) at.oews 00" 16 a Mon tl..n hrY1mWY rrom.n 90-" tM 50 . (7) 1.8 4544 1 126 576 p9) 52:6 65+ 709 214 (36) 9213 wo, 3DY 781003 (10) 2.9 aS6H 1 a2a 411 (26) ~ 161.7 65 ~ 2 114 61.1,(36)i 12484 .1900. .O o.ah. OW b ETEt /Fenpub.11. 2 m. 1~36 4 6C1 sa16 tew 1W V* 4 a+.o.ws Qeatq a o ra. Nrs w.raw Y 0 0 30-" 2440161 (41) 1.6 9D46 4 45-64 2 602 232 (66) ]2:e 21048 1r5 a5+ 1 33! 545 (69) 921.3 '1W8 2525 Wer, xY" 3671518 (46) 2.6 .1900 ,a 45da 47aG7e7(70) 16117 .1YOO' 144r 65+. 3 512 504 . (78) : 1246. .190p! l332 teea 12 i07 'IMD md,utesisUhert.c Mr.n Am.rse; t<d ETS,.rniibrverrtd fotrcoo. Rrorce. The Mn oorr.irs tlwaYa d the* tomwws ndso;. 1Thses rat+dsrs ars tM pWe! d wr OW Yi fr..nnd. frisf. OlO luurfi mbmn. Multiplying these rates and popula- tion estimates by the etiologic 5action+ approximately 28027 deaths among US never-smokers are estimated to have occurred annually in the 1980s as a re- sult of ETS exposure (Table 2): Similar data for former smokers who quit 15 or more years previously are sum- marized in Table 3, based on the asstunp- tion that they have approximately the same relative risk and proportions ex- posed as never-smokers. Also presented are attributable deaths under the assump- tion that former smokers who quit 5 or more years previbusly have the same iisks from passive smoking as do never-smok- ers. These last data indicate how attrib- utable deaths increase under a variety of assumptions about the return of former smokers to baseline (never-amoker) risk. The final estimates of attributable deaths are presented as a range, assuming that the true number of ETS-attribut'able IHD deaths among former smokers lies between the number derived in oonsid- eri.ng only former smokers who quitmore than 15 years previously and the number derived by considering former smokers with 5 ormore years since quitting (Table 3). Cbrnbining the above estimates, the overall estirnate ofETS-attributableheart disease deaths for never-amokers and former smokers is 3500D to 40000. IHD RtSK FOR NEVER-SMOKERS • LJWING YIHTH SMOKERS Individual excess risk of death for a never-smoker exposed to ETS can be derived using an RR estimate and con- verting rates for never-smokers to a cu- mulktive risk of IHD death by a given age, using formula 4 below, which ac- counts for competing causes of death•':. 74 Excess risk= 7, (1tR,-1)QX+1 ,-X1 exp ~-~(RR~-1)qr (1?+'q.V)~ In formula 4 excess risk refers to cumu- lative excess risk of IHD death by the age of 74 years„q, is the IHD mortality rate for nonexposed (truly nonexposed, no ETS exposure in home or elsewhere), q, is the overall'all=csuses awrtality rate for the nonexposed (here assumed to be the all-causes mortality rate for never- ernokers) (age- and sex-specific data for 1982 through ~1984 provided by Lawrence Garfinkel„ MA, American Cancer Sod- ety, written communication, June 1990); RR is the rate ratio for the exposed vs the nonexposed (assumed to be constant over age), and i and j'index ages. Background risk for never-smokers may be calculated by, omitting the terms using the RRs. An Axelson-type adjustment- was used to derive the background IHD rate for the tnilynonexposed. The Axelson technique consists of partitioning the overall IHD mortality rate for never-amokens into a weighted ~ average of the rate for those with background exposure (the rate for the truly nonexposed times the RR for never-smokers with background ETS ex- posure) and the rate for those living with smokers (the rate for the truly nonex- posed times the RR for never-smokers living with amokera)i The resulting equa- tion may then be solved for the rate of those truly nonexposed. For a female never-smoker with, no. ETS exposure (truly nonexposed), the lifetime (tu an average age of 79 years) risk of.IHD death is 4.4%. The risk for a female never-smoker exposed to back- ground ETS exposure is 4.99E,,while the risk for a female never-smoker living with a amoker is 6.1%. Corresponding results for men from age 30 to an average age of 74 years are 6.3%, 7.4%, and 9.6%. These results should be viewed as crude estimates, given the multiple as- sumptions involved. These risks apply to long-term former smokers. The estimated inceased risks of death from IHD! due t.o ETS exposure are higher than those accepted in regulat- ing environrnental'toxins. For example, environmental limits for toxins are of- ten set to limit the number of excess deaths resulting from exposure to one in Wor one in 10`,° whereas the excess risks calculated are in the range of one to three per 100: There are currently no federal regulations regarding exposure to ETS, with the exception of regula- tion for domestic airline flights. CONC LUSION A number of assumptions are in- volved in estimating the heart disease mortality due to ETS, adding an unfor- tunate level of uncertainty. The most' important assumption is that the rela- tive risks for ETS and'heart disease„de- rived from the epidemiologic evidence, are reasonably accurate. The epidemio- logic results may be questioned; given the inherent uncertainties of any epide- miologic study. Differential misclassifi- cation of ever-smokers as never- smokers and uncontrolled confounding are possible explanations for the excess risk observed in the epidemiologic stud- ies. However, neither of these likely ar counts for the observed risks. The epide- miologic data are strengthened because multiple studies now are consistent and reasonably wellidesigned.. Considerable uncertainty is involved in extrapolating from the epidemiologic data, .vhich~consider the relative risks for never-smokers living with smokers,, to estimating relative risks for those exposed to ETS (anywhere) vs, those truly not expoeed'(anywhere): This lat- ter population of the truly nonexpoeed is largely hypothetical, in that virtually' everyone is exposed to background lev- els. This extrapolation was made based on observed relative risks for ETS ex- posure at home and on urinary cotinine measurements, but is necessarily a crude estimate. If it were assumed that b4ck- ground (not frorm spouse) ~ exposure causes no increase in risk (6e, a thresh- old effect), then the number of, annual IHD-attributable deaths (due solely to !E JAM& January 1.1992-VOl'267, No. t Pssivs Smdunp and hbert Dnease-Steenlard

:be .ng -ng ige iad .as a8- Ply #h ire atr i1e, of- eas tne ess me no u'e .>a- In- A8e or- o6t .]e- Llr ,en de- -i& Iff- mg ess ud- ac- 11Be And .ed Igic sks :rs, ose OBe latr sed ally ev- sed ex- llne >lde tch- 131'e •-s2t- exposure from a spouse) dt'ops to about 15000: to 1900(). The above estimate of 35 004 to 40000 IIHD deaths attributable to ETS among never-smokers and former smokers iss based on data from the early 1980s. TThe current number of attributable deaths is likely to be lower, given the declining prevalence of smoking, declining heart disease mortality, and the increased so- eietal trend to limit exposure to ETS. One prior risk assessment of ETS and heart disease exists, using methods sim- 1'Iar to those used here. W ells''' estimated that in the United: States in 1985 tben were 32 000 heart disease deaths among nonsmokers (never-smokers and former smokers) attributable to ETS. His esti- mate is surprisingly close to the one pre- sented herein, despite the fact that he used different data and assumptions in his estimate. Wells included all former smokers in the population at risk for ETS- induced heart disease, while my discus- sion is restr'icted to never-smokers and former smokers with at least 5 years since n.r.e.eb.a 1. National Research Coancili Enti+onmenta! To-. bacco S+noke: Measuring Exposure asd A.aesinp Health Effects. Sl'aahington, DC: National Aod. emy Prens; 1986. i 1lealth Ef,l2da of Poa+ire Smoking: Msesnneni of Iwrnp Cancer in Adullb and Respirolory Duor- dns in Ckild+en. Washingtan, DC: Env'aanmental Protection Agency: May 1890. , Publiaxtioa EPA/ EO01&9Q/Df16A, review dra}L a. Hamnwnd S, Leaderer B: A diIIttcoo monitorto measure ezpasure: to passive smoiting. Enrirmt Sci T.ehwol. 1967t21:494-497. l. Rep.ce J: Exposiue asaeeeamnt in parsive smok, img.la Entironmental Tobacco Smoke: A Carapen-, dium ofliifonnation Washington: DC: IndoorAir Divioon, Envvonmentsl Protection Agenel;1989. 8. Cummings K, Idllioney M, Bliaiyava A. Ilet amsmentof cureent exposure to envaonnwi4l to bacco smoke. Arch EnldTon Health: 199D;*5:76-79. i. Haley N, Colosimo S, Axelrad C, et al: Bicl ehetmnl valldatfonE of Klf-reporGed exposure to ETS. Enriron Rea. 1989;49:127-135. 7. Wald N, Ritchie C. V.alidrtion of st>Kfies on htnt cancer in nonsmokers msrried to smokers. lawael. 1984;1:1067. 8. Huagatvel-Pureiaiiren K, Sona lrl , EnBsttom K, et al: Passive smoking at work: , biotheminl and biological measures of exposure to ETS. !wt Arch Uctup Envirnn Health: 1987;69:337345. 8. Surgeon General. TAe Health Coauqreus+ of lsuolrntary Smoking: A Report of Ge Srry.on General. Roeln+iUe„lld: US Public Healt.b Service, Dept of Health and Human,6trvioes; 1996. 10. Helaing R, Sandl'er D, Comstoek G, et aL Heart disease mortalityin notumokers bving.9tb sawk- Qs. Am J Epidynsiol. 1888;1279I6-Sp2. 11: Hde D, GOlia C, CFwpea C. Passive amolmng ad en&mespirstmy L'ealtB in a 8enerai popdation in the wert of Smtl.nd. BMJ: 19W299:423-W. IZ. Humble C, Croft J, Gerber A, et a1: Passive smoking and 2ayear cardiovascular disease mor- tatity among nonsmoking wives, Evans County:. Georgit Am J Public HeaftA: 1990 80599601. 1f:' l.ee P, CbamberLin J, Alderwn 6!. Relation- sliip of passive smoking to risk of luns cancer and otlter amoidag-aesa-iated di.emea. B* J Cancer. 1996:61.97-105. 14. He Y. Women's passive smoking and eaweary 6eart diaease. CAin J Pm, Med: 1l":19•2Z. JAMk January 1, 1992-Vot 267; No. 1 quitting. Wells e>vlrapolatzd American Cancer Society dats from the 1960s for heart dieeaee rates among never-smok- ers to estimate these ratss for the 1960s: I have used never-aznoker heart disease rates from four cohortsthdies in the 1970s and 1990s. Wells used a different prvoP dure to estimate the prevalence and ef- fect of exposure outside the home. In this article the assumption has been made that never-amokers living with cur- rent or former smokers have an increased risk of heart disease. That is, both short- term and long-term effects of ETS on the beart have been .eetmled. TTlis aestmlp- tion agrees with the epidemiologir evi- dence, which in mo6t studies has defined exposure as living with a curre=]t emoker or an ex-smoker. If one were to aesturte the effect of ETS on the heart were only ahort-ter'm (eg, via an increase in COHb), then one would have to use RRs from studies in which exposure is defined as living with a current smoker. There are two US studies using this definition of exposure.l=•ta Both ahow a higfler RR for 16. Dobaauo A. A{e:atder H, Helkr R, et aL P.a- dve amoking and the risk of heart auarlc or ooro nary death. Med4 Aiast. 1991;154:793-797:. 16. 6verdsen K, Kulkr L, Nartin:M, et aL Effede of passive smoking in the Multiple Rudc Fattar ln-. twention Tr411 Am J Epidemsol. 18B7i126:783796. 17. GariandC, Barrett-Cannne E, Suares L, et all Eifecta of passive atooi=4 an iadhemk beart dis- aase mortaiity of nonamoken. Am J Epide+nial: , 1985;121 6~5•619. 19. Hi=ayanu T. Lung eanee cm Japan: effects of autrition and pasaive smokin&: In: )ifisellild,,Car r*s P, eda:In+np Cancer: Causes and Prnention. Ne.rYork, NY: VerlagChemie lnternatiotu1.19B1. li. GLnts s, Parmky W. Passive amoldng and Aeart disease. , Circ•ulation. 1991:83:1 4z. !0. Sargeon General. Tk H'aolth' Cowr.7iienees of Smoking: Ca+diooa+erlar Dis.aae: A Report of tAe Stirpeon General: Roekville, Md: US PublibHeaAii Service, Dept of Health and H unun Services;19B9.' 21. Marchand L, Wilken. L, Hankin J, et al. Di- etary pattertr affemale aon®oken.it6 and.rt2, out exposure to envirvunental tobaaoo smoke. Can Causes Cont.oL 1991;2:11-16. 22. Sidney S,,Caan B,,Friedman G. Dietary intake of carotene in natumoker+ with and: vithout pa sfve smoking at home. Am J Epidemiol. 1989;129 1805-1309. 23. Allred E, Bk»d ier E; Chaitman B, et al: Short- term effeete of carbon monoxide exposure on the exercise performance of subjeeta.itb eoronary ar tery disease. N Engl J' Med: 1989;321i1426-1475., 24. 6lieps D;,Herbat]L, Hmdertita A, et al. Pro duttion of rrrbTthmias by aievatsd COHb in pa- tients with eorvaary artery dis,e.se. Ann lwlerw Mad: 1890;113B43.861. !6. 8tenn F, H.lpcain W, Haebutt8 R. Ha.rt dis- aa.e mortality among bcid8e and tunnel officen exposed to carbon monoride. Am J Epidemiot. 19B8;128:1296-1288. 76. Surgeon Getera1. TAr Health Bewdus of Sieot'- iwg Cessation: A Repcrrt of We Svrp.ai , Gewe+ul. Roekvi[le, Idds US Public Health Serviae, Dept of Healtb and Human Bervices; 1i80. 27. Feldman J; 8henker I, Etsel R, at al!,Pasdve smoking aher+ Jfpid pro6les in .do7ascenta. P.di- at+tin. 19812354864:. !& Rep.oe J. Lowery A. Risk aase.sment metlr odoio8ia for parive.maionQ-itrdtmed lan( eanar. , exposed vs nonezpased; (approxilnately 1.6) ~ than the one assumed here (1.2' to 1.3)! However, these studies have limi- tations for use in alcnlating attributable risk. Olleu included only wanlen and the outcome was based on all cireulaiory dis- ease (including stroke), not IHD: The other'6 included only men at high risk of heart disease. In conclusion, assuming the epidemi- oiogic evidence is valid and assuming our estimate of 35 000 to 40 000 annual excess heart disease deaths among never-slnoker8 and long-term former smokers due to passive smoking is cor- rect, then heart disease mortali ty is con, tributing the btilk of the public health burden imposed by passive smoking. Lung cancer,, the previous main culprit, has been estimated to cause approxi- mYtely 3000 excess deaths per year among never-amokers. Coeuoenta on.uiier versions of the taanusaipt were kindly provided by DrsJudson Welis•,Peter Gann„Ls.iie Stayner, Lrwic Kulkr, and TTomas Robuu. Risk Aaal.' 1890;10'Ti37. 2!. Hleinbaum K• Kupper L, liorganatern H. Ey ideneiatopic Raeanl, San Joae. Calit: Lifetime I:aarninQ PuElxstlont; 1983: , 10. Vainio H! Partarien T. Population bardeni of lung encer due to enviiotvnental tobacco smoke. Mvta7: Re+. 11i8?Z221:1II7-140. 81. W ald N. NbrKhahal K. Tbompaon S, eta1. Does beeatliitg otJw people's tobacco smoke cause luate anoer', BIfJ. I9B6;,253:1217-17?2. 32. Kabat G. Wynder E. Lung cancer in nooatnok- era. , Cancer. 18Bi:53:1214-1221. 33. Humble C, SamevJ, Pathak D. Yarri.ge to a sa+oker and hms esrner risk. Am J Public H.aitk. 1997t77:596-602: 34., GarlSnkelL, Auerbach 0, Joubert L. lnvolun- tary smoking and lung cancer. a casr-eontrolstudy. J Natl Cancer ln+t. 1985;76:463a69. aS. Coerea P: Pickle L, Fontlum E„et a1: Passive ataoking and lung arxer. Lancrt: 198:t2:59&-597. !6. Ste!lman S. Gartfnkel L. Smoking Aabit, and tar levels in a new Ameriran Cancer Society pro- spective study of 1.2 million men and .omen. J' Natl Canexr 1*+t. 19B6i76:1067-105i: $7. RoBot E, Murray J. Smoking and eauses of death among US veter.ns:16 yeus of observation. J Na/f ~ Cancer Iws1. 19BU95:21Ii222: 88:, Beeson W, Mills P. Roland P, et a). Chronic duea.e among Seventh-Day Adventiats, a low-riak group. Cawcer. 1989,fi1:570581. 39. Caklit: G. Willett A, 8t.mpfer )d; et al. A peoapective study of p.rental history of m,voordial i mtuction I and aoronary heart disease in .amen:. A.n J' Epiderniod. 1866;129'4&58. 40. Br.ckbill R• Fraaer T, ShillinQ S: Bnrolmg ebararteristirs of US sar3een, 197S1fIB0.. A1n: J lwd Md. 1988;1'9i5-11. 41. Gail 16. Measuring the benefSta of tedue:d ex- ponue to envvonmcntal nrdratrna. J CAron Du- I9762b:1ab147. 42. Axebon 0. Aspects aa eanfoundirK in aacupe tional epidemioloLv. Saowd J WorkEe eiron H.a1 tX. 1478:4:8.4-89. • 43. Rodericks J. Brett S• Weenn:G. 61wni6nnt risk d.oaions in federal regulYtory.jenos. Rqrl' Tou•,eol PAa.rwcol. 1987:7:80732Q: 44. WelL A. An estimate of s,duh mortality trvm passive saalriiK. Ewmron,lwt. 1988;14:7A1L265. Passivs Srttokirq vK! 1•1sM Deease-StesN.ntl 99