Tobacco Documents Online

), smokrng•RV/ared Dls. J 993 ~ afl); 27-?6 ENVIRONMENTAL TOBACCO SMOKE AND CORONARY HEART DISEASE A.K. Arr»itage Consultant Pharmacologist and'Toxicologisz Knaresboroug>7, UK Abstract The possitiilitythat exposure taenMronmental tobacra smoke (ETS) may increase the risk of non-smokers devei'oping coronary heart disease (CHD) is a matter of debate. The results of published epidemiology studies harre been considerrd according to the dassicai criteria for jixigment of causality. Tweiv* epiderrnobgicai studies have been published onithe alleged relation.ship between exposzre to ETS and ttv risk of CHD ih6 non-smokers. The relative risks generaiyfall between 1 and'2, which represent3 a 'weak •assrxiation'. Although the reported association shows some signs of oons isten cy; it is not strong, nor it is specific. The ladc of consistency between dosimetry and epiderniology casts doubt on the biobgical ptausibiiity;'further daims of a dosdresporrse reiatiorship, which are based entire#y on reported cigarette usage, are not wholy convincing. On current euidence a causal re{ationship betweesti exposure to ETS and the developrnent of CHD has not been proved. Key vNords: Environmental toDacco smoke (ETS), coronary heart disease (CHD), assocfation and'causality, dcsii'netty, pubication bias, biol'ogicai plaus+bi(ity Introduction. The debate regard'ing the association between smoking and cardiovascular dis- eases, especially coronary heart disease (CHD) has, during, recent years, led to consideration of the possibiiitythat exposure of non-smokers to environmental tobacco smoke (ETS) might increase the risk of such disease. The uncertainty aboutthe existence of such an association is evident from the pronouncements of groups of scientists who carefully and aitically reviewed the relevant data available at the time of their review. 1986: US Surgeon General (1J - "Further studies on the relationship between involuntary smoking and cardiovascular disease are needed in order to determine whether involuntary smoking increases the nsk of cardiovascular disease". This viewwas based on four studies which had' been published at that time. 1986: National Rese•arch Council [2 J-"1Nith respect to chronic cardiovascular morbidity andmortality; although biologicaliyplausible, there is no evidence of.statisticaliy significant effects due to ETS exposure, apart from the study of Hirayama in Japan" (four studies considered, three of which were also considered by the US Surgeon General). 1990: Environmental Tobacco Smoke - Proceedings of the Intemational Sympcuiurn atMcGill Univefsiry,Montrea]{3)-Dr Wexier and discussants unanimously concluded that the data from seven studies would not pro- vide any basis for altering the Surgeon,Gen- erai's and the National Research Council~s conclusions concerning ETS and cardio- vascular disease. Address correspondence to Dr,a.K. Arrr,itage. Sytamore Lod9e. Abbey Road, KnaresCorough, Norrh Yorkshire HGS 8xY, uK. (Accepre0 for publicapon 30.7.92.) 27

1990: UnT*+ ksbetween passive smoking and' dsease, a best-evidence synthesis [4) - This report of an 11-mernber working group on passive smoking stated in 1990 that "after an initial review of'! the literature, we deter- mined that insuff icientstudies of high o,uality had'been done in the area of cardiovascular disease to warrant comment in our review at the present tlme". Most of the studies upon which these opinions were based were not designed' specifically to investigate the effects of ETS exposure on CHD. According to Weetman & Munby (5) they were later adapted to this purpose after the initial claims that ETS affected liealth adversely had appeared in, 1981-1982. Thus, neither of the two~large prospective studies [6,7], was conducted~ according to accepte6 principles, i.e. systematic coliection ofi. risk factor data at intervals and corrnpletefoliow-up of all deaths. With the exception of the recent study of Oobson and associates [8), it is notable that all the other published~ studies involved'only very, small numbers of', deaths [9]. Epidemiology data and the question of causality Twelve epidemiolbgy studies have now been published ([I6-8,10-20] - see Table 1), on which to base a judgement but, few details were provided for two of the case-control studies and the studies of Martin er ai: [11 ]„ Palmer er al: [13] and But3er [19] have only been published as abstracts; full papers are still awaited. Some of the data in the Table are based on figures quoted by Lee [21 ] and Glantz & Parmley [221. In attempting to interpret the results of these studies it is appropriate to recalltheelernents of evidence laid down by the United States Surgeon General (23) and the late Sir Austin Bradford Hill (!24) that need to be considered before a reasonable inference of causation forstudies that purport to show association can be made. Hill's elements were as follows: - aniassociation should be strong - it should be consistent across a variety of subjects and' circumstances - it should be specific - exposure must precede the observed outcome - there should be a dose-response relationship, - the association should be biologically plausible = evidence from all Vvan't scientific disciplines should be coherent - experimental evidence should be sought whenever possible - anaiogy to similar cause and effect relationships should be considered. Hill's criteria did not actually include - "'sourcesof bias mustbe ezcluded"-though it is implicit in the condition - "an association should be strong". Sources of bias and confounders, which feature so prominentiy in all papers concerned with the association of ETS and the d'evelbprnent of disease are, therefore, logically considered understrength, of association. The United States Surgeon General [23] wisely commented in 1964'that "Statistical methods cannot establish, proof of a causal relationship in an association, The causal significance of an association is a matter of judgementwhich goes beyond anystatement of statistical probability". It is regrettable, however, that three decades on judgements tend'to get doud'ed to fit in withibeliefs. Each of Hill's elements will be considered in turn, Strength of association, The possibility of an association between CHD and alleged exposure to ETS has arisen from observational studies (prospective and retrospective) of groups considered to, be either exposed or not exposed to ETS. Risk of' death from CHD in the exposed group relative to the unexposed group is calcuiated' by application of appropriate statistical techniques. There are considerable difficulties in conductingland interpreting epidemiology studies when the 'increased risk' is srnalll Indeed, it has been, stated (25-28] that epidemiology cannot usually reliably predict relative risks of less than two. 28

Tabte 1. Epirbemiotogical studies of erwironmentat totweco smok.e (ETS) and coronary heart disease (CHD). CHD deaths or caaes Relative 95% oonf. Reference Location Sex Unexposed Exposed risk Grnits Factors adjusted for Came control stud}es: teeetaL/101 UK M 26 15 1.34 0.64-2.80 None F 22 55 .97 0.56-1.69 Martinetal (111 US F 23' 2.6 1.2-5.7 Not known He(121 China ' F 9 25 1.5 Unknown Age,, race, residence, oecrrpatoon, almer et a! .(13( SA F ' .2 nknown hypcrtension, family history of hypertension or CHII, alcohol, exercise, hypercholesterolaemia Not known Dobson et a1. (8) Australia M 161 22 0.97 0.50-1.86 Age and history of myocardial F 117 43 2.46 1.47-4.t3 infarction . I'yospective sturk--- ' Hole et aL (151 UK M+ 30 54 2.01 1.21-3.35 Age, sex, class, blood pressure, Svendsen et al. 1161 US F M 8 5 2.23 0.72-6.92 cholesterol, body mass Age, blood pressure, cholesterol, "' Hclsing et a!. (61 US M 248 122 1.31 1.05-1.64 weight, education, drinks Ac3e, marital status, schooling, F 437 551 1.19 1.04-1 _36 housing llirayarna (71 Japan F 118 376 1.15 0.94-1.42 Age Gailand ef a!. (171 US F 2 17 2.7 0.90-13.6 Age Humble ef al. (18( US F 76' 5 9 1. O.g9-2.57 Age, cholesterol, fiullcr (19( US F 64' _ _ 1.4 0.50-3.80 blood pressure, body mass Age • I lole et at is an update of Gilfis et al. (141 but is not a separate study '• I khirm data extracted from lalcr paper by Sandler et al. 1201 '= total number of cases TsazTsCzoz

The 12 studies provided' 15 different estimates of relative risk for either men or women; and 95% confidence intervals are available for all but tvvo of these. Table 1 shows that 13 out of' 15 Indicate a relative risk of > 1 but' eight of these have a lower confid'ence limit of <1, indicative of a non-significant effect. Although the studies suggest an association; because most of the claimed relative risks lie between one and' two, Hill's criterion of strength isnot satisfied. The studies by Butler (19J, Dobson etal. [8) and Svendsen et aL [16) give some d'at;a on ETS exposure at home, work, leisure and travel situations. None of these supplementary results was statistically significant. Meta,analysis Recently the results of all studies have been combined using the technique of ineta- analiysrs (29;22). Wells (29); using~ studies available in 1987, computed a pooled relative risk of 1.3 (95% confidence interval, 1.1~-1.6) for rTmen and 1.2 (95% confidence interval, 1.1-1.4) for women. Glanu & Parmley [22]'reported a relative risk of 1,3 (95% confidence interval, 1.2-1.4) for men and women combined'. Meta-analysis is a mathematicaltechnique primarily developed for handling data from multi-centre randomised clinical trial+s where thesame protocolforthe selectionof patiepts, the selection of control subjects, dosing and observation is followed ih each of the collaborating centres. Whether such techniques are scientifically appropriate for epidemiologlcall studies conducted in different countries by investigators using, widely different methods and' criteria is doubtfwl'(30r-34j. Meta-analysis does not increase the strength of the association, although, by reason of narrowing the 95% confidence intervals it may, produce a mathematically, though notnecessarilybiologically, significant relative risk. The Glantz and Parmiey paper (22J has recently aroused criticism and debate (,35-3'8) on a number of iss.jes including the mathematicall model used by Weils [29). Misclassification bias According to Lee (39,40j most, if not all, of, the apparently higher risk of lung cancer among non-smoking women married to smokers compared to non-smokingwomen married torion-smokers can be explained by a bias introduced when, smoking (or ex- smoking) women, claimi untruthfully to be non-smokers. Such rnisclassification leads not only to imprecision but to bias, because smokers tend to marry, smokers and " non-smokers tend to marry non-smokers more than would be expected by chance. Misclassification bias of this kind affects not only the interpretation of data relevant to lung cancer risk from ETS but also to some extent that relevant to CHD risk [21). The matter of smoking~ habit concordance has been fully discussed by Lee ('211. Publication bias Lee [21 ] has shown that there is a tendency for the studies with small numbers of deaths or cases to have the largest reiative risk estimates. For example, Table 1' shows that the highest relative risk reported, i.e. 2.7 by Garland et al. (17) was based on only 1.9 deaths, whereas one of the lowest relative risks (1.19)reported by Hel;;ing (6) was based on, 1,358 deaths in the largest of' the 12 studies. Lee (211 suggests that this is a manifestation of publication, bias. He also points outthat the American Cancer Society has relevant data on many thousands of deaths from heart disease among women, who have never smoked, but has not, so far reported results which, according to Lee, suggests that no relattonship with ETS was found. The possibility of publication bias has been strengthened by the paper of Easterbrook et al. [411] which states, "we have confirmed the presence ofia systematic selection bias in the publication process according to study results. Studies with a statistically significant result for the main outcome of interest were more likely to be submitted for publication and more likely to be published than studies with null results, after adjustment for confounding factors". Thus, conclusions of ineta~analyses based only on published work may produce a misleading high estimate of relative risk. 30

other potential confounders Neither of the two large studies 16,7) recorded detaiis of major nsk factors for CHD such as blood pressure, blood cholesterol levels, obesity, etc, an omission criticised bya number of reviewers [3, 5,21). inthe studies by Helsingl Sandler, [i6,20) andby Hole (15) the index of ETS exposure was based on living with a smoker, but no adjustrnent was made forthe number of people in the household. Since household size may correlate with various facets of disease, adjustment, should'have been made for it in the analysis. Dobson etaG [SJ collected data on smoking behaviour for cases and controls by different methods and showed that, for the controls, smoking frequency varied according to the location of data coliectiom This provides a further important example of potential bias. Consistency of associaticn Stati'stically, assuming no systematic bias, it would appear unlikely that 13 out of 15 estimates of relative risk> 1 could have ansen by chance. Although, some element of consistency amongst the published studies is indicated the possibility of publication bias may have accentuated the consistency of the associatiom Specificity of association There are two distinct aspects of specificity- one relating to the exposure and one to the disease. CHD is a common cause of death in non-smokers, which is hardiysurprisingwhen almost 300 so-called risk factors have been described [42)'and the literature on such risk factors "seems to expand at almost exponential rates" [43). For this reason, the cause of CHD Is frequently stated to be multifactorial in origin, It has been said that "the phrase is a synonym for'unknown' and thus a euphemism for ignorance" [44]. CHD is of course not the only disease to be linked' with ETS.. There is, therefore, an absence of specificity in the association, as Wexler [3) also concluded. Temporaliry Hill! (24) stressed the need to consider the temporal relationship of any association, particularly for diseases of'slow development~. There are no data available on which to address the importance of the temporality of the association between ETS and CHD. In, view of the multirfactorial nature of the disease, which inevitably complicates its precise development, it is doubtful if there ever will be any unequivocal data in this context. Doselresponse relauonship Exposure ETS isa mixture of sid'estream tobacco smoke (the smoke that originates from the smouldering end of a cigarette, cigar or a pipe between puffs), waste mainstream smoke (i.e. that expelled prior to inhalation)) and exhaled mainstream smoke diluted;with variable but much largervolbmes ofambient air. Additional physicalland chemical changes occurasthe mixture ages. The concentration and nature of ETS to which an individual is exposed depends, among other things, on: - type and'number of cigarettes smoked ha given time - volume of room - ventilation rate - proximity of burning cigarette. The effective daily dose for any exposed individual depends upon the concentration in various environments and the time spentt in these environments, which highlights the diffiCulty, of assessing, exposure under 'real life' conditions (45). Additionai(y, exposure rnust relate to a time-span that is appropriate to the development of the disease. ETS isa complexmix of manycomponents and, because no known substance is rep- resentative of all of these, it is not possible to monitor ETS exposure meaningfully [5]. Nicotine comes closest to fulfilling some of the requirements for a suitable marker [46). Kirketal, (47]Imeasured atmospheric nicotine, carbon monoxide and particulate matter inn nearty3,000locations over 30 minute periods in travel„work, home and leisure situations. Their data show that levels vary greatly both within, and between,, situations. They illustrate how inadequate and misleading is 31

an assessment of ETS exposure in epidemiology studies based merely on reported smoking habits of couples living together, determined from questionnaires. The reiiabilityof questionnaires represents a weakness of all experimentall designs, particularly when smoking spouses them- selves were not always questioned (43). Ad- dirionally, some are untruthfull abouttheir smoking status and misclassification of non-smokers who are really smokers invariably results (39,40). In the large Helsing study 16), the only smoking data that were collectedion every person was in 1963, and no attempt was made tosee whether changes in smoking habits occurred during the study: Dosimetry Dosimetry assessment involves the measurement of a substance in biological fluids. If ETS has any affect whatsoever on CHD then the only reasonable way that this could come about is by absorption of a component or components of ETS into the systemic circulation. No such measurements have been incorporated in any epiderni- ological study: As with markers of exposure, there is no single substance that is reliably representative of ETS. Both nicotine and carbon monoxide have been implicated in possible mechanisms by which active cigarette smoking may accelerate the development of CHD (49)'. Nicotine and its metabolite cotinine, which has a haif-life of 15-20 hours, have been measured inc (a) ETS exposed non-smokers, (b) non-smokers not exposed to ETS, (c) active smokers (150). Very small amounts of cotinine are found in the plasma, saliva and urine of nonrsrnokers. Whilst non-smokers allegedly exposed to ETS are reported to have higher cotinine levels than non~srnokers not exposed, these levels are minute compared with those found in active smokers (about 1/300) and variation between subjects is large (up to 10-fold). Ihinon-smokers, endogenous production of carbon monoxide results in low blood levels of carboxyhaemoglobin (COHb) in the range 0.5-1.5% (51). Small increases of COHb due to ETS exposure may be indistinguishable from those due to en- dogenous production and non-tobacco related sources [1). Clearly, the amounts of nicotine a nd'carbon monoxide absorbedifrorn ETS are small and it is unlikely that either compound plays any role in the alleged increased relative risk for CHD of non-smokers exposed to ETS. It would be surprising, if any other smoke component was absorbed in significant amounts from ETS, Dose/response Some authors (7,12,15,16) of the studies listed in Table 1 claim that their data demonstrate a dose-response relationship between ETS exposure (based on the number of cigarettes smoked by the active smoking partner) an6death from CHD. Ir,iview of the imprecision of such dosage assessment, these dairns should'be viewed with caution (52). Biological plausibility Although ETS is a different entity, from mainstream smoke, it contains many of, the same components, and attempts have been made to equate ETS exposure with active smoking in terms of 'cigarette equivalents'. On such exposure/dosimetry considerations, the relative risks of more than two reported in some ETS studies [8,11,15,16,17) are irnplausibly high, set against the reported relative risk for active smoking of 1.9 in males and 1.8 in females [53). Professor Wald con- cedes thatthe association seems "surprisingly li3rge" and requires explanation [54] and has commented that the risk of 1.3 given by Glantz & Parmley (22) is "too high to be biologically plausible" [55). ThaUoneauthor (17)can incorrectiy report a relative risk of 14.9 (the majorfinding of the study) and then subsequently 'correct' it to 2.7 whilst maintaining that this "does not affect the condusions" (56), casts grave doubt on the conduct of the whole study. Likewise, when, Hirayama [57]'reported on the first 14 years of his prospective study, there was no mention of a higher mortality rate from CHD of non-smoking women married to smokers. Armitage (45j pointed out that "it is difficult to believe that a 32

J previously unsuspected risk coul'd become apparent merely as a consequence of three more years of follow-up". Doll (58J has commented that if you find something unexpected but of social significance you have a responsibility to be sure that you are right before you publicise your results to the world. In, the same paper he also offered additional advice that ". .. as a research worker, you always ought to try to disprove your own findings". Pipe smoking Pipe smokers inhale tobacco smoke bothi actively (to an extent) and passively. They surround themselves in dense clouds of tobacco smoke and are regularly exposed to high concentrations of ETS, yet experience "little if anyexcess risk of'oeathfrorn coronary heart disease" 159J. 1t is unlikeliy that the chemical composition of pipe sidestream, smoke wouid'differ appreciably from that of cigarette sidestream srrmoke. Therefore, these observations provide suggestive evidence against a role of ETS as a cause of death ifrom i CHD. Coherence This criterion addresses the effects of ETS on biological systems which have beeniclaimed to support, a causative hypothesis. Platelet function Glantz & Parmley (22) claimed that ETS adversely affects platelet function in a way' that increases the risk of heart disease. This condusion is not warranted considering that the study cited [60] measured a response to an ETS insultin non-smokers keptIogetherin an 1ft= room for 20 minutes, while testers smoked "30 heavy brand" cigarettes. The relevance of these 'artificial"experiments to C HD problems is questionable. It is not known whether, unden normal conditions of ETS exposure encountered in real-life situations, the platelet function of non-smokers exposed to ETS differs from the platelet function of non-smokers not so exposed. Fibrinogen levels and thrombogenesis Dobson,and associates (8) reported plasma fibrinogen concentrations in the controlF subjects of their case-control study in ani attempt to provide mechanistic supoort to their conclusion that exposure to ETS in the home increases the risk of fatal and'nonrfatall heart attack. The authors state that "people exposed to passive smoking had higher levels than those not exposed~ (except for passive smoking at home for wornenY". However, the results were not significant and the workplace meansfor non-smoking men and women (exposed vs. non-exposed) were v'rrtuallythe same, Effects of ETS/carbon monoxide on exercise tolerance Healthy people can inhale enough carbon monoxide to reduce the oxygen carrying capacity of the blood by 10% or more,, without ill,effect (61 ]. However, people with reduced circulatory efficiency (e,g. because of existing heart disease) rnay be temporarily compromised by exposure to carbon monoxide; their exercise tolerance may be reduced and they may develop anginaf pain sooner than in the absence of carbon mon- oxide. Aronow (62] reported a reduction in the duration of exercise until' the onset of, pain in 10 patients exposed, to ETS, which resulted in mean COHb levels of 1.77% and concluded that exposure to ETS was harmful to patients with angina pectoris. Aronow's experiments were not, conducted on!a strict double-blind basis and the measured end-point was a subjective one. Muchof his work has now been generally discredited (1,63,64]. Sheps etal. [,65] have since shown that blood levels as high as 4% COHb have no significantieffect on a range of parameters examined in: patients with ischaemic heart, disease. This well-controlled study casts considerable doubt on the suggestion that elevation of carbon monoxide levels in the blood resulting from ETS exposure is demonstrably harmful. Experirnen r Hill [24] asked whether the frequency of the associated event (CHD, death) was affected by preventative action (modifying ETS exposure): No data relevant to this question are currently available. Whether it would be possible to design a sufficiently sensitive 33

study to demonstrate objectivelythevalue of preventative action in any specific situation seems questionable. Analogy ETS is one of the many problematical materials on which toxicologists have had to make safety evaluatiornjudgements because of its variable composition,and complexphysico- chernical'properties, The obvious analogy is vvithi mainstream tobacco smoke to which reference has been made in the section on biological plausibility, Great caution, how- ever, needs to be exercised in making judge- ments in this context (52). Conclusion. The case for ETS exposure causing CHD is wholly unconvincing because almost all' of the accepted 'causationa' criteria remain unsatisfied. Of particular concern is the weakness of the association, the likelihood of the existence of publication bias resulting in an overestimate of a very lbwrelative risk, the lack of biological plausibility and the anecdotall nature of dosimetry assessment. At the present time, therefore, one is not able to conclude categorically that ETS is, orr is not:, harmful in a cardiovascular context. It is debatable whether the conduct of further epidemiological studies, frequently recorn- mended; is practical and just'rfied. Small studies are a waste of: time and money because at best~ they can only detect large risks as significant. Of course it is LheoretJcally possible to conduct a prospective study, as envisaged by Wexier [3 ], which wouldtontrol , for all confounding variables an6 involve a sufficient number of subjects to provide reasonable statistical power. However, even large studies cannot distinguish with any certainty between a very low risk and no risk. Furthermore, without meaningful measure- ment of prolonged ETS exposure, results mightstill be inconclusive regarding the ques- tion of causatiorn, in the meantime, it is wrong that an ETS/CHD health scare has been blown up out of all proportion, in the last fewyears by a passionate anti-ETS health lobby. Interpretive opinions are not proven facts; they must be challenged and' a more balanced point of, view presented to the general public.. References 1. U5 Surgeon General. The Health Consequences of Involuntary Srnoking; US Departrt,entof Health and Human Services, Washington DC, 1986; 2. National. Research Council. Environmental tobacco smoke: Measuring exposures and assessing health effects. National Academy Press, Washington DC, 1986. 3. Wexler, L.M.; Environmental tobacco smoke and cardiovascular disease: A critique of the epidemiological Iiterature and recommendations for future research. In: EnvironmentallTobacco Smoke: Proceedings of the Intemationall Symposiurn at McGill University lEds. D,J: Ecobichon, J.M. Wu). Lexington Books, Lexington Ky, 1990: pp139-152. 4. Spitzer, W:O:, Lawrence, V., Dales R. et al.; Links between passive smoking and disease: A best-evidence synthesis. Clin. Invest. Med. 1990: 13; 17-42. 5. Weetman, D.F., Munby, J.; Environmental, tobacco smoke (ETS),and cardiovascular disease. In: indoor Air Quality and Ventilation (EdS. F. Lunau, G.L. Reynolds): Selper, London, 1990: pp211-215. 6. Helsing, K.J., Sandler, D.P., Comstock, G,W., Chee, E.; Heart disease mortality ininon-smokers living with smokers. Am. J, EpidQmiol! 1988: 127; 915-922. 7. Hirayarrna„ T.; Lung cancer in Japan: Effects of nutrition and passive smoking. ln: Lung cancer: Causes and Prevention (Eds. M; Mizell, P. Correa) Verlag Chemie InternationaU New York. 1984: pp175-195. 8. Dobson, A.J., Alexander„H:M., Heller, R,F., Lioyd; D.M.; Passivesmoking and the risk of,heart attack or, coronary death. Med. J. Australia. 1991: 154; 793-797. 9. Lee, P.N1; Weakness in recent risk assessments of environmentalltobacco smoke. Environmental T!echnology;,19911: 12: 193-208. 10. Lee, P.N., Chamberlain, J., Alderson~ M.R.; Relationship of passive smoking to risk of lung cancer and other smoking-associated diseases. Br. J. Cancer. 1986: 54; 97-105. 34

11', Martin, M.J.,, Svendsen, K.H., KuAer, L.H.;, Non-smoking men married to smokers are similar to non-smoking men married to non-smokers. Paper presented'at 7th Annual Meet ng, Society for Behavioral Medicine, 1986. 12. He, Y:; Women's passive smoking and coronary heart disease. J. Tradit. Chin. Med. 1989: 23(1); 19-22. 13. Palmer, J.. Rosenberg, L., Shapiro; S.; Passive smokingland myocardial infarct'ion (abstract)', CVD Epidemiology Newsletter. 1988: 43; 29. 14. Gillis, C.R,, Hole, D.J,, Hawthorne, V,M„ Boyle;,P,; The effect ofI environmental tobacco smoke in two urban communitles in the West of Scotland! Eur. J! Res. Dis. 1984: 65(Suppl. 133); 1i21-126. 15, Hole, D-J„ Gillis, C.R., Chopra, C., Hawthome, V.M.; Passive smoking and cardiorespiratory health in a general population in the West of Scotland. Br. Med. J. 1989: 299; 423-427. - 16. Svendsen, K.H., Kuller, L.H., M'artin~ MJ., Ockene, JLK.; Effects of passive smoking in the Muhiple Risk Factor Intervention Trial. Am, J. Epidemiol. 1987: 126; 783-795: 17. Garland; C., Barrett-Connor, E., Suarez, L. et al.; Effects of pas;sive smoking on ischemic heart disease mortality of non-smokers. Am, J! Epidemiol. 1985: 121; 645-650. 18. Humble, C., Croft, J., Gerber, A. er al:; Passive smoking and; 20-year cardiovascular disease mortality among non-smoking wives, Evans County, Georgia. Am. J. Public Health. 1990: 80; 599-601. 19. Butler, T.; The relationship of passive smoking to various health outcomes among Seventh Day Adventists in Caiifornia (AbstraR). Seventh World Conference on Tobacco and'Health. 1990! 3116, 20. Sarndler, D.P., Comstock, G.W:, Helsing, KJ., Shore, D.L.; Deaths from all causes in non-smokers who lived with smokers. Am J. Public Health; 1989: 79; 163-167. 21. Lee, P.N:; Environmental Tobacco Smoke and'Mortality: Karger, Basel. 1992. 22. Glantz, S.A., Parmley, W.W.; Passive smoking and heart disease - Epidemioibgy„physiolbgy an6 biochemistry. Circulation 1991,: 83; 1-12. 23', US Surgeon General. Smoking and Health. US Department of Health, Education and Welfare. Public Health Service PaDer No. 1103, 1964. 24. Hill, A.B.; The environment and disease: Association or causation? Proc. R. Soc. Med. 1965: 58; 295-300. 25; Doll, R., Peto, R.; The causes of cancer: Quantitative estimates of avoidable risks of cancer in the United States today. J. Natl, Cancer Institute. 19811: 66;,1192-1305, 26: Alderson, M.R,;,An Introduction to Epidemiology. Macmillan, London. 1983. 27. Roe, F.J.C.; Panel discussion on lung and' other cancers. In: Environmental Tobacco Smoke: Proceedings of the International Symposium at McGill University, (Eds. D.J, Ecobichon, J.M. Wu), Lexington Books, LexingtoniKy, 1990: pp128-132. 28. Wynder, E.L.; Epidemiological issues in weak assocdations. Int. J. Epid'emiol. 1990::19(3, Suppll 1); 55-57. 29. Wells, AJ.; An estimate of adult mortality in the United States from passive smoking. Environment International, 1988: 14; 249-265. 30! Flpiss, J.L., Gross. A.1.; Meta-analysis in, epidemiology, with special reference to studies of the association between exposure to environmental tobacco smoke and lung cancer: A critique, J: Clin, Epidemiol. 1991:44; 127-139. 31. Greenland, S.; Quantitative methods inithe review of epidemiologic literature. Epidemiol. Rev: 1987: 9; 140. 32. Thacker, S.B.; Meta-analysis: A qwantitative approach to research integration. JAMA 1988: 259;. 1685-1689. 33. Letzel; H., Blumner, E., Uberla, K.; Meta-analyses on passive smoking and lung cancer. Effects of study selection and misclassification of' exposure. Environ; Tech. Lett. 1988; 9; 491-500. 34. Spitzer, W.O.; Meta-meta-analysis: Unanswered' questions about aggregating data. J. Cfin. Epidemiol, 1991 !: 44; 103-107: 'N O W:S.; Environmental tobacco smoke and cardiovascular disease (Letter)1 Circulation 35. Simmons ~ , 1991: 84; 956. Ul 36. Decker, W.J,; Environmental tobacco smoke and cardiovascular disease (Letter). Circulation 19911: ~ 84; 956-957: 37. Nolcomb, L.C.; Environmentall tobacco smoke and cardiovascular diseasQ (Letter)i Circulation 0 1991: 84;, 957-958. 38. Glantz, S.A., ParmiEy, W.W.; Environmental tobacco smoke and cardiovascular disease (Letter); ~ Circulation 1991 84; 958-959: 35

39: Lee, P,N!; Lung cancer and passive smoking; Association or artefact due to misctassification of smoking habits? Toxicol. Lett. 1987: 35; 157-162. 40. Lee, P.N!; Misclassification of Smoking Habits and Passive Smoking. A' Review of the Evidence. ' Springer-Verlag, Heidelberg. 1988. 41, Easterbrook, P.1i, Berlin, J.A., Gopalan, R., Matthews, D.R.; Publication bias in cllnicall research. Lancet. 1991: 337; 867-872. 42. Hopkins, P.N:, Williams, R.R.; A survey of 246 suggested coronary risk factors. Atherosclerosis. 1981: 40; 1-52. 43, Hopkins, P:N., Williams, R.R:; Ident#ication and relative weight of cardiovascular risk factors. Cardiol. Ciin. 1986: 4; 3-31. 44. Skrabanek, P., McCormick, J.; Facts and'Fallacies in Medicine; ThQ Terragon Press, Glasgow. 1989: p92. 45. Armitage, A.K.; Environmental tobacco smoke and coronary heart disease. Ih: Other People's Tobacco Smoke (Ed.: A.K. Armitage). Galen Press, Beverley, UK. 1991: pp109-116.. 46. Roe;, F.J.C., Perry, R.; The physical and chemical characteristics of environmental tobacco smoke with special reference to exposure dase. In: Other People's Tobacco Smoke, (Ed, A.K. Arrnitage)! Galen Press, Beverley, UK. 1991: pp21-39. 47. Kirk, P.W,, Hunter, M,, Baek, S.O. etaL; Environmental tobacco smoke in indoor, air, ln: Indoor and Ambient Air Quality; (Eds. R, Perry; P:W; Kirk): Selper, 1988: pp99-112. 48. Thiery, J.; Paper given at Toxicology Forum. 1990. Budapest. pp213-219.. 49: Cohen, A.1., Roe, F.1.C.; Monograph on, the pharmacology and toxicology of nicotine. Tobacco Advisory Council Occasional Paper 4,1981, London. 50. Jarvis, M.J., Tunstall-Pedoe, H., Feyerabend, C,, Vesey, C., Salloojee, Y.; Biochemical markers of smoke absorptioniand self-reported'exposure to passive smoking. J. Epidemiol. Comm. Hralth 1,984: 38; 335-339. 51. Coburn, R.F., Williams, W.11, Forster, R.E.;',Effects of erythrocyte destruction on eatbon monoxide produCion in man.1. Clin. Invest. 1964: 43; 1098-1103. 52. Reasor, M.J., Will, J.A.; Assessing exposure to environmental tobacco smoke: 15 it vali6 to extrapolate from acti've smoking? J. Smoking-Related'Dis. 1991: 2; 111-127. 53. US Surgeon General. Reducing the health consequences of'smoking. 25 years of progress. US Departmen; of Health and Human Services, Washington DC, 1989. 54. Wald, N:J.; Honizon: BBC TV Programme, 23rd January, 1991. 55. Wald, NJ.; Ihdependent on Sunday: 13th January; 1991. 56. Garland, C,; Erraturn: Am. J. Epidemiol; 1986: 122; 1112. 57. Hirayama, T.; Non-smoking wives of heavy smokers have a higher risk of lung cancer: Astudy from Japan. Br. Med. J. 1981: 282:,183-185. 58. Doll, R,; Cornversation with Sir Richard Dolli Br. J. Addiction 1991: 86; 365-377. 59. Wald, N.J.. Idle, M., Boreharn„1.; Serum cotinine levels in pipe smokers: Evidence againstt nicotine as cause of coronary heart disease. Lancet 1981: ii; 775-777. 60. Burghuber, O.C., Punzengruber, C.L., Sinzinger, H. et aL; Platelet sensitivity to prostacyclin in smokers and non-smokers. Chest 1986: 90; 34-38: 61, Roe„ F.J.C.; How Inadequate ventilation may impair health: and well-bQing, In: Other People's Tobacco Smoke (Ed. A.K. Armitage). Galen Press, Beverley; UK. 1991: pp5-20. 62. Aronow, W.S.; The effect of passive smoking on angina pectoris. N. Engl. J! Med. 1978: 299; 21-24. 63, Budiansky, S.; Food and drug data fudged. Nature 1983: 302; 560. 64. Horvath, S,M.; Investigation of studies by Aronow of the effects of low levels of carbon monoxide on heart patients. 1983 Statement by the Environmental Protection Agency (EPA). PA Ref: (202) 382-4355, p2. 65: Sheps, D.S., Adams, K.F.,,Bromberg, P.A.; Lack of effect of low levels of carboryhaemoglobin on cardiovascularfunaion in patients with ischemic heart disease. Areh, Environ, Health 1987: 42; 108-116. 36 O