Tobacco Documents Online

INDOOR AIR QUALITY AND VENTILATION Ediced by F. Lunau and G.L Reynolds lW 4,e ttN.a&., '~b. F. o..J Mu.t^~Y, J. ~.;sQaw Ve2:~N -2iE

INDOOR AIR QUALITY AN D VENTILATION Sold and distributed by Publications Division, Selper Ltd., 79 Rusthall Avenue, Chiswick. London W4 1 BN ISBN 0~ 948411 06 6 Copyright held by QSclper Ltd, 1990 Published by Publications Division, Selper Ltd., London Printed by Pri.ntext Ltd., London 00

~E may be :opyright .S. Code). ENVIRONMENTAL TOBACCO SMOKE (ETS) AN'D CARDIOVASCULAR DISEASE t).R ,,.d 1 %l..nt.,. C.f....~{..6 n,..,. 1.." c..... ..l.nd r.,k,..I ...... C.....t- la,.d .CRI .cU. t, A ASSTRAtT The epSdesiological evidence relating ezpoaure to ETS and cardiovascular dlseases has beea ezaained: al'1 of it is fTaved. Host of the di'fficulties arise fras the different studX deslgns., Three of six studies reporti an increased RR for cardiovascular diseases, although the others failed to do so. It is concluded that no :ncreased RR has been established unequivocally, either because there is none, or bscause the inadequate design of the stud5es frustrated their objective. INT.RODDCTI!ON Ezposuresoto -mven.iron.ental--tobaceo - s.oke=(ETS) -'-=Aas --baen associated_,vith a nuaber of.serious _diseases in aan•. In virtually all casea, it has not been possible to siaulate In anisab sodels the adverse health effects reported in •an, so the evidence depends upon the flndings of epidesiological studies. There is distinct veakness in the design of these studies, so.e of which are peculiar to the evaluation of effects of ETS (',1Y, and others which are cosson to all epid'esiological i'nvestigations (2). Vith respect to 5TS, two factors prevent us reaching unasbiguous anavers; first, there is the poor assesssent of the extent of exposure (3), and secondly, there is the atsclasaification of soae cigarette ssokers or ex-ssokers as non-saokerr (4). bn the report of the Surgeon General (5), less than 2 of 359 pages are dedYeated to E7S and cardiovascular disease. In another coeparable reviev, conducted by the Nev Sork AradesT of Science (3), onlx, 11 of 337 pages refer to cardiovascular diseases. The present paper considers six epidesiolsagtcal studl.s identified In an extensive search of the literature (6-11). The firet thing to note fro this databu e In that the .ajority ef the studies were not designed specifically to in.estigate the effects of ETS exposure on the incidence of cardios u cular disease, but were adapted to this purpose froa sose other, once the initial clLaisa that ETS affect health adversely bad appeared in 19$1'-2 (see 12, 13). This adaptation of existing studies has led to effects being sought in populations that are not representative of the population at large. Hov is exposure to ETS quantified? No substanee is known that 211 ~

Is representative of all the coaponent's of ETS (3:Y, so it Is not possible to onltor such exposure Sn aeaningful anner. Instead, epideslologists havs to resort to soae subjective Index of exposure, usually in the form of the smoking behaviour of couples 1Sv1ng together. In eost studies, the tncldence of a specific medical outcome i determined in a group of non-sackers arried to clgarette eaokers, and thi'.s ratie Ss then~ compared with that in non-saokers arried to non-saoliera. In this way, exposure to ETS in the hose can be assessed, although no~ allowance is aade for exposure of both groups to ETS outside the hoae, or for any effects arising fro exposure to other, potentially toxic, agents. It seems likely that any effe:ts of ETS, Sf there are any, vould' be masked by the variability Sntroduced by theme confound3'ng Snfluences. The smoking status of the partiel'pants in epidemiological etudles is determined from quest'Sonnaires. The reliablIl'ty of the answers to the queitiona that eonetitutes this part of the experimental d.s1'gn is low. It Ss inevitable that soae subjects are iscLassified with respect to their cigarette smoking behaviour; this probably results from slaple failures Sn a uory, or is the consequence of giving false answers to avoid adaitting that they have a habit which is considered to be socially undesirabl~e. All the reports considered in this review have appeared since 1985, 1.e. after the first suggestion in (1981)' that ezposure to ETS may be associated with serious health problems (12,13). H'owever, the aeasuraaents on the subjects in the trials were aade la the 1970s, S.e. retrospectively, and thus can be considered to be the result of data dredging. REVIEW The Garland~ et a1 study (8) was perforaed aver a 10 year period after enrolment between 1972 and 1974 of 922 of: the aduLts aged betweeo 50 and 79 in a eoaaunity of San DS'ego~, in the U.S.A. In the period under conslderation, there were 19 deaths from ischaealc heart disease (determined by analysis of death certificates).: Only two deaths were recorde&in the controL group (non-sackers arrled to non-saokers), which probably represents too~ low a baseline level to persit safe pred'ictions from these data to the population at large. The age-adjusted death rat'es for Sschaealc heart diseaae were not significantly elevated in th. subjects considered to be exposed to LT,S (P > 0.1), and were higher In those arried to ss-saorers than in thoee arried to current saokers. Tbe Snvestigatlon reported by Lee et al (11) was a case-control study initially designed to examine lung cancer risk. Yith rsspect to isehasaic heart disease, no statistically significant increu ed risk was associated with supposed exposure to ETS in the hose, at work, or from travel and Ieisure. Svendsen et al (9) selected a sub-group of patients for analysis fro the cohort in the multiple risk factor intervention t'rial (MRFIT). MRPIT was designed to aeasure the effect of different interventions on ortality of patients Sdentified as being at high risk of coronary heart disease. Men aged 35 to 57 x,ears old were recruited in 18 cities in the D.S.A., and fol'lowed' 212

for aean tiss of eeven years. THe smoking status of the cohort .as daterained by qusstlonnai're and, unusually, for such studies, confiraed' by objd) etive aasures (S.a. aeasurament of seru• thd'ocyanata levels and e:haLed carbon onozi'de);, the cause of death vss determined by a committee of three cardiologists reviewing the case-papers on a blind basls (:1.a. not avare of the treatment aseignaent, or, In this case, the spousall r.oking status). Thare vas a saall' apparent increased relative rl4k associated with exposure to ETS in the hoas„ but this did not reach Llie leval of etatistlcal si'gnificance (P > 0.05 for all co.parieons„ except for d'eath fro any cause, when P• 0.01'). This very carefully conducted Snvestigat'ion was characterised by the small sise of the groups under conelderatlon (controls 56 non-fatal and fatal eventa in 959'subject.;, spouse smokers 26 events In 286 subjectr), and the sub-group selected ay have been atypical, bacause it vas chosen fso the highest 15= of those at risk fro• cardiovascular disease. Halsing at al (7) considered a population of 91,909 vhlte people from Yashington County, Maryland, U.S.A. aged 25 or older on entry in July 11963. S.oking staLus was det'erai'~ned froa rasponses .ade to a private census conducted in 1963, i.e. before the publication of the Surgeon Generall's fi'rst report in 1964 (14'), and then allocated a score on the basis of the extent of smoking, whether or not it was current, and the type of tobacco consumed (cigarette, pipe or cigar). The population vas followed for 12 years, with the cause of all deaths determined fro death certificates. There were 2022 deaths fro. arteriosclerotic heart disease 1n non-saokers. The adjusted' (for age, marital status, years of schooling and quality of housing) rates of death fro arteriosclerotle heart d'lsease of the population vas then ssessed vith rsapect to ezposure to ETS In the home; a stat3sLleally significant relative risk was detected for both sen (1.31, 95Z confidence li.its 1.1-1:.6)' and vo.an (1.2G, 951 confidence li.it. 1.1-1.4). Hove.er, St vau not possible to da.onatrate any incrasa in risk vitbincraased exposure in •en, but there was such a relationship In women (P < 0.005). The study of Helsing et aL can be critlcl'sed in a nu.ber of ways. First, the smoking status of the subjects was determined once in 1963, so there was no possibility of alloving for any subsequent changes in behaviour. Secondly, there was no infor.ation on the established risk factors for the diseaae, such as blood pressure and serum chol'asteroL levels In the exposed and uaezposed groups. Pinally, the sod-point used in the study relied on death certifieates, which are prone to considerable inaccuracy (15). The Oillls et al study (6, 16) was set up in 1972-76 in R.nfrev and Paisle~ in an attempt to detect any special circumstances that aT relaqe to the very high rates of lung cancer and cardio.aseular di'sease that occnr in tbe vest of Stot'land. Men aod voaeo batvean tbe agea of ('6 and 64 vere recruited and s.oking behaviour deter.ined fro• a self-advlnistered qusstionnaire, and subsequently chacked by an experienced Snter.iever vhen the subjects attended a screening oentra. A cohort of 15399 subjects vas identified (80Z of those quallified to participate), and folloved for an average of 11.5 years. Cause of death vas determined from death eartifieates. From these data, it was 213

rossible to calculate the relative risk associated with exposure to rTS in the hose. T.he on:y statistlcalIy sigr.ificanL RR was for ischaemie heart d:sease i'n non-smokers (2.01, F= O.C08?, which is a resarkablr n'Sgh,value, because the RR'fros smoking vas only 2.27. '"ere were 30 deaths 1'n , the control group and 54 in thooe considered to be exposed to ETS In thel'r homes. khen :ardlovascul'ar symptoss vere detected in the eeeening eoaponent of this study, there were no statistically significant differences Eetveen the exaosed and unexposed groups vi'2h respect to engine and sa;or abnormalitles of the electrocardiograa. t Ss possible that '!ie ischaeaic heart disease zor:a'_iTy rate fcr those _:ns'_''dered . .e exposed to '_".S represents a spurious f!n:ing, because of a•, a=stnee of aw effect on the pre-tersl'nel car.isvsscu_ar s;+mptoms, nnd because of the magnitude of the effect relative to that seen `-n saok'ers. The final study in t?i1s database Ss the one fro. Japan, described by ?irayaaa (10). A prospective epi'deslologlcal study on a cohort of 2e5,,118 Japanese peoplie was initiated in '-966, with a vlev to deters:ntng the incidence of serious disease. The cause of death used as an end point in the investigation was taken fros death certlf:cates. '.li1s study _'ed to the first suggestion that exrosure to T_'_S was assoclated vith an increased risk of lung cancer (1'?):' further consi'd'erat.on,has resulted in add'Stiona: clai'as of' adverse health effects due to spousal cigarette s.oki'ng, _nc'-uding a R3 of 1.31 for 1se!rsemic heart disease ('P'• 0.019)' 1r. -on-smoiing vcman in '98G (17). Thd's is a surprising finding, because a report three years earlier on the same cohort led to the conclusl'an that "passive smok'ing did not see to lncraase the risk of d'evel'opi'ng ischaemic heart' dlseass" (13). The importance of 9lrayama'a various reports In the field of aaverse health effects of ETS cannot be underestimatad. Jnicrtunataly, most' of the influence has arisen from a poorly. 9eaigned' study , v!iich has been uch criticised, partly because of the c.elear .ay it has been, preaented in the literature. The cohort was assembled as a convenient sample, rather than as a reprasentative one, which has resulted in over-dependencs on certain catagorles of the population at large, e.g. agricultural workers and young people (only 2Z were over 60, whereas 121 of the Japanese population fall 1'nto this category)'. Japanese vomen spend much of their tlae 1n ssal'1 rooms, where aay effect of ETS voul~d be greater than 1n the larger Sndoor air spaces frequented by those who live in tl:e vest. T.~a cooking habits of the wives of Japanese egrleultura'- vorkers ay have confounded the study, because Rerosene stoves would Aave been used extensively, and these are kncwn to em1'S large quantities of potanti'ally toxic gasses and particulate matter P18). The miscliassification of saokers and ex-ssokars of cigarettes as non-ssoker may also have contributed to the amplrical fimdi'ngs. No doubt the Snfluence of any of these factors could have been taken i'nto account In evaluating the validity of the result's, but this has not been possible because !?irayama has refused to give other epidemiologlsL access to hi's data (19). Such secrecy doas not inspire confidence im the objectivity of the conclusions reached in the Japanese study. 214

coNCLaSIOgS Of the six studies consid.red here, only three revealed anyy significant effecta. Each of the positive studies was flawed ln some way. Tvo consistent probLsas are apparent: too strong a reliance on the aeeuracy of death certificates, and the unre11ab1Ti'ty surrounding the deteradnation of the smoking status of the subjects and thelr spouses in order to measure of exposure to ETS. It is eoncluded'tha: no Sncreased risk of cardiovascul'ar disease can be associated unequivocal'ly with exposure to ETS, and it seema probable that this will continue to be the case until specifically designed t'siels are Snstigated', and some obj:ective aeasure of degree of exPosure ean, be devised. REFERERCES 1. Veetman, D. F. (19q09 Indoor Air Pollution: Proble.s and Priori'ties, ed. F. V. Lunau and C. H., Leslie, in the press. 2. Feinstein, A. R. (1'988) Science, 242, 1257-1263. 3:. Hulka, B. S. (chalraan)i(1986). Environmental Tobacco Saoke. yeasurlne Ez osuras and Assessl'n Realth Effects. 4. Lee P. N. (1988). Mieclaseification of Saoking Rabtts and Passive S.okinA: a Reviev- of t'_he Evidence. 5. The Surgeon General's Rsport (1986). The Rsalth Conseouences of InvoluntarY S.okina. 6. Hcle, D. J., Cillis, C. R., Chopra, C. and Havthorn., V. M. (1989) Br. med'. J., 299, 423-427,. 7. Helsing, I. J., Sandler, D. P., Comstock, G. V. and Chee, E. (1988) Am. J. Eo d'emiol'., 127, 915-922. 8. Garland, C., Barrett-Connor, E., Suarez, L., Criquil, M. R. and Vingard, Ds L. (1985) Am. J. EDideaiol., 121, 645-650. 9. Svendeen, I. B., fiuller, L. H., Martin, M. J. and Ockene, J. (,1987). Am. J. Eflidemiol., 126, 783-795. 10. Hirayama, T. (1985) Tokal J. ezo. clin. M.d., 10, 287-293. 11i. L.e, P. U., Chaaberlain, J. and Alderson, M. R. (1986Y Br. J. Cancer, 54, 97-105- 12. Trichopoulos, D., Ealandidi, A., Sparros, L. and MacMahon, B. (1981) Int. J. Cancer, 27, 1-4. 13. Hirayasa, T. (1981,) Br. Med. J., 282, 183'-185. 14'. Surgeon Gan.ral's First Report (1964)' Smoking and Health. 215

15. R1111,, A.B. !1q77) A Stiort '.ezrbook of yedl_al,_St_etiatlce' 16. G111'Sa, C.R.. Hole, D.J., Ravthorne, V.M. and Boyle, P , f,i98G). Eur_- J. DSe.,6S (Suppl 133), 12'-1'26. 1'-7. Rirayaaa, ". (198C) LunoCancer: Causea and Prevention, pp 175-195, ed Mixell:, M. and Correa, P. 18. Saeet, J.M., M.rDury, M.C. and Spengler, J.D., (1987) Aa. Rer Resy. DIa., 1i36, 1G86-1508. 10. Db.rla, IC. '~08?]. :ndoor A_r uallty, pp <5-60, Ratior.al Acadsey of S'eien<ee, Argentl'na. 216