Tobacco Documents Online

UNDOOR AIR QUALITY Il~T ASTA Proceedings of the International Conference He1d at the Central Plaza Hotel', Bangkok, Thailand on 28'-29th November, 1991. Edited by B. R. Reverente Jun Philippine Refining Co Inc 1351 United~ Nations Avenue Mani1ia PO BOx 1176 PliilipPpines D. F. Weetman School of Health, Sciences University of Sunderland Sunderlan& England~ and' AS. Wongphanich Occupational: Health Department Faculty of Public Health Mahidol University 420/1 Rajvidhi Road Banokok 10400 Thailand Published by Indoor Air International The InternationalU Associataon for Indoor 4irQualitv, Postfach 2 CH-4467 Rothenfluh Switzerland ISB.N 3-906470=001S April 1993 1

, Indoor Air Qualin,, Ventilation and Enerp7 Saving Qigao Chen , Refrigeration Using Waste Heat To Improve Working Condiuons. Jiajie Liu, Xianying Liu and Chunw-u Sun~ Reliability of Quesuonnaire Data. Jun b:aSa,wai A Survey of' the Quality of Indoor Aii in Manila Office Buildings. Benito R. Revererite„Jr. Sources of Polltition The Effect of Outdoor Air Pollution on The Mortality of Indoor Residents. l:in-hua Chen and Qih-uan Deng Industrial Pollution in Vietr,am: An Important Ea-ternal' Source of Ind'oor Air Polllrtion. Le Van Trung, The Environmental Impacr of Carcinogens from Motor-Car Exhausts. Manfred Buck Identification of the Sources of Airborne Particulate Matter in Urban, Suburban„ and Industrial Areas of Korea. Sung-Ok Baek and Myuns Ho Hahn~ Non-occupational Exposure to Lead in an~ Occupational Setung; Eiji Yano Sources And Indicators Of Indoor Air Pollution In The Occupauonal! Setting. D. Soedirman Airborne Asbestos Concentrations in, Rooms 'with Sprayed-On :A-sbestos Materials. Iziiyoshi Sakai,. Naomi Hisanaga:, I:azunori Mitani„ Hironobu Tsuchiva, Itian Huang, Eiji Shibata, Yuichiro Ono, Akinori Kojima and Yasuhiro Takeuchi' The Contamination of Indoor Air with Asbestos and Man-made Mineral Fibres. John A. Hoskins and Robert C. Bro'wn. V1 83 97 107 117 121 127 133 141 155 161 171 ~ ~ 185 w ~ F"a 0 Ph ~

Concentration And Exposure Dose Of Radon And Its Progeny In Public Bath Houses at Two Hot Springs. Qing-Xiang Zeng, Yan Li, Zhao-Hui Huang, Hui Su, Hong-LiniWang, and~hiu-Ian Hui 95 Level and Dose of Radon and Its Progeny in linderground Buildings in Wuhan City. Qing-Xiang Zeng, Yan Li, Zhao-Hui Huang and Hong-Lin Wang. 199 RadoniConcentration in Drinkino Water and the Dose Received by the Exposed' Population: Yibin Chen 205 Control of Welding Fumes in the Metal Fabrication Industry: R.C. Panjwani ?13. Control of Dusts for Small Cement Factories with Shaft Kilns. Guangquan Liu, Jiang Liu, Chen Zhou andi Gushua Hu 2?9' Asbestos Concentration In Indoor Air And Management, Of Asbestos-Removal Works In Japan. Susumu Tod'a and Mikiharu Aoyagi 237 Standardization, of the Measurement of Asbestos Fibres in Indoor, Air. Norbert Hoefert 243 Continuous Measurement' of Indoor, Air Quality in Office Buildings in Japan. Akiyoshi Ito 251 PIXE Application to Indoor Air Quality Measurement. Toshitami Ro, Takeshi IShiguro;,Mitsuru Fujimura and Yoshikazu Hashimoto 261 Health Effects A critique of the methods used to assess the toxic effects on maniof combustion products, Di F. aJeetman 275

Health Problerns in a Thermoelectric Plant in Vietnam, Nguyen Do Nguyen~ Nguyen The Dung and Pham Nboc Len ?S7 Concentrauons of Pyrethroids in the Air of Dividins and Packing Workshops. P Yao, J Sun; Y W.u, S Wang, L Liu and F He 293 Prevalence of Welders' Pneumoconiosis in Workers Exposed to Welding Fumes. Changqi Zou,. Kangji Xng„ Yanhua Yuan, Qingchens Du„ Yuxi~ Hbu and! Zonsshu Mao 299 Key T infect Neurotoaic Effects Of Carbon Disulfide On Ravon Workers: Fengsheng He, Zhemin Zhang; Shi Yang, and' Shoulin Zhang 305 Further Anallhses of the Role of' Confounding Variables in Epidemiolobic Studies of Environmental Tobacco Smoke and the Respirator}'System in School'-age Chiidren. Raphael J, W'itorsch, JosephiM. Wu, Ronald Di Hood and Philip Witorsch 313 n envirc partict. Factor Th( t}ie pu kind c inapp-n

E 0 in order to seek a. j within buiidings ,f air conditiotdng ronment„ and willl :hanism Of indoor A CRTTIQUE OF THE METHODS USED TO ASSESS 'I'HE TOMC EFFECTS ON MAN OF COMBUSTION PRODUCTS. D, F. R'TEETMAN'N spaces". n ' n. R _, DOE-ER.6()493-7-vol ,~ 1 Analysis. John W-del events in Japan duriq sources of indoor a:id -336. as detennined by PDcE Nucll lns4r, h'larhPho School of Health Sciences, University of SunderJand , Sunderland, Tyne and Wear, England Key Words: combustion~producrs, epidemiology, risk assessment, environmental tobacco smoke, cardiovascular diseases. ABSTRACT Combustion of organic material results inahe release ofpaniclds, gases,,and~pyrolyuc products. all of which can accumulate in the indoor environment, and could damage health. The methods of assessing risR to mamare reviewed, and it is conduded that epidemiology provides the bestsingle approach. The problems in unerpretatiorrof epiderniological studies are reviewed; with, particular, eemphasis on the specific problem of environmental tobacco smoke (ETS) and cardiovascular disease. tt is cond'uded that too many important potentially confounding factors have been overlooked to decide if there is an, association between exposure to ETS and cardiovascular diseases. NATURE AND SOURCE OF CO."vfBUSTION' PRODUCTS Iti is difficult to imagine life v,ri•,hout~ combustion. In addition to domestic heating, combustion of some form of fuel occurs in cooking, many forms of transportation, most industrial processes and most of the generation of electrical power. The common factor is that some form of fuel is burnt, and the fuel is derived from organic matter, a•ith the inevitable release of pollutants. Not all combustion, contributes pollutants to the indoor environment, but in those situations where this effect appears to be minimal„it should be remembered that the indoor air is derived from, that outdoors, so the dirtier the air outside a building; the more polluted it will be inside.. When organic matter is burnt, three classes of pollutant, are formed. Firsr.there are gases. As the pred'orninant chemical process is oxidation, which usually occurs without sufficient oxygen to allow complete oaidation, there is the release of 2: complex mixture of'~ oxides. Thus amongst the gases generated bvy combustion, there are oaides of carbon, nitrogen and sulphur. 275 3SM

Assessmenr of the Toxic Effects of ComBusrion Products Organic molecules can undergo complex re-arrangements without, the consumption ofoaygen in a!process called pyrolysis; which is largel', deterrriir,ed by the temperature of' combustion. Finally, there is the generatjon of pardcles which are the small globules of organic matter that give rise to the visibility. of smoke. Particles vary in size, and undergo complex changes in shape andli surface area when they cool down to the ambient temperature. Large panicle5 do not stav suspended in the air for long. Particles up to 10}en are readily i.nhaled, and may not be rapidly cleared by from the lung;, large particles (1oj,, aerodynamic diameter) are deposited in the upper respiratory tract„ whereas small ones (O1yn) are exhaled Ill. The important aerodynamic diameter with respeet to potenval pulinonarytoxiciry is probably about O.S;an [1]. Much the same tnucture of gases, pyrolyuc products and particles are generated from the combustion of any organic matter. For example, there is more similarity than difference in the products released from burning woo& and tobacco. With combustion of organic matter that has undergone some degree of inetamorpho- sis; as with coal, oil and natural gas, there is variation in the proportions o; components generated from eachs,but again there is a mixture of gases, pvrol±,tic products and particles. METHODS OF RISK DETERNII.hA.TIOh investigations. Any reliable findings about risks to the health from combustion products can be used i.n attempts to regulate levels of'the offending substances, and~ thus protect man., justifi-the funds they consume„so there need to be an acceptable purpose oflthe ' The purpose of determining the risk to rnanftom combustion products arises from~ the ubiquitous distributioni of these substances. Scientific knowledge is sought because of man~s insatiable curiosity;, but increasingly scientists have to Iaboratory Studies Laboratorv animalscan be exposed to smoke and then assessed for anv effect. However, this type of eaperimentation is fraught with difficulty. There is always some carbon monoxide generated by combustion, which prev.ents high~doses of smoke being administered. The anatomy of'the respiratory system of, say.,,rats, is quite different from that of man, so it is difficult to~predict the outcome in this latter species from effects seen in the former. Any experience of pharmaceutical research teaches that there is no completely reliable way of translating effects seen in laboratory animals to man. To determine what happens in man, ii is necessaryto investigate in man. Direct eaperimentation in man can be achieved'in exposure chambers. The problems here are that exposure must be shon-teran, and there is still an upper ?76 ltmit to the dosc course,,it is pos to volunteers ir but then it is b:, components. administering in vitro svster inadequate for Epidemiolog, It is possil combustion p population of outcome in th study designs with a conditi to the suspea, The second'a; determine wi method does causes of d'ea summarises t: Table 1 Prc The bes: gation is to regulators 2 exposure to ate exampl( disease. so the investii products fr

Assessment of the Toxic Effects of'Combustton Products without the ~ determine& of particles, e visibility of' nishape and irge particles 3dilv inhaled, -icles (10Fan -act, whereas iiameter with _1 . Much the ated from the irnilarity dian bacco With : netamorpho- roportions of ises, pyrolytic ~1 )ts arises knv.wledge is 2ntists have to ourpose of the rn combustion ag substances, i for any effect- here is always nts high doses :em of, say, rats )utcome in this :)harmaceutical aslating effects zs in man; it i5 -hambers. 'rne :s sull an uppef limit to the dose that can be tested due to the presence of carbon monoxide. Of course, it is possible to separate the various chemicals in smoke, and apply them to volunteers in exposure chambers, either individually, or in defined mixtures, but then it is barely possible to measure any interactions between the differenrt components. Anv reduction in the scale of the test system, for example by administering smoke to tissue cultures, still involves an extrapolation from the in vitro svstem to intact man. The experimental methods available are inadequate for this purpose. Epidemiology It is possible to measure the effects of long-term exposure of man~ to combusnon products by epidemiological techniques. With this approach; a population of exposed individhals is identified and the rate of any medical outcome in this group is compared with that in a suitable controligroup. Two study designs are possible. First, one can either start by identifying individualss with a condition (i.e: cases) and attempt to show a greater exposure in; the past to the suspected cause thanloccurs in a demographically matched control group.. The second approach is to assemble a population (i.e. cohort) of individuals and determine what happens to them medically over several' years. The cohort method does not provide rapid answers, especially with the most frequent causes of death„which frequently have a slowly developed pathology. Table 1 summarises the problems associated with such studies. Table 1 Problems in Epidemiology 4. 5:. Selection of exposure and control groups Multifactorial nature of disease Difficulty imcontroV?ing,confounding variables Only associations detected Intervention studies are difficult. The best way to understand the difficulties of such epidemiological investi- gation is to examine the quality of the evidence in a specific case. As the regulators are currently considering the health effects that may result from exposure to environmental tobacco smoke (ETS), this will serve as an appropri- ate example. The latesn claims are that ETS is causally relatedto ischaemic heartt disease, so the quality of this evidence will be considered in detail. However, the investigator is confronted by comparable difficulties with, combustion products from any fuel source. 277 E M 0 0 ®

Assessment of the ToxicEffects ofCombusrion Products Study Design The study design is shown in figure 1. The exposed group are obtainedby selecting non-smokers who are married to smokers,, whereas non-smokers married to non-smokers provide the unexposed group: The rarlonale of tlis design is that the non-smokers (usually wives) would be exposed to ETS in the home from the smoking of their spouses. With these two populations, it is possible to compare the rates for ischaemic heart disease, either in ~ case-control 1 or cohort studies, The problems of interpretation arise from the imprecise distinction between the two exposure groups, and because there is no allowance made for exposure to ETS outside the home. A second difficultyy is the determination of' the smoking, status of the spouse, whichi is usually achieved from the response of individuals to questionnaires. Any wTongl~ classified partners tends to reduce the precision, of the study. Exposed rate - events/population J' [ adjusted I t RR EXPOSED RATE UNEXPOSED RATE I [ adjusted J 1 Unexposed'rate - eventsLpopulation 1 NON-SMOKER married to a NON-SMOKER . A qu ~-holha that the [2). As eaampl I [3]), the patient If t}i scienris~ second to card cardiov risk fac: scicntis sugges not to c or behh cardio% studies to be r,Mucc by Gla: respon studies suppos Glantz connec study, [I Tat cardio, contair follow: arteriostrokel inform

0 Dbtained by:)n-smokers nale of this .) ETS in the .ations„ it is :ase-controt : imprecise :) allowance =ulry is the 'y achieved v classified M Assessment of tbe ToxicEJJects ofCombustion Products A quite differenuset of'problpms relate to the definition of an effect. i.e. those who have the disease. If one relies on deatli certificate information, then the fact that the individual has died is reliable, but the cause of death is prone to error 12). As few autopsies are now performed in most countries of the world (for er.ample, currently, less than 1'3°/p of bodies nthe USA are subjected to autopsy [3l), the cause of death lias to be deduced' from ~ the signs and ~ symptoms of the patient alone: If the possible causes of' ischaemic heart disease are considered, medicall scientists do not have an unequivocal answer. This has resulted in a sort of' second level approach, where risk factors apparently predisposing individuals to cardiovascular disease are identified. In other words, the aetiology of' cardiovascular disease is considered to be multifactorial. To date„over 200 such risk factors have been proposed by various research groups [4) (not allimedical scientists agree with, this approach: McCormick and Skrabanek [5) have suggested that we should refer to risk markers, as opposed to risk factors, so as not to confuse association with causatdon): However, if there are constitutional or behavioral characteristics of individllals that could predispose them to cardiovascular disease, each one should be controlled! for in epidemiological studies, otherwise they may acuas confounders, allowinganiincorrecrconclusion to be reached. Much of the interest in ETS and cardiovascular disease arises from a: review by Glanz and Parmley 161, in which, the authors presented! a case that' ETS was responsible for a proportion of the cases of the disease. Nine epidemiotogical studies were identified in the literature (5 with a statisticallv significant effect, supposedly from spousal exposure (7-11)), and'4 without (12-151, which allowed Glantz and Parmley to conclude "These epidemiological studies demonstrate a connection between ETS exposure and' death from heart disease." One other study [16] has been added to the ones covered by Glanz and Parmley: Table 2 contains a summary of ohe epidemiological i studies linking ETS with cardiovascular disease. The medical endpoint varies from study to study, and contains both "death from" and "possession of" the specified condition;, the following conditions were taken as the endpoint: ischaemic heart disease, arteriosclerotic heart disease, coronary heart disease, myocardial infarction, stroke, and all cardiovascular diseases. The reliance on~ death certificate information as the medical endpoint is also indicated in table 2. 279 M ~ ® ® FS © 0

Assessment of tbe Tozic EJfects of Combustion Products Table 2 The Epidemiological studies of ETS and cardiovascular diseases. STUDY REFERENCE DISEASE END POIN'T TOTAL EVELTS 7ESTS sig/tota] BUTLER [121 CHD D.C ~ 0/1 HIRAYAb1A 1101 IHD D:C. 494 1/3, GAR1A.h'D 113]1 I.HD D:C 19 0/1,0 LEE (a51' IHD * STROKE DIAGNOSIS 121 0,130 SVENDSEN 1111 CHD PA','EL 13 ?j17 HE 181! CHD DIAGNOSIS 34 24Y47. HELSING 171 AHD D.C. 2014 10/22 HOLE I 191 ' IHD D:C + PA'.'EL 84 1/5 HUMBLE (14] All CVD D.C. 76 0/24 DOBSON [aG hiI D.C - PAN'EL 382 2!1~ Under EYELTS ;,the numbers refer, to those in the CONTROL * E7TOSED groups for the cohor, studies. and'thenumber of'cases in the case-control studies, and give an indicauon of'the ability of the tests to detect aneffect. The values are for males and itmalts combinedi,and both home and work place exposure to ETS. IHD - ischaermc heart diseasc CHD = coronary, heart disease AHD - anenosclerouc heart disease Ml - myocardial' infarction CVD - cardiovascutar disease D.C. - death ceruficate PANEL - evidence reviewed by a panel of' experu DIAGNOSIS - disease deternuned by standard diagnostic techniques The large number of comparisons made in the papers is indicated in Table 2. When multiple comparisons are made on one set of data, the comparisons are not,truly independent(some values will be used andthen re-used indifferent tests); so anv study containing a stausucally significant outcome willl be considered furuher. The positive studies are re-considered in Table 3. It has already been stated' that cardiovascular diseases are thought to, be multifactorial in their aeuology, so anv putauve causes other than exposure to ETS that are not controlled for will be capable of confounding.the epidemiologi- cal studies. For this reason, some of the best established risk factors for cardiovas cular diseases have been identified from the literature (Table 4), and the ex[ent of the control over these potential confounders has been detemlined (Table 3). -i Table 3 Missing epidem S7L'D1.. HIRA1'A'`!A 110) SvEI\'DSEI\ I1'1]1 HE [8] HELStN'G [ 71 HOLE 19) DOBSON' 1]61 in no study was the : Fami]y history of ca: [81- The Svenden behaviou:. Table 4 The be: RISK FACTOR Family history, of' d. H,vpenension Cigarette smoking Dietrv fat loa& Diabetes Lack of exercise h9enopausal' status. Alcohol l consumpuc Obesu}• From table 3 i same standard. Svendsen et a] (`. exposure to ET` concentration). dietan~ fat intakc population selecti the U.S. populati The cardiovascui were high blood and those most a anvy effect of ETS

diseases. TESTS sig/total 0/1 1/3 0/10 0/30 2/17 24/47 10/22 1/5 0/24, 2/12 -ated in Table - comparisons ed in different =ome will be .ble 3, :hOughL to be n exposure to epid'emiologi- for cardiovas- and'the extent 'Table 3). Assessment of the Toxic Effects of Combusnon Products, Table 3]vfissing evidence (potential confounding variables) in the positive epid'emiological studies STL'DY Obesitv I Hypenension Alcohol I Fatty Diet HIRAY;tiMP. [1011 NO NO YES I NO SVENDSEN 111] YES YHS YES YES HE [8] YES YES YES YES HELSING [7] NO NO I NO NO HOLE [9) YE.S YES NO YES DOBSON 1161 NO NO: NO NO. In no study was the following controlled'foc diabetes. exercise and menopausal'status in women. Family history ofi cardiovascular disease was not controlled for, except in the study by He et al [81. The Svendsen ~ et a1 study ll i l was the onN., one to employ a marker to detect smoking behaviour. Table 4 The best established cardiovascular risk factors RISK FACTOR SELECTED REFERENCES Family history of disease [18, 19, 201 Hypertension [21, 22] Cigarette smoking 117, 221 Dieery fat load 122. 2311 Diabetes 11811 Lack of exercise 124;,251 Menopausal status [18, 26, 27) Al'cohol consumption, l4) Obesity 128] From table 3 it is apparent that the epiderniological studies are not all of the same standard. Perhaps the best designed! study was that performed by Svendsen et al (11]. This was the only study that attempted to confirm the exposure to ETS by measuring a marker of exposure (serum thiocyanate concentration). The potential confounders of hypertension, bod'y weight, dietary fat intake and alcohol consumption were all controlled for, but the population selected for study was atypical. The subjects were from the 15% of the U.S. population thought to be at greatest risk from cardiovascul.ar disease. The cardiovascular disease risk factors considered to be important in this study were high blood pressure, cigarette smoking and high blood cholesterol levels, and those most at risk possessedtwo of the three risk factors. When it came to any effect of ETS, this was measured' in non-smokers by the spousal smoking 281 ® M 9 19M e ® a m

Nam Y ~ 9 MM Assessment of rhe Toxic Effects of Combustion Products status method. Thus all ETS exposed subjects must have been both hypenensive and had high blood cholesterol levels. H'owever„3 rea11y important cardiovas- cular disease risk factors (family history of cardiovascular disease, glucose tolerance and'whether or not the subjects exercised) were not controlled for. The final point thar: needs to be made from this welld designed study was that exposure to ETS was not'associated with possession of a cardiovascular disease in a statistically significant manner: the significant result that qualifies it for examina_ don in Table 3 was between surrogate exposure to ETS and death from all causes,. The ra ostudies that provided the highest proportion of statisvcally significant associations between ETS exposure and~death from cardiovascular disease were by Helsing and others f71 and by He et al [6]. The study by Helsinget al was the least well controlled of all the studies considered here. There was no attempt to confirm exposure to: tobacco smoke (either from ETS or undisclosed~ smoking). No information was reponed about blood pressure, bod'}7 weight, dietary fat intake, alcohol consumption, familv history of cardiovascular disease„ glucose tolerance, exercise, and menopausal status of the female subjects. In fact this study is best considered to be a linkins,of death certificate information to the response to a self-administered questionnaire: When one considers the absence of control over potential confound'ers, no reliance can be placed on the findings. The case-control study from. China [811 has only been~ published in Chinese; but the 34 female coronary hearr disease patients were shown to be at risk from spousall smoking (OR = 3.52, confidence limits, P' = 0.05, 1.26 - 7.17). This remarkable level ofl risk greatly exceeds many estimates of the direcr effects of' smoking (9, 1711 The effects of potential confounding influences was assessed in a multivariate logistical regressionanahysis, where it was shown that the effects of surrogate exposure to ETS persisted when the following risk factors were controlled for;: previous history of hypertension, family history of hypertension, family history of coronary heart disease, history of passive smoking, amount of' exercise, and previous history of hyper-cholesterolaemia. However, this study on a!small group of only 34 patients needs to be extended, evidence of the effects of direct smoking d'etermined, and evidence of difference in diet between the cases and controls added. The quality of the other studies considered here lie between those of' Svendsen et 21 and Helsing etal. All are poorlytontrolled. A study of appropriate standard and size has not yet been performed, so it is not yet possible to decide whether or northere is an association between exposure to ETS and cardiovas- cular disease. The fundamental problems in study design consist of: a: the selection of the exposed' and control groups,, b: the exact classification,of disease; and c: the exercise of adequate control over the numerous potential confounding variables, h9uch, the same difficuln• arises withi anu attempts to examine the ® possible associauor. Table 5 To progrc medicall c taken. T criteria. The associauom, less than 3 1291 The effi:at shou There should t more cases tha: All the evidene tn vinually all cases, d the more persuasive 3 If one were a', cornbustion.produ, the case-controll st cause the diseasee causes the conditi properves of an causation can be amongst epidemic all cases, there is ry outcorne is causec WurcentJH Th: "n Occ1y1+ H Hill AB A Shc Hill RB, Ander~ 195E 1.2941

M Assessment of'trye Toxic Effects of'Gomba.estion Products n hypertensive ant cardiovas- sease, glucose :rolled for. The ; that exposure 3r disease in a it for examiraa- from all causes. ;allv significant ar disease were -1g et al was the was no attempt or undisclose6 _, body weight, ascular disease, subjects. In fact = information ~ to e considers the )e placed on the h iil Chinese„ ): nisk from -'6 - 7.17). This direct effects of es was assessed 'n that the effects -isk factors were of hypertension, )king, amount of vever, this study ~nce of the effects diet between the °tween those of dy of appropriate )ossible to decide i S and cardiovas- ,nsist of: possible association between combustion products and common diseases. Table 5 To progress from the associatzon~ of an environmental factor with a medical outcome to establishing causaliry, several' steps have to be taken. The whole of the evidence should conform to the following criteria. 1. The association should be strong enough to be persuasive: it is seldom the case a•ith RR less than 3 129); 3' There should be consistency of findings between diffuent studies. 3: The effect should be specific, or as near to this as possimle, with the exposed group. 4. The temporal! relationship with respect to exposure should be appropnate to the pat.hologrcal sequence oflthe disease. 5. There should be a dose-response relationship„whereby, the greater exposures result in more cases than occurs in the less exposed group of individuals. 6: There should be a'freedom from implausibility" with respecrto biological mechansms. ',:. All the evidence should be coherent and poinutoa-ards one concilrsion,. In virtually all cases, the full set of cnteria are not fulfilled, burithe nearer one is to achieving this,, the more persuasive is the argument, (Adaptedlfrom I21i) If one were able to show a statistically significant, association betweenn combusdonproducts anddisease inan epidemiologicalstnldy, as is the case with, the case-control study from China [$J', it is still! not evidence that the products cause the disease. Association is only association: to conclude that exposure causes the condition, further steps are needed. Table 5 indicates some of the properties of an association that would have to be demonstrated before causation cam be concluded. It should! be noted that there is much debate amongst epidemiologists as to the exact criteria for taking this further step. In all cases, there is an element of' subjectivity in reaching the decision that the outcome is causedl bv the infVuence studied. REFEREI\ C:l;.S M a © ntial confounding s to examine the 1. Vincent7H`.Thefateofinhaledaerosols:areviewoflobservedtrendsandsomegeneralizatioris: Ann Occun Hy 1996,34(6)`.623-637: 2. Hill AB: A Short Textbook of Medical I Statistics. London: Edward Arnold, 1984. 3. Hill RB. Anderson RE:-Rte Autopsy-Medical Practice and Public Policy. Boston: Butterworths, 196B:1-294. 283 a

Assessment of the Toxic Effects of Combustio1l Products 4: Hopkins Pr; Willianss RR: A survey of 246 sugge-sted'coronary+ risk:factors., h,r r 1981;40;1-52. 5. McCotTnick J, Skrabanek P: Coronary, heart disease is not preventable byy populauon interventions. Lancet 1988;3-839-841. 6. Glantz SA, Patmley )X'W: Passive smoking and heart disease: epidemiology„physiology, and biochemistry,. Circulation 1991;83:1-12: 7. Helsing I€J, Sandier, Comstocl: Gu'', Chee E: Hean disease mortality in nonsmokers livtng o,.itb smokers. Amencan loumal of Epidemioloev 1988;127c915-9Z2. 8: He Y, Li L, Wan Z, Li L, Zheng )C, Jia G. Passive smoking and coronary hean disease in wome~ China loumal of Preventive Medicine 1989;23;19-22. 9. Hole DJ, Gillis CR, Chopn C, Hawthome VM: Passive smoking and tardiorespirntory liealth in a general population in the west of Scotland. Br Med 1 1989:299,423-427. 10, Hirayarrta T:,Lung cancer in Japaa: effects of nutnuon and passive smoking. in: Mizell lL Coaea P ed. Lung cancer: Causes and Prevenuon. New York: Veriag Chemie lntemauortaL 1984: 175-195. 11. Svendsen L'Ii„Kuller LH, Martin MJ, Ockene JH: Effects of passive smoking in:ttie multiple risk factor intervention tnal. pmerican: loumal of Er?idemiolqM~ 1987;126:783-795. 12. Bualer •P: The relationship of passwe smoking to various health ~outcomes among Seventh- Day ttdven[ists in California. VII World Conference on Tobacco and Health 1990,316:(Absuaq). 13. Garland C„Barren-Connor E, Suarez L, Criqui MH, Wingard DL Effeots of passive smoking on ischemic heart disease mortality of nonsmokers: a prospecuve study. Am 1 Fnideminl 1985:121:645-650. 14. Humble C„Croft J', Gerber A.,Casper M, Hames CG: Tyroler HA: Passive smoking and 20-year cardiovascular disease mortality among nonsmoking wives, Evans County, Georgia. Amenca n, loumal of'Public Health, 1990;80:~599-601. 15. Lee P, Chaatberiain J, Ald'erson M: Relationship of passivc smoking to risk of lung cancer and other smoking-associated diseases. $ntish loumal of'Cancer 1986;54:97-105. 16. , Dobson A), Alexander HM! Heller RF, Lloy,d DM: Passive smoking and the risk of heart attack or coronary death. n at I991;154:793-797. 17. Weintraub WS: Cigarette smoking as a risk factor for coronary artery disease. In: Diana JI< ed. Tobacco Smoking and Atherosclerosis. Pathogenesis and Cellular Mechanisms. Nea• York: Plenum Press, 1990: 27, 37. 16. Lemer D, I+annel W: Patterns of coronary hean disease morbidity and mortality in the sexes: a 26-year follow-up of the Framingham population. American Hear ioumal 1956t111:383-390 19. O'Connor GT, Buring JE, Moore LL, Goldhaber SZ, Stampfer MJ, Willen WC, Hennekens CH:. Familv history of premature myocardial infarcuon and the risk of nonfatel myocardial infarcuon. Am I Epidemiol 1988:126(4):916, 20. Slack J, Evans h1t: The increased risk of death from ischaemic heart disease in first degree relatives of'121 men and 96 women with ischaemic hean disease. I Med'Genetica 19663:239-257:. 21. Nora JJ, Lonscher RH„Spangler RD, Nora AH, Kimberling u7; Geneuc-epidenvologic study of earlv.onset ischaemic heart disease. Circulanon 198Q,61:503-506. 22: Pekkanen J, Nusinen A, Puska P, Punsar S• I:arvonen MJ: Risk factors and 25 year risk of coronary , hean disease in a male population with a high incidence of the disease: the Firtnish cohorts of the seven countnes study. r, M 1989s299:81-85: 23. MacmahonS,PetoR,,CuderJ,CollinsR,SorlieP,NeatonJlAbbottR,GodR•inJ,DyerA,Stamler J: Blood pressure, stroke, and coronary heart disease. Pan 1, prolongcd difference in blood pressure:, prospective obsennuonal studies corrected for regression dilution bias. LIncei 1940;335`.765-774: 24. Morris JN, Everiit MG, PollardR, Chave SPW: Vigorous exercise in liesure time: protection agarrsst coronary heart disease. Lancet 1960:2:1207-1210:, 25. Morns )N;,Cla}ton DG, Eventt,MGi Semmence Ahil Burgess EH: Exercise in liessure time: coronary attack and death dates. Br Heart l 1990;63325-334.. 284 26. Karinel WB. Framinghan 27. Eaker ED; Pz Am li Cardi, . 28'Hopkuu P,\ Cardiol' Clln 29, Wynd'erEL Mtd 1987:1 :

0 risk faetors. Atherosclerosis preventablt by population idemiology; physiologyand' ry in nonsmokers living with .ary hean disease in women. and rudiorespintory health .299:423-427: sive smoking. In: Mizel] Mi, eriag Chemie intetnationalJ ve smoking in the multiple 1987:126:753-79>. outcomes among 5eventh- i Health 1990;316:(Absuact): fiffects of passive smoking ive study. Am I Enidemiol assive smoking and 20-year County, Georgia. *rt+erican i ) ofJung cancer and y¢, ,-105. and the risk oftiean attack -ry disease. In: Diana JN ed ir Mechanisms. New YO& and morraliry in the sexes: irt Iournal 1986;111:3833390. uillett WC, Hennekens GK ;k of nonfatel! myocardial can disease in first degree .l.ri1956:3:239•257. :netic-epidemiologic study 16. aaors and 25 year risk of' of the disease: the Firmisb God,bvin J, pyer A, 5t2ralcr onged difference in blood sion dilution bias. LUIW :n liesure time: protecuon Exercise in hessure tiID~ M Assessment of the Toxic Effects of Combustion Products 26. Kannel WB: Metabolic risk factors for coronary heart disease in women: perspective from the Framingham study: Am Hean I 1987;114(2):413-419. 27. Eaker ED, Packard B, WengerK, Clarkson TB, Tyroler HA: Coronary artery disease in women. Am I Cardiol 1988:61:641:~644:. 28. Hopkins PN, Witlisms RRldenaficuion and relative weight of cardiovascvlar risk factors. Cardiol lini c 1986:4:3•31. 29. Wynder El: Workshop on guidlines to the epidemiology of weak associations. Preventive !kd 1987;16:139-141L 285 I I

---