Document Q9jKqyMLzzQ3Y470zBqxr91k

Asbestos Exposure--Quantitative Assessment of Risk1-*3 JANET M. HUGHES and HANS WEILL4 Introduction Knowledge regarding the health effects of asbestos exposure has been emerging since early in the century, but the recent public concerns over the consequences of this naturally-occurring and useful mineral have been unprecedented. Anxi ety concerning these exposures escalated when these risks became associated not only with primary industrial contact (mining, milling, manufacturing), but also with less predictable and more numerous occupational exposures, such as among insulators and shipyard work ers, some with only indirect (though per haps heavy) exposures. Currently, there are two areas of major concern: (/) whether today's workplace exposure lev els, much lower than in the past, have, in fact, been lowered sufficiently; and (2) whether there is a potential risk in nonoccupational circumstances, such as from asbestos products inside public buildings. Although these concerns are wide spread, there has generally been little in formation available to the public, ad ministrators, and concerned profes sionals regarding how great the potential risks might be in various circumstances. In fact, the dose-relatedness of the risk is often not well understood, and many have concluded that all exposures may be equivalent to those experienced by heavily exposed workers several decades ago, whose subsequent health effects have justifiably received considerable public attention. Exposure to asbestos results from many sources, both from the natural en vironment and from commercial asbestos use, and it has been noted that asbestos fibers can be found in the lungs of al most everyone in the population (1). This fact does not indicate that such low ex posures as occur in the general popula tion are necessarily without risk, but it does emphasize that there are many sources of asbestos exposure and that any particular exposure source (e.g., products inside buildings), while contributing to the overall exposure of individuals, may well not be the sole or even primary source of exposure. In fact, asbestos ex SUMMARY Method! tor deriving quantitative Mtimataa of aobaatoa-aaeociatad haaith date am reviewed and thoir numerous aaaumpttona and uncartaintiaa described. These methods bwetva extrapolation of risks otiaarvad at past relatively high asbestos eoncantration lavata down to uauatty much Ibwsr concentration levels of Intsrsettoday-tosome cases, orders of magnitude imror. These models am uaad to calculate aatimataa of tt>a potantiai dak to ewrkers manufacturing aabeetoa products and to students anretlad In aetioot* containing asbestos products. The potential dak to workers exposed tar 40 yr to 03 fibers per milliliter (t/ml) of mixed aebestoa fiber type (a permissible workplace exposure limit under consideration by the Occupational Safety and Haaith Administra tion (OSHA)) am estimated as S2 lifetime excess cancers per 10,000 exposed. The risk to students exposed to an average asbestos concentration of 0JW1 t/ml of mixed asbestos fiber types tor an average enrollment period oft school yearn is estimated as 5 llfotimo excess cancers parone million exposed. If the school exposure ie to chryeotile asbestos only, then the estimated risk la 1.5 lifetime excess cancers par million. Risks from other causes art presented tor comparison; e.g., annuli rates (per million) of 10 deaths from high school football, 14 from bicycling (10-14 yr of age), 5 to 20 for whooping cough vaccination. Decisions concerning asbestos products require participation of all parties involved and should only be made after a scientifically defensible estimate of the as sociated risk has been obtained. In many cases to date, such decisions have been made without adequate consideration of the level of risk or the com effectiveness of attempts to lower the poten- tial risk. ah nev nesaw on teas; posure levels in the ambient air, especially in urban areas, can be higher than within buildings containing asbestos products, especially if these products are wellmaintained. While there has been debate in many countries about the potential health risks from both occupational and nonoccupational asbestos exposures, much of the recent concern in the U.S. has focused on exposure in schools. The Environmen tal Protection Agency (EPA) directed all U.S. schools to test for asbestos-con taining products, and if asbestos was found, to notify parents and school em ployees. To date, however, EPA has provided these schools with neither guid ance as to what further action should be taken if asbestos materials are found, nor estimates of the magnitude of the poten tial risk associated with these products. Inevitably, some parents and school ad ministrators have concluded that these materials pose a high cancer risk and have taken action consistent with this conclu sion. Others have postponed action. It is likely that in some instances, asbestos products in excellent condition, emitting few or no fibers, have been removed un necessarily, while in others, materials emitting fibers at an unacceptably high level have not received attention. This report will derive estimates of the potential risk from asbestos exposure in two situations: the workplace and schools. In deriving these estimates, im portant factors and issues in asbestos risk estimation will be discussed, as will the assumptions and uncertainties of the methodology. A summary of risk esti mates derived by several governmental advisory groups (2-8) will be presented for comparison purposes and to illustrate the important effect of various factors on the resulting risk values. The findings of these advisory groups, although likely to influence governmental decision making with respect to asbestos and its future use, have, in general, not reached (Received in original form June 14, 1985 and in revised form October H,'1985) 1 From TUlane University School of Medicine, Department of Medicine, Pulmonary Diseases Sec tion, and the School of Public Health, Department of Biostatistics and Epidemiology, New Orleans, Louisiana. * Supported in part by Specialized Center of Re search (SCOR) Grant No. HL-15092 from the Na tional Heart, Lung and Blood Institute. 1 Requests for reprints should be addressed to Dr. Janet M. Hughes, Titlane University School of Medicine, Pulmonary Diseases Section, 1700 Per dido St., New Orleans, LA 70112. 4 Wbrk partially performed while a Fellow in Science and Public Policy, The Brookings institu tion, Washington, DC. 6113 22385 5  huohq mo wbll a wide scientific audience, nor the con cerned public. Moreover, the risk esti mates developed by these groups have been based on widely differing assump tions concerning the level and duration of the asbestos exposure, making com parisons difficult. In this report, the es timates of these groups will be adjusted to common exposure assumptions in or der to allow direct comparisons; differ ences in the risk values will emphasize important differences in the methodolo gies and assumptions of these advisory groups. Methods Most studies of the effects of asbestos ex posure have been of workers exposed to as bestos in industry, most employed many years ago. These asbestos concentration levels were considerably higher than in industry today, and several orders of magnitude higher than environmental exposures. Estimates of risk, especially for environmental asbestos ex posures, are therefore not based on direct ob servation of these risks, but instead, on ex trapolation of risk observed at higher levels down to these lower levels. Estimates of potential asbestos risk will be derived for lung cancer and mesothelioma, two causes of death repeatedly shown to be elevated among past asbestos workers. Some studies of asbestos workers have observed an increased risk of cancer of other sites; e.g., larynx, gastrointestinal tract and kidney (9). These findings, however, have not been con sistent and there is considerable controversy as to whether these cancers are associated with asbestos exposure. There is general agreement, however, that any potential excess cancer risk at these sites combined would be relatively small - less than lCWo of the excess lung can cer risk. Asbestosis, when it occurs as the result of relatively recent industrial exposure levels, is generally very slow in progressing and does not now often result in serious disability or in increased mortality. There is no evidence of asbestosis occurring as a result ofenviron mental exposures. Risk of this disease, there fore, will not-be considered. Results Lung Cancer Risk The data from several studies suggest that lung cancer risk, as measured by the Stan dardized Mortality Ratio (SMR, the ra tio of the observed number of cases [O] to the number expected [E] in the absence of asbestos exposure, multiplied by 100; ie., SMR = 100 [O/E]) is linearly related to cumulative asbestos exposure. Because the SMR should be 100 (no excess) when there is no asbestos exposure, this linear model may be written as: SMR = 100 + bx (Equation l) where b is the slope of the line and x is the cumulative asbestos exposure (usu ally recorded in fibers per milliliter years (f/ml - yr). For example, if b = 2 and x = 10 f/ml - yr, then SMR = 120; ie., the observed number of cases is 120% of that expected. Because SMR = 100 (O/E), Equation 1 may be rewritten in terms of the num ber of excess cases (O - E) occurring: O - E = E (b/100) x (Equation 2). It is this last equation that will be used to estimate the potential number of ex cess cases for a specified cumulative ex posure x. Estimates of the slope b may be ob tained from epidemiologic studies that have demonstrated a dose-response rela tionship (by calculating risk for popula tion subgroups defined on the basis of cumulative exposure level). Demonstra tion of increasing lung cancer risk with increasing estimated exposure is an im portant requirement for an epidemiologic study because it allows critical examina tion of its findings. If observed risk in a population shows no sensible pattern with increasing exposure, then the valid ity of the exposure estimates and/or the risk estimates must be questioned; if risk is substantially elevated at very low ex posure levels, then systematic underesti mation of exposure or overestimation of risk must be considered. Errors in risk estimation can occur as a result of many factors, including incom plete trace of the population, inaccurate ascertainment of causes of death, and use of an inappropriate comparison popu lation for the expected mortality rates. Accurately estimating past exposure lev els for workers is always difficult because only limited past exposure measurements may be available; and job records may not fully reflect work activities. Because of these potential problems, demonstra tion of a gradient of risk with estimated exposure is essential because it provides at least partial validation of the exposure estimates. There have been 7 industrial cohorts for which a reasonable pattern of risk has been observed for categories of in creasing cumulative asbestos exposure (10-16). Estimated slopes from the 7 co horts vary considerably but exhibit a rela tionship with industrial process: the lowest values are found in studies of miners and friction product manufactur ing workers, higher values Cor asbestos cement and other manufacturing em ployees, and the highest among textile workers (figure 1). These results indicate that for the same estimated cumulative asbestos exposure, textile workers have the highest potential lung cancer risk, while miners and friction products work ers have the lowest. It is believed that, because of the differing state and physi cal treatment of the asbestos in these different processes, the dust clouds con tain asbestos fibers of differing physical dimension, which is related to the level of carcinogenicity. Specifically, exposures in textile manufacturing are more likely to include long, thin fibers than in min ing and other manufacturing. Moreover, animal experiments have demonstrated that long, thin fibers are more pathogenic than short, coarse fibers (17, 18). Uncertainties of the Lung Cancer Model The linear model for lung cancer risk at low exposure levels is based on several assumptions: that the linearity of risk ob served in industrial cohorts applies to all exposure levels (even very low),, that the risk for a cumulative exposure is indepen dent of smoking and remains constant after an initial lag or latency period (usu ally 10 years), and that risk is linear with respect to both components of cumula tive exposure, duration and concentration. It is doubtful if the shape of the doseresponse relationship for low amounts of asbestos exposure can ever be determined by direct observation because of the large number of non-asbestos-related lung cancers in the population. However, the linear model is consistent with observa tions at higher exposure levels and is sup ported by the mathematical theory of CumutotW* Asbastoa Expoaua (f/cc-yrs.) Fig. 1. Estimated lung cancar dosa^asponsa relation ships from 7 epidemiologic studias (Ik taxtilss. Ma manufacturing. Ca asbastos camant, Mn - min ing, Fp - friction products). 6113 22386 MSITM CVOCUW-OUkNTraTtVf ASSESSNBTT OF WSK 7 multistage carcinogenicity for low doses of a carcinogen (19). Moreover, there is general agreement that risk is unlikely to increase more rapidly than linearly at low cumulative asbestos levels. The assumption that asbestos lung cancer risk (as measured by the SMR) is independent of smoking is equivalent to assuming that the effect of asbestos is the same among smokers as among nonsmokers, i.e., that asbestos exposure increases the lung cancer risk among smokers and nonsmokers by the same percentage. Contrary to this assumption, some of the available data suggest, though not conclusively, that the percent age increase among smokers may be somewhat less than among nonsmokers (20, 21). Similarly, there is limited evi dence that the assumption of a constant relative risk (observed number/expected number) over time may not be valid: data from North American insulation work ers (22) exhibit a decline in the relative risk for lung cancer after 35 years from initial exposure. Both these issues remain controversial and cannot yet be resolved from the available evidence. Although there are some data (10,15) to support the assumption that risk is linearly related to both concentration and duration of exposure, the evidence to date is quite limited. In spite of these uncertainties, the as sumptions of the lung cancer model are conservative in the sense of possibly lead ing to an overestimate of potential risk but, in all likelihood, not underestimat ing risk. Having accepted the model, however, there remain considerable un certainties concerning the estimate of the slope b most appropriate for an exposure situation; therefore, a range of slopes is often considered when assessing risk in an exposed population. Mesothelioma Risk Mesothelioma, a primary pleural or peri toneal malignancy, is a rare disease, ev6n among asbestos workers (a total of ap proximately 1,500 cases in the U.S. an nually, compared with approximately 130,000 lung cancer cases). Consequent ly, in comparison with lung cancer, there is much less quantitative information for mesothelioma for establishing a doseresponse relationship. As a result of the very low incidence of this disease in the general population and conclusive evidence of an increased risk among asbestos workers (the only known occupational cause), any case oc curring in a person with an industrial as bestos exposure is assumed to be a result TABLE 1 MESOTHELIOMA IN POPULATIONS EXPOSED TO DIFFERENT FIBER TYPES Fiber Type Chrysotile (11. 15. 18. 27-30)t Amosite (31. 32) Crocidolite (33-35) Mixed fiber (12. 14. 22. 29. 36-40) Total Oeeths Obeerved Lung Cancer ------------------------------------ Observed Number Excess' Mescxnelioma ------------------------------ Observed As 4* of Excess Number Lung Cancer 5.500 407 99.0 12 12.1 861 144 85.5 19 22.2 735 80 31.5 52 165.1 6.751 960 465.3 320 65.9 * Studies comOined. For eacn study, lictu lung cancan odaaivad numear minus numoar aipactad eased on compamon population. __ T References in paranllmss. of that exposure. However, virtually all studies of mesothelioma cases have in cluded persons for whom no known as bestos exposure could be identified. Moreover, a substantial number of cases have been reported from rural Tkirkey (23) (attributed to the indigenous occurrence of another fibrous mineral, erionite, a form of the mineral zeolite (24)), and an imal studies have demonstrated that fibers other than asbestos can induce mesothelioma in animals (25, 26). For these reasons, it is concluded that there are background cases of mesothelioma in the general population; i.e., cases with no relationship to asbestos exposure. Of the several minerals commonly re ferred to as asbestos (chrysotile and the amphiboles) crocidolite and amosite, have accounted for almost all of the com mercially used asbestos. There is strong qualitative evidence of a differential in mesothelioma risk by type of asbestos fiber exposure, with crocidolite, and probably amosite, posing greater risk than chrysotile. (This difference has been recognized by most industrialized nations by setting different control limits for these fibers.) Across-study comparisons of the observed numbers of mesotheli omas cannot be made directly because differences would be partially due to differences in the sizes and exposure lev els of the cohorts studied. However, the number of excess lung cancers in a study should also be related to these factors, so this number can be used, in effect, as a measure of these variables for a study. Therefore, by considering the ratio of the number of mesotheliomas to the num ber of excess lung cancers, an approxi mate adjustment for exposure level and size differences of the study populations can be made. As shown in table 1, the number of mesotheliomas was 12% of the excess lung cancers among chrysotileexposed populations, but 165% for crocidolite exposure. Populations ex posed to a mixture of fibers exhibited in termediate results (66%). These data sug gest that chrysotile exposure poses ap proximately one fifth the risk of a mixed fiber exposure (12 versus 66%). This pat tern was observed for both pleural and peritoneal mesothelioma (see (5) for sep arate results), although the evidence is more striking for peritoneal tumors. Similar results have been reported by two case-control studies (41, 42), which found cases to have had greater exposure to crocidolite than noncases, but no differences in chrysotile exposure: Tis sue analysis studies (43, 44) have sup ported these findings, with greater num bers of crocidolite fibers but similar num bers of chrysotile fibers found in the lungs of mesothelioma cases compared with noncases. A recent study (45) of per sons exposed to chrysotile asbestos in mining (contaminated with the amphibole tremolitc) found, on average, a greater number of chrysotile fibers in the lungs of mesothelioma cases than in matched controls. These findings are consistent with the concept of doserelatedness of risk with chrysotile ex posure or its contaminant, which showed an even larger gradient. It cannot, how ever, address the question of a difference in risk between chrysotile and crocidolite/amosite exposures because neither crocidolite nor amosite was present. Concerning the question of whether mesothelioma risk is related to industrial process, results of the three crocidolite exposure studies are suggestive of a difference by process: mesothelioma as a percentage of excess lung cancer risk was 120% among crocidolite miners (33) but 225 and 298% among crocidolite gas 6113 22387 I HUQNQ AMO *tA^ mask manufacturing workers (34, 35). TABLE 2 However, other study differences (e.g., ESTIMATES OF THE MESOTHELIOMA MOOEL PARAMETER K FROM length of follow-up) might account for 7 EPIDEMIOLOGIC STUDIES these findings, and similar comparisons of populations exposed to other fiber types are difficult. With only twelve Industry Maaothaiioma Casas Maan Duration (yd Mean Concentration Kx 10* mesothelioma cases among the chrysotile studies (table 1), no meaningful com Mixad Fibtra In (22) 180 25' 15* 0.29 parison by process can be made for chrysotile; the two amosite populations were both involved in manufacturing; and within the mixed fiber studies, vary ing fiber type mixtures could account for observed differences in risk. It must therefore be concluded that the data cur rently available are not adequate to de termine if there is a difference in Tx (48) Ca(*9> Ct (50) AmostM Ma (31) Chrysotil* Mn (SI) Tx (52)t 25* 7 25 17 10* 84 30* 30 9* 11 14 15 35* 1 31* 1 10 40* 7.5 0.15 0.11 2.18 05 u 0.006 0.07 mesothelioma risk by industrial process, after accounting for fiber type. Because there is no reliable quantita tive dose-response information by which Otfrvfte 0/ atorwMiooflr Mn mining, Fp fnctton products. Mt manufacturing, <3m gas mask manufacturing, Ca camant products. In * insulators. Tx > tsxtiia manufacturing. Sh - shipyard. * Exposure reformation net avaiiapfa/raponad for cohort; astimatad from bast availabta sourca. * Oataifad data presantad m reftranc# 53. mesothelioma risk may be related to as bestos exposure, mesothelioma modeling is based on the observation among as bestos workers that incidence rates in crease as a power of time since initial ex posure. Peto and associates (46), using data from the study of insulators (22), found that incidence was approximately proportional to the 3.2nd power of time since initial exposure, irrespective of age at initial exposure. Using this observa time since initial exposure to a carcino gen; e.g., lung cancer and exposure to both smoking and asbestos. Because of this, the question may be raised of why the approaches to modeling lung cancer and mesothelioma risks differ substan tially (modeling lung cancer relative risk but mesothelioma incidence). In estimat ing asbestos-associated lung cancer risk, sulators (22) and British textile workers (48) were 15 to 20 times the rate in U.S. asbestos cement workers (50). However, according to the model, the differences in exposures (assuming concentrations of 30 versus 11 f/ml and durations 25 versus 4 yr) would result in incidence rates differing by a factor of 9 or more. Thus, if differences in exposures are taken into tion, together with a mathematical model of carcinogenesis (47) and the assump tion of a linear relationship between in cidence and exposure concentration, the age-specific background rates (to calcu late an expected number E) are available so that relative risk o.f lung cancer may be considered. However, because of the account, rates in these 3 populations are relatively similar. However, rates in the Canadian asbestos cement workers' study (49) were approximately triple those in following model is used for relating the small number of mesothelioma cases, the insulators' and textile workers' mesothelioma incidence rate I(t) at time background rates for mesothelioma are studies, even though the assumed ex t since initial exposure to mean asbestos not available and, therefore, relative risk posures were lower. These differences in concentration c and duration d: cannot be investigated. rates are reflected in the estimated values .... _ K c , 0 <t <d w " K c(t2 2 - (t-d)22 ), t > d (Equation 3). There have been 7 studies (table 2) for which sufficient information is available to estimate the parameter K of the meso thelioma model (Equation 3). Most of of the parameter K from these studies, with that based on the Canadian study (49) considerably higher than those from the other mixed fiber studies. Some of Thus, during exposure (0 < t < d), inci these 7 studies do not meet minimal the differences in incidence rates in these dence rises as a power of length of ex criteria concerning availability and vali 4 studies may be due to differences in the posure; after exposure ceases (t > d), in dation of exposure information. For ex mixture of fibers, but little quantitative cidence rises less steeply. ample, the study of insulators (22), with information on this point is available. Using this model and an estimate of by far the largest number of mesotheli The observed and fitted mesothelioma the parameter K, ordinary lifetable methods are used to predict the total number of lifetime mesothelioma cases occurring in a cohort exposed to asbestos for d years at a concentration of c f/ml. A slightly different model incorporates a latency period for disease development by replacing elapsed time since first ex posure (t) with elapsed time minus 10 yr (t - 10); however, there are only mini mal differences between the two models in estimated lifelong risk. As for mesothelioma, incidence rates of many cancers are power functions of oma cases, had neither job histories nor exposure measurements available. (See the Consumer Product Safety Commis sion (CPSC) report (2) for a more detailed review of these studies.) With these limitations in mind, and using the best available estimates of the mean du ration and concentration of exposure for each cohort, estimates of K may be de rived for each study (table 2). Observed mesothelioma incidence rates in the 4 cohorts exposed to mixed fibers varied considerably; at 32 yr fol lowing initial exposure, rates among in incidence rates for 3 of these studies (fig ure 2) show that the model provides a rea sonable fit to the observed data. (Data for the fourth mixed fiber study (50), omitted from figure 2 because the lower rates require a smaller scale, are also con sistent with the model.) For exposure to mixed asbestos fibers, the best available estimate for the param eter K of the model will be taken to be 0.2(10"*), approximately the median of the 4 estimates (table 2). To reflect the uncertainties of the value K, values rang ing from 0.1 (10**) to 0.4 (10"*) will also 6113 22388 txpo*w*-QUAirmxTivi amosiicnt or mi 10 20 30 40 50 Years Since Initial Exposure Fig. 2. Observed and fitted annual mesothelioma inctdanca ratas by years since initial exposure m 3 popula tions with mixed asbestos fiberexposure (C asbestos cement. In insulation, Tx > textile). be considered. Data for chrysotile ex posures (table 2) suggest a value for K of 0.04(10"*), or one fifth the value for a mixed fiber exposure. This is consis tent with the evidence in table 1 that chrysotile exposure may pose one fifth the risk of a mixed fiber exposure. A rea sonable range for the chrysotile parame ter will be taken as 0.02 (10**) to 0.08 (10-). Because rates are assumed to increase with time since initial exposure and to be independent of age, the estimated number of lifetime cases among persons exposed early in life will be greater than for those exposed late in life. Thus, in estimating lifelong risk in an exposed population it is necessary to specify the age at initial exposure for the cohort. application of the model to a population initially exposed at 25 yr of age finds that doubling exposure duration from 10 to 20 years multiplies predicted risk by 1.4 (not by 2.0). However, for short dura tions, the model is approximately linear; e.g., doubling exposure from 1 to 2 yr results in approximately a doubling of estimated risk. As with lung cancer, there is limited evidence that mesothelioma risk after many years following initial exposure rises less steeply than would be estimated by the model. If this evidence is valid, then the model could substantially over estimate lifetime mesothelioma risk. Industrial Asbestos Exposure While keeping in mind the stated uncer tainties of the models, they are useful for obtaining an approximate estimate of the magnitude of asbestos-associated risk. As an example, an industrial cohort of asbestos cement manufacturing work ers who began employment at 25 yr of age will be considered. Of the studies rep resented in figure 1, the 2 studies with intermediate slopes (asbestos cement and mixed product manufacturing workers) were those concerned with the most simi lar industrial processes. The slopes for these studies were approximately 0.5, which will be taken as the value of the slope for lung cancer risk for this exam ple. To reflect uncertainties in the esti mated slope, slopes ranging from 0.25 to 1.0 will also be considered. Based on current U.S. rates, the prob ability of a 25-yr-old man eventually dy ing of lung cancer is approximately 0.051. Thus, in a cohort of 10,000, approxi mately 510 lung cancer deaths would be expected (E = 510 in Equation 2). If the cohort is exposed to 0.5 f/ml (currently being considered by the U.S. Occupa tional Safety and Health Administration (OSHA) as the permissible exposure limit) for 20 yr, then cumulative exposure (x) is 10 f/ml - yr. Using Equation 2 and a slope (b) of 0.5, the lung cancer model estimates a total of 25.5 (or 26) lung cancers attributable to the asbestos exposure (in addition to the 510 back ground cases). Doubling the exposure du ration would double the lung cancer risk to 51 excess cases. The estimated num ber of lung cancer cases would be the same if the exposure is assumed to be to chrysotile only. See table 3. For mesothelioma, the model (Equa tion 3) estimates a total of 27 lifetime mesotheliomas (using life table methods and 1980 U.S. mortality rates) if exposure to mixed asbestos fibers lasts for 20 yr; this estimate becomes 31 cases if the ex posure continues for 40 yr (table 3). If these exposures are to chrysotile alone, then the estimated mesothelioma risk is reduced by a factor of 5. Thus, the estimated total number of cancers attributed to 20 yr of exposure at 0.5 f/ml would be 53 cases for a mixed fiber exposure and 31 cases for a chryso tile exposure: Reducing the assumed as bestos exposure concentration level by a factor of 2 would similarly halve the es timated risk. Such calculations may be performed for other industrial cohorts, with the lung cancer slope chosen appropriately. In the Uncertainties of the Mesothelioma Model The model for estimating the mesotheli oma risk associated with asbestos ex posure is based on many more assump tions and considerably less quantitative evidence than the model for lung cancer risk. Mesothelioma incidence as a power of time since initial exposure is consis tent with other cancers, and the proposed model provides a reasonable fit to the available data sets. There is no quantita tive evidence, however, that risk is linearly related to exposure concentration or that exposure duration should be factored into the model as in Equation 3 (a result of mathematical modeling of multistage carcinogenicity). It should be noted that, in contrast to the model for lung cancer, the mesothelioma model does not assume linearity of risk with exposure duration; i.e., doubling the exposure duration does not necessarily double the risk. In fact, TABLE 3 LIFETIME RISK ESTIMATES FOR AN INDUSTRIAL COHORT (MALES) OF ASBESTOS CEMENT MANUFACTURING WORKERS EXPOSED FROM 25 YR OF AGE TO AN AVERAGE ASBESTOS CONCENTRATION OF 0.5 F/ML. BY DURATION AND FIBER TYPE Duration (yr) Estimated Attributable Cases per 10* Exposed Lung Cancer' Meaotheiiomat Total MixaO Fiber Exposure 20 26 (13 to 51)4 40 51 (26 to 102) Chrysotile Exposure Only 20 26 (13 to 51) 40 51 (26 to 102) 27 (13 to 54)* 31 (16 to 62) 5 (3 to 11) 6 (3 to 12) 53 (26 to 105) 82 (42 to 164) 31 (16 to 62) 57 (29 to 114) * Baaso on an assumed c< 0.5 (ineienee in the lung cancer Standardized MortaKy Ratio per uni incraaaa m cumubdve T Baaed on an aaaumad exposure Vti* - yr)of 0.2 (10*) tor Ilia modal paramatar K; aaa lan. * Range d asomatee. baaed on a ranga al atopaa 025 to 1.0. * Range ol eeUmatoe. on a ranga ol paramaiata 0.1 (10*) to 0.4 (10*) tar mixed fiber ax- poauraa and on* im< ones lor chryaodla exposures. 6113 22389 to huohcs aho *tu TABLE 4 expected to die of lung cancer based on ESTIMATED NUMBER OF LIFETIME DEATHS ATTRIBUTABLE TO 8 YR recent U.S. rates; ii, these would not be OF ASBESTOS EXPOSURE IN SCHOOLS (PER 10* EXPOSED), BY FIBER TYPE AND AVERAGE CONCENTRATION attributable to the school exposure to as bestos (more than 90ft due to smoking). Lung Cancer* Meeotfteiiomet Total For estimating background mesothe 0.001 f/ml* Mixed fibers 0.6 (0.3 1.2)4 4.4 (22 to 8.8)* 5.0 (2.S to 10.0) lioma cases, Peto and associates (61), using data on Los Angeles residents, found that mesothelioma incidence rates Chrysotile only 0.8 (0.3 to 1.2) 0.9 (0.4 to 1.8) 1.5 (0.7 to 3.0) for persons with no known asbestos ex posure were approximately proportional 0.003 f/ml* Mixed fibers Chrysotile only 1.9 (0.9 to 3.7) 1.9 (0.9 to 3.7) 13.2 (6.6 to 28.4) 2.6 (1.3 to 5.3) 15.1 (7.5 to 30.1) 4.5 (2.2 to 9.0) to the 3.5 power of age. Checking the va lidity of this background model by ap plying it (with a coefficient of 2.3(10*11]) to the 1980 U.S. population, this model ' Stud on an aaaumed slop* of 0.5: m footnof*. I*M* 3. t Basso on an assumed value of 0.2 (10*) for tft* modM parameter K: as* tan. 1 Rang* of estimates. based on a range of slope* 0.2S Bio. Rang* of astimatea. baaed on a rang* of model parameters: see footnote, tab!* 3. estimates a total of 385 mesothelioma cases during 1980 among persons with out known asbestos exposure. This would be an annual rate of approximately 1.7 cases per million, which is in agreement calculations, current lifelong risk of lung cancer was assumed to be valid for the cohort; in fact, there is evidence of a de crease in male lung cancer risk, both in the U.S. (54) and in Great Britain (55), primarily among those in the younger age groups (believed to be a result of the de crease in tobacco use). If this trend con tinues, then the number of background Students will be enrolled in schools containing asbestos products for a vary ing number of years, starting at various ages. On average, they will be assumed to be enrolled for 6 school years, at a mean initial age of 9 yr. Because a stu dent spends approximately 36 wk per yr, 35 h per wk, at school, a school year is equivalent to 0.656 of a work year (48 with an estimate of 2.0 made recently by McDonald (62) using a different ap proach. Applying this model to a birth cohort of one milion, a total of 140 life time background cases are estimated. The estimated numbers of cases in the as bestos exposed students would be in ad dition to these background cases. Other Known Risks cases (E of Equation 2) should be lower, wk per yr, 40 h per wk). Therefore, the For school asbestos exposure, the upper resulting in a correspondingly lower es exposure accumulated over a school year estimate of approximately 15 lifetime ex timate of the excess cancers attributable at 0.001 f/ml would be equivalent to that cess deaths (mixed fibers at 0.003 f/ml) to asbestos exposure. School Asbestos Exposure accumulated during a work year at would constitute an average annual rate 0.000656 f/ml. of approximately 0.25 deaths per million Because the exposures of asbestos ce exposed. To place this risk in perspective, In its recently released report (6), the ment and muted products manufactur some other estimated and observed an Ontario Royal Commission presented a ing workers were in large part to dust of nual death rates per million in the U.S. complete review of the data (56-60) con building products, these exposures may are: approximately 1,200 for long-term cerning asbestos exposure levels inside be assumed to be qualitatively similar to smoking (63), 15 for bicycling (10 to 14 buildings containing friable asbestos. Af exposures inside buildings. Therefore, a yr of age (64)), 15 for inhalation/ inges ter reviewing and summarizing these slope of 0.5 will again be used as the best tion of foreign objects (65), 1 for living data, the Commission concluded that estimate of the slope in the lung cancer for 2 months with a cigarette smoker (66), "the majority of exposures of building model (with a range of 0.25 to 1.0). This and 10 from playing high school foot occupants in buildings with substantial may be an overestimate of the slope be ball (1970 to 1980) (67). Although not amounts of friable asbestos would be to cause the dimensional distribution of the converted to rates, school bus accidents fiber levels less than 0.001 f/ml, with a fibers emitted from finished products resulted in approximately 200 deaths in few single readings as high as 0.01 f/mi may be skewed more to shorter, coarser 1975 (68) and whooping cough vaccina representing the highest likely exposure" fibers than fibers used in manufacturing tion, 5 to 20 deaths annually during 1970 (reference 6, p. 577). these products (including raw fibers). to 1980 (69). By contrast, for school as Based on this report, the best estimate Among a cohort of one million school- bestos exposure, the upper estimate of of the mean fiber concentration in U.S. children exposed to 0.001 f/mi of mixed an average annual rate of 0.25 deaths per schools containing asbestos products will fibers, the models estimate 0.6 lifetime million exposed would be equivalent to be assumed to be 0.001 f/ml. To take into lung cancers and 4.4 lifetime mesotheli 0.75 annual deaths, based on EPA's esti account the possibility that some stu omas (table 4), for a total of 5 lifetime mate of 3 million students currently ex dents could be exposed to levels as high excess deaths. If exposure is to chryso- posed (60). as 0.01 f/ml occasionally, an upper esti mate of the possible mean exposure lev els will be taken as 0.003 f/ml (the geo metric mean of 0.001 and 0.01 f/ml). For each concentration level, it is assumed that students are exposed to this concen tration of asbestos, on average, through out every school day, for their entire length of enrollment in the school. tile only (the fiber type used in acousti cal plaster of ceilings), then the estimated mesothelioma cases would be 0.9 and the total cancers 1.5. If an asbestos concen tration of 0.003 f/ml is assumed, then these estimates triple (table 4). By contrast, in this population of one million students (halfmale, half female), approximately 32,000 (3.2ft) would be Other Estimates ofAsbestos-Associated Risk Several advisory groups (primarily gov ernmental) have derived quantitative es timates of asbestos-associated risk of lung cancer and mesothelioma under var ious exposure assumptions (concentra tion, duration, h/wk). In order to com pare these estimates, they have been ad- 6113 22390 AMCgTQ* tXMtUW-OUAMTItKnVt AMCSSttCNT OF 11 justed (where possible) to a cohort of 10,000 workers exposed starting at 20 yr of age, for 20 yr (for 40 h/wk) to 0.5 f/ml of mixed asbestos (table 5). (Where esti mates were reported for smokers and nonsmokers separately, they have been combined in table 5 assuming half smokers.) Because there was fundamental agree ment among these reports in the general methodology, there is reasonably good agreement in the risk estimates. Never theless, important variability in the esti mates remains. These differences are pri marily the result of variability in the values assumed for the model param eters. Similarly, each reported range of risks reflects a corresponding range of parameter values. For a particular ex posure situation of interest, the range for lung cancer risk can be narrowed con siderably based on the appropriate indus trial process. The importance of the model parameters in estimating risk emphasizes the need for critical review of the epidemiologic studies used for estimat ing these parameters. This is especially true in modeling mesothelioma risk, be cause the studies with the highest esti mated parameter values (31, 49) have potentially important limitations in their exposure estimates (table 2). In the study of Canadian asbestos cement workers (49), which yields the highest estimate for the model parameter K, workers in the highest exposure category were found to have the lowest lung cancer risk, bring ing into question the validity of the ex posure estimates. The study of amosite workers (31) had no exposure measure ments performed at the work site. Be cause the value of the parameter K esti mated from a study is directly related to the assumed exposure for that cohort, er rors in exposure estimates would be cor respondingly reflected as errors in K. Differences in estimated background risk for lung cancer also account for some of the variability in the estimates of lung cancer risk in these reports. This is espe cially clear for the estimates from the CPSC and EPA reports in which the lung cancer risk estimates differ by a factor of 2, even though they used the same range of slopes for the model. Discussion Because of a general awareness that as bestos is a human carcinogen, the public has become increasingly concerned about the use of asbestos in our society and is often alarmed to learn of asbestos ex TABLE S ASBESTOS RISK ESTIMATES BASED ON VARIOUS REPORTS FOR A COHORT OF 10.000 MALES EXPOSED FOR 20 YR IN THE WORKPLACE TO 0.5 F/ML OF MIXED ASBESTOS FIBERS FROM 20 YR OF AGE Lung Cancar Rpo<t Slop* b Excwa Death* OMtht CPSC (2) EPA (4) HSEt (7,8) NRC* (3) OSHAl This report 0.3 to 3.0 0.3 to 3.0 0.04 to 5.3 2.0 1.0 0.05 to 2.0 10 to 101 21 to 212 2 to 250 120 89 3 to 102 15 to 154 16 to 159 NA 105 52 38* (19 to 76) Dthntnn ot aMxmwtfont: CPSC - Coneumei Product Safety Cawnaicn: EPA Environmental Protactio.r Agency; HSE Health and Safety Executive; MAC - National Reteeren Council. ' Sacauaa ot miner dWNtancaa m the maaonraiioma model*. tna aatimtfad parameters K of these modal* ar* not comparable and tfrerefote are not repotted hate. T Eatimatee baaed on tafifee 38 and 38 ot 1919 report and table 7 ot 1963 report. _ * Meeotnalioma eetimate incorreetty caiculafed in report recalculated here using NRCs eanmate of K. I Beat eatimate and range of etttmatea: see footnote, table 3. I Pan of OSHA's 1983 rule-making procedure for a revaed aabeatoa standard: derived by same con tractor aa EPA eatimatee. posures in some work situations (e.g., brake lining repair) or of the presence of asbestos-containing products in public buildings, especially schools. Generally, however, little information has been available to workers, consumers, parents, or school administrators to assist them in estimating the likely magnitude of the potential risk from other occupational or nonoccupational exposures. Some governmental agencies con cerned with the issues of asbestos use have sought scientific input into asbestos- associated risk estimation as part of their decision-making process. While making valuable contributions, much of this work would likely have been improved by expanded use of the scientific review process. Moreover, these risk estimates have, for the most part, not reached the general public or the medical and health professionals who are often asked for ad vice in these issues. Risk estimates such as appear in ta bles 3 and 4 should provide a basis for rational decision-making regarding as bestos, with input from all concerned parties. Such estimates may be used to identify exposure levels for which the as sociated risk is deemed "acceptable" by those exposed, and levels deemed unac ceptable. In making these judgments, these risks may be placed in context by comparison with other risks in our society. With respect to asbestos products in the schools, such a rational approach could be used to determine when removal is called for and when proper main tenance of these products is preferable to removal. In fact, some experts have argued that removal of building materials in good condition could pose greater risk than proper maintenance of these prod ucts because of the likelihood of in creased exposure levels during and fol lowing removal. The process of risk assessment has be come increasingly accepted as a neces sary base on which to base rational decision-making. However, just as it should not be denigrated as unhelpful be cause of its inevitable limitations, neither should it be oversold as a panacea. There is evidence that the public now has more limited (and realistic) expectations of science than it once did, while continu ing to hold a favorable view of the value of scientific endeavor (70). This situation provides an opportunity for an improved dialogue between scientists and the pub lic on subjects of common concern, and for scientists to play an important role in shaping public policy that is scientifi cally defensible. References 1. Churg A. Current issues in the pathologic and mineiaiogic diagnosis ofasbestos-induced disease; Chest 1983; 84(3)^75-80. 2. Consumer Product Safety Commission. Report of Use Chronic Hazard Advisory Panel on Asbestos. Washington, DC,, 1983. 3. National Research Council. Committee on Nonoccupational Health Risks. Asbestiform fibers--nonoccupational health risks. Washington, DC: National Academy Press, 1984. 4. Environmental Protection Agency. Asbestos health assessment update; Prepared by Nicholson W. Washington DC: U.S. Environmental Protec tion Agency, 1984. 6113 22391 12 MMMCS AMO WSU. J. Department of National Health and Welfare. Canada. Report of the committee of experts con cerning the scientific basis for occupational stan dards for asbestos. OttawaCanada. 1984. 6. Royal Commission on Matters of Health and Safety, Canada. Report on matters of health and safety arising from the use of asbestos in Ontario. Toronto: Ontario Ministry of Government Services, 1984. 7. Health and Safety Executive (U.K.). Asbestos: the control limit for asbestos. Prepared by Acheson AD, Gardner MJ. London: Health and Safety Commission, HMSO, 1979. 8. Health and Safety Executive (U.K.). Asbestos: the control limit for asbestos. An update of the relevant sections of the ill effects of asbestos upon health. Prepared by Acheson AD, Gardner MJ. London: Health and Safety Commission, HMSO, 1983. 9. Miller AB. Asbestos fibre dust and gastroin testinal malignancies. Review of literature with re gard to a cause/effect relationship. J Chronic Dis 1978; 31:22-33. 10. Weill H, Hughes J, Waggenspack C. Influence of dose and fiber type on respiratory malignancy risk in asbestos cement manufacturing. Am Rev Respir Dis 1979: 120:345-54. 11. McDonald AD, Fry JS, Woolley AJ, McDonald JC. Dust exposure and mortality in an American chrysotile textile plant. Br J Ind Med 1983; 40:361-7. 12. McDonald AD, Fry JS, Woolley AJ, McDonald JC. Dust exposure and mortality in an American factory using chrysotile, amosite, and crocidolite in mainly textile manufacture. Br J lnd Med 1983; 40:368-74. 13. Henderson VI, Enterline PE. Asbestos ex posure: factors associated with excess cancer and respiratory disease mortality. Ann NY Acad Sci 1979; 330:117-26. 14. Berry G, Newhouse ML. Mortality of work ers manufacturing friction materials using asbestos. Br J Ind Med 1983; 40:1-7. 15. McDonald JC. Liddell FDK, Gibbs GW, Eyssen GE, McDonald AD. Dust exposure and mor tality in chrysotile mining, 1910-1975. Br J lnd Med 1980; 37:11-24. 16. McDonald AD, Fry JS, Woolley AJ, McDonald JC. Dust exposure and mortality in an American chrysotile asbestos friction products plant. Br J Ind Med 1984; 41:151-7. 17. Pott F. Some aspects of dosimetry of the car cinogenic potency of asbestos and other fibrous dusts. Staub-Reinhait Luft 1978; 38:486-90. 18. Stanton MF, LayardM, Tfegeris A, etal Rela tion of panicle dimension to carcinogenicity in amphibole asbestos and other fibrous minerals. JNCI 1981; 67:965-75. 19. Crump KS. Hod DG, Langley CH. Peto R. Fundamental carcinogenic processes and their im plications to low dose risk assessment. Cancer Res 1976; 36:2973-9. 20. Saracd R. Personal environmental interactions in occupational epidemiology. In: McDonald JC, ed. Recent advances in occupational health. Lon don: Churchill Livingstone, 1981:119-28. 21. Berry G, Newhouse ML, Antonis P. Combined effect of asbestos and smoking on mortality from lung cancer and mesothelioma in factory workers. Br J Ind Med 1985; 42:12-8. 22. Selikoff U, Hammond EC, Seidman H. Mor tality experience of insulation workers in the United States and Canada, 1943-1976. Ann NY Acad Sci 1979; 33091-116. 23. Baris Yl.FahinAA.OzesmiM.eraf An out break of pleural mesothelioma and chronic fibros ing pleurisy in the village of Karain/Urgup in Anatolia. Thorax 1978; 33:181. 24. Baris Y. The clinical and radiological aspects of 185 cases of malignant pleural mesothelioma. In: Wagner JC, Davis W, eds. Biological effects of mineral fibres. Vol. 2. Lyon: IARC Scientific Publications, 1981W37-47. 25. Stanton MF, Wrench C. Mechanisms of mesothelioma induction with asbestos and fibrous glass. J Natl Ca Inst 1972; 48:797-821. 26. Davis JMG, Addison J, Bolton RE, Donald son K, Jones AD, Wright A. The pathogenic ef fects of fibrous ceramic aluminum silicate glass ad ministered to rats by inhalation or peritoneal in jection. In: Biological effects of man-made mineral fibres. Copenhagen: World Health Organization. 1984; 303-22. 27. Rubino GF, Piolatto G, Newhouse ML, Scansetti G, Aresini GA, Murray R. Mortality ofchryso tile asbestos miners at the Balangero mine, North ern Italy. Br J Ind Med 1979; 36:187. 28. Weiss W. Mortality of a cohort exposed to chrysotile asbestos. JOM 1977; 19:737. 29. Acheson AD, Gardner MJ, Pippard EC Grime LP. The mortality of two groups of women who manufactured gas masks from chrysotile and crocidolite asbestos: a four-year follow-up. Br J Ind Med 1982; 39:344-8. 30. Thomas HF, Benjamin IT, Elwood PC, Sweetnam PM. Further follow-up study of workers from an asbestos cement factory. Br J Ind Med 1982; 39:273-6. 31. Seidman H, Selikoff IJ, Hammond EC. Short term asbestos work exposure and long-term obser vation. Ann NY Acad Sci 1979; 330:61-89. 32. Acheson ED, Gardner MJ, Winter PD, Ben nett C. Cancer in a factory using amosite asbestos. Int J Epidemiol 1984; 13(1):3-I0. 33. Hobbs MST, Woodward SD, Murphy B. Musk AW, Elder JE. The incidence of pneumoconiosis, mesothelioma, and other respiratory cancer in men engaged in mining and milling crocidolitein West ern Australia. In: Wagner JC, Davis W, eds. Bio logical effects of mineral fibres. Vol. 2. Lyons: In ternational Agency for Research on Cancer, 1980; 615-25. (IARC Scientific Publications No. 30). 34. McDonald AO, McDonald JC. Mesothelioma after crocidolite exposure during gas mask manufac ture. Environ Res 1978; 17:340-6. 35. Jones JSP, Smith PG, Pooley FD, et aL The consequences of exposure to asbestos dust in a war time gas mask factory. Biological effects ofmineral fibres. Lyon: International Agency for Research on Cancer, 1980:637-53. 36. Newhouse ML. Berry G. Patterns of mortal ity in asbestos factory workers in London. Ann NY Acad Sci 1979; 330:53-9. 37. Fmkelstein MM. Mortality among long-term employees of an Ontario asbestos-cement factory. Br J Ind Med 1983; 40:138-44. 38. Elmes PC, Simpson MJC. Insulation work ers in Belfast. A further study of mortality due to asbestos exposure (1940-1975). Br J Ind Med 1977; 34:174-80. 39. Peto J. Lung cancer mortality in relation to measured dust levels in an asbestos textile factory. In: Wagner JC, Davis W, eds. Biological effects of mineral fibres. Lyons: International Agency for Research on Cancer. Vol. 2.1980; 829-36. (IARC Scientific Publications No. 30). 40. Rossiter CE, Coles RM. HM dockyard. Dcvoaport: 1947 mortality study. In: Wagner JC, Davis W, eds. Biological effects of mineral fibres. Lyon: International Agency for Research on Cancer. Vol. 2. 1980; 713-21. (IARC Scientific Publications Na 30). 41. McDonald AD, McDonald JC. Malignant mesotheliomas in North America. Cancer 1980; 46<7):16JO-6. 42. Newhouse ML, Berry G, Skidmore JW. A mor tality study of workers manufacturing friction materials with chrysotile asbestos. Ann Occup Hyg 1982; 26:899-909. 43. Jones JSP, Roberts GH, Pooley FD, et al. The pathology and mineral content of lungs in cases of mesothelioma in the United Kingdom in 1976. In: Wagner JC, Davis W, eds. Biological effects of mineral fibres. Vol. 1. Lyon: international Agency for Research on Cancer, 1980; 187-99. (IARC Sci entific Publications No. 30). 44. McDonald AD. Mineral fibre content of lung in mesothelioma tumours: preliminary report, in: Wagner JC, Davis W, eds. Biological effects of mineral fibres. Lyon: International Agency for Re search on Cancer. Vol. 2. 1980:681-5. (IARC Sci entific Publications No. 30). 45. Churg A, Wiggs B, Depaoli, L. Kampe B, Stevens B. Lung asbestos content in chrysotile work ers with mesothelioma. Am Rev Respir Dis 1984; 130:1042-5. 46. Peto J, Seidman H, Selikoff IJ. Mesothelioma mortality in asbestos workers: implications for models of carcinogenesis and risk assessment. Br J Cancer 1982; 45:124-35. 47. Whittemore A. The age distribution of hu man cancer for carcinogenic exposure of varying intensity. Am J Epidemiol 1977; 106(5):418-32. 48. Peto J. The incidence of pleural mesothelioma in chrysotile asbestos textile workers. In: Biologi cal effects of mineral fibres. Lyon: international Agency for Research on Cancer, 1980; 703. (IARC Scientific Publications Na 30). 49. Fmkelstein MM. Mortality among employees of an Ontario asbestos-cement factory. Am Rev Respir Dis 1984; 129:754-61. 50. Hughes JM, Weill H. Update of mortality study of workers engaged in asbestos cement prod ucts manufacturing. Submitted for publication. 51. Nicholson WJ, Selikoff U, Seidman H, Lilis R, Formby P. Long-term mortality experience of chrysotile miners and millers in Thetford mines, Quebec Ann NY Acad Sci 1979; 330:11-21. 52. Dement JM, Harris RL, Symons MJ, Shy C. Estimates of dose-response for respiratory cancer among chrysotile asbestos textile workers. In: Wal ton WH, ed. Inhaled particles V. Oxford: Pergamon Press, 1982; 869-83. 53. Crump KS. Testimony to the U.S. Department of Labor concerning a proposed revision of the as bestos standard. Washington. D.C., 1984. 54. DevesaS,Horm J, Connelly R.TYends in lung cancer incidence and mortality intheU-S. In: Mizeil M, Correa P, eds. Lung cancer causes and preven tion. Vbrlag-Chemie International, 1984; 33-46. 55. Doll R, Peto R. The causes of cancer: quan titative estimates of avoidable risks of cancer in the U.S. today. JNCI 1981; 66:1193-308. 56. Sebastien P, Billion-Galland MA, Dufour G, Bignon J. Measurement of asbestos air pollution inside buildings sprayed with asbestos. 1980. (EPA 560/13-804)26). 57. Nicholson WJ, Rohl AN, Weisman 1. Asbestos contamination ofair in public buildings. Research Triangle Park, N.C.: U.S Environmental Protec tion Agency, 1975. 58. Sawyer RN, Spooner CM. Sprayed asbestos- 6113 22392 usirrcf uwsunf-oiwnwTTvj ujuswht of mk 13 containing materials in buildings: a guidance docu ment, Part 2. Washington, D.C- U.S. Environmental Protection Agency, 1979. 59. Pinchin DJ. Asbestos in buildings. Toronto: Royal Commission on Asbestos Study, Series Na 8, 1982. 60. U.S. Environmental Protection Agency, Of fice of Toxic Substances. Support document for final rule on friable asbestos-containing materials in school buildings: health effects and magnitude of exposure. Washington, D.C.: U.S Environmen tal Protection Agency, 1982. 61. Peto J, Handerson B, Pike M. Trends in mesothelioma incidence in the U.S. and the fore cast epidemic due to asbestos exposure during World War II. In: Peto R. Schneiderman M, eds. Ban bury report na 9, Quantification of occupational cancer. Cold Spring Harbor Laboratory, 1981; 51-72. 62. McDonald JC. Health implications ofenviron mental exposure to asbestos. Environ Health Perspect 1985 (in press). 63. Wilson R. Risk/benefit analysis for toxic chemicals. Endotoxical Environ Safety 1980; 4:370-83. 64. Metropolitan Life Insurance Company. Statistical Bulletin 1981; 62:5. 65. Crouch E, Wilson R. Risk/benefit analysis. Cambridge, MA: Ballinter Publishing Co.. 1982. 66. Wilson R. Analyzing the daily risks of life. Technology Review 1979; February:4!-6. 67. Gerberich SC, Priest 3D, Boen JR, Straub CP. Maxwell RE. Concussion incidences and seventy in secondary school vanity football players. Am J Public Health 1983; 73 (12):1370-5. 68. National Safety Council. Accident facts. 1976. 69. Sun M. Whooping cough vaccine research revs up. Science 1985; 227:1184-6. 70. Yankelovich D. Science and the public pro cess: why the gap must dose. Issues in Science and Technology 1984; Fali:6-I2. 6113 22393