Document b2xgwX15ZZ6k1Lg8mQvd2kJO

EPIDEMIOLOGY RESOURCES, INC Fb/2- 'S'Ax/V VFC/S/CV PROJECTIONS OF ASBESTOS-RELATED DISEASE 1980-2009 FINAL REPORT August 2, 1982 By ALEXANDER M. WALKER, M.D., Dr.P.H. A ; 777C P.O. Box 57, Chestnut Hill, Massachusetts 02167 (617) 734-9100 TABLE OF CONTENTS Has List of Tables and Figures................................................................................. ii Overview............................................................................................................... I Task 1: Determine the effective number of past asbestos workers.............. 3 Task 2: Project mesothelioma incidence........................................................ 16 Task 3: Project lung cancer incidence............................................................ 17 Task 4: Estimate asbestosis prevalencenow and in the future..................... 21 Task 5: Estimate the amount of asbestos-related disease likely to occur in women............................................................................................. 27 References............................................................................................................. 28 i A 1777 1 Tble LIST OF TABLES & FIGURES Pl 1 Annual Age-Specific Mesothelioma Incidence for Males .............. 3 2 Proportion of Mesothelioma Cases with Asbestos Exposure History 5 3 Distribution (%) of 185 Canadian and 159 U.S. Male Case-Control Pairs According to the Probability of Occupational Asbestos Ex posure as Classified in Four Centers............................................. 7 4 Relative Frequencies of Peritoneal and Pleural Mesothelioma In Heavily Asbestos-Exposed Groups................................................ 8 5 Relative Frequencies of Peritoneal and Pleural Mesothelioma In Less Heavily Asbestos-Exposed Groups........................................ 9 6 Relative Frequencies of Peritoneal and Pleural Mesothelioma In General Populations and In Cases With No Identifiable Asbestos Exposure............................................................................................ 11 7 Numbers of Pleural and Peritoneal Mesotheliomas........................ 12 8 Year of Entry into the Asbestos-Exposed Workforce...................... 14 9 The Dissemination of Asbestos Exposure In Insulation Worker Equivalents Entering the U.S. Labor Force.................................. 15 10 Projected Numbers of New Mesothelioma Cases 1980-2009 in Men With Plausible Asbestos Exposure Using Two Models of In cidence ............................................................................................. 17 11 Incidence of Lung Cancer--Underlying Risk.................................. 18 12 Selikoff Asbestos-Multiplier................................................................... 18 13 Risk Multipliers for Lung Cancer From Cohort Studies................ 19 14 Projected Numbers of New Lung Cancer Cases 1980-2009 in U.S. Men Plausibly Exposed to Asbestos (According to Exposure In tensity: Heavy Exposure and Light Exposure, Under Three As sumptions Regarding the Distribution of Light Exposures) .... 20 15 Lung Cancer: Percent Distribution of Year of First Exposure to Asbestos Products Alleged in Lawsuits (1975-1981) and Predicted by Model (1975-1979) ..................................................... 21 16 Projections of the Number of Prevalent Cases of Asbestosis in U.S. Males 1980-2009, based on the Incidence of Mesothelioma in Asbestotics........................................................................................... 23 17 X-Ray Changes in Asbestos Insulation Workers ............................... 23 18 Projections of the Number of Prevalent Cases of Asbestosis in U.S. Males 1980-2009, Based on the Equivalence of Asbestosis and Mesothelioma Rates .................................................................... 25 . 19 Projections of Asbestosis Lawsuits Assuming 9% of Prevalent Cases (Non-litigants) Bring Suit Each Year............................................. 26 Figure Pip 1 Annual Incidence of Mesothelioma per 100,000 Workers by Years Elapsed Since First Exposure to Asbestos....................................... 2 Mesothelioma Mortality--Asbestosis Mortality................................. 24 24 " A ; 7772 PROJECTIONS OF ASBESTOS-RELATED DISEASE 1980-2009 Overview It is possible to estimate the amount of asbestos-related disease that will occur in the foreseeable future. One method of projection, together with its assumptions and limitations, will form the subject of this report. The method presented here is distinguished from other possible techniques in that it is tied to events currently being observed. Projections of future cases of mesothelioma and lung cancer depend on a number of steps, all of them starting with the number of cases of mesothelioma occurring in the United States. We begin with the cases of asbestos-related mesothelioma occurring in the late 1970's and infer the size of the asbestos-exposed populations which would be required to produce that number of cases. Using actuarial techniques and Johns-Manville (J-M) case data to provide some details of exposure timing, we then back-calculate to estimate the size of the original exposed work force from the 1930's on, whose remnants are now alive. Again using actuarial methods and drawing on known exposure-disease incidence curves, we then project the number of cases of mesothelioma and lung cancer which are likely to occur in the future in the exposed worker population. Since the clinical manifestations of mesothelioma vary according to the intensity of asbestos exposure, we can make a limited refinement of the procedure above by classifying the projected cases as deriving from relatively heavy or relatively light exposures. This is achieved by comparing the distribution of sites of mesothelioma in known relatively heavily exposed cohorts, such as insulation workers, in whom disease is peritoneal at least half the time, with that observed in the U.S. as a whole, in which peritoneal disease occurs only 10% of the time. Projection of the burden of asbestosis in the United States depends both on the rate of occurrence of new disease and on the prevalence of old disease, which may or may not yet have been medically diagnosed in any given individual. Accounting for prevalent, but, as yet, undiagnosed cases of old disease is necessary for asbestosis, but not for mesothelioma or lung cancer because the latter cancers are either detected or lead to death (usually both) within a period less than two years after clinical onset of disease. Asbestosis sufferers, on the other hand, may live for many years with diagnosable (but undiagnosed) disease, and represent a large pool of potential litigants who can "behave like" new cases at any time, simply by being made aware of the nature and probable origin of their disease. We have inferred the number of prevalent asbestosis cases in the late 1970's by two principal methods. In the first, we estimate what proportion of mesothelioma cases have diagnosable asbestosis, and we compare this figure to the rate at which people with asbestosis develop mesothelioma. Knowing the approximate number of mesothelioma cases with asbestosis and the rate at which mesothelioma occurs in asbestotics, we can calculate about how many asbestosis sufferers there must be in order to account for the observed number of mesothelioma cases occurring among them. In the second method, we have drawn on another, independent aspect of the asbestosis-mesothelioma relationship. In studies of groups occupationally exposed to asbestos, the death rates from mesothelioma and asbestosis have been found to be nearly identical. Thus, projections of mesothelioma deaths in persons exposed to asbestos can be expected to give a good estimate of the number of asbestosis deaths. Several studies have provided data on the mortality rate in asbestotics, and on what proportion of the mortality is due to asbestos. The number of asbestotics is then inferred from the number of asbestosis deaths and the mortality rates in persons with asbestosis. At this point the asbestosis projections depend on a scientific judgment which has to be made on the basis of almost no directly relevant data. The question is: do new cases of asbestosis continue to appear many years after the cessation of exposure or, after a certain lag period, does the occurrence of new cases stop? Put another way: does cleaning up the workplace diminsh the risk of those workers who have already been heavily exposed to asbestos but, as yet, have no disease? Preliminary J-M data 1 A ;7773 suggest that cleaning up the work environment does reduce the risk of those already exposed, although not immediately to zero. Asbestosis prevalence projections have been made on the assumption that new cases actually occur through 1984, and that from 1985 on, any apparent new cases represent new diagnoses of existing disease. Several other "quick and dirty" asbestosis projections methods are also available. These depend on drawing analogies between the United Kingdom and the United States, on use of relative mortality figures, and on the extrapolation of data from x-ray surveys of exposed workers. J-M data on women are too scanty to allow any direct estimation of the number of cases of asbestos-related disease on them. (Less than 5% of suits derive from women.) Estimating female disease has to be based on deriving approximate proportionality constants: based on the probable historical timing of female workers exposed to asbestos, the current 5% figure can be expected to decline to nearly zero over the next two decades. The tasks and subtasks which are involved in disease projection are as follows: 1. Determine the effective number of past asbestos workers. a. Determine the number of cases of mesothelioma in the U.S., 1975-1979. b. Calculate the fraction of mesothelioma cases with a documentable history of asbestos exposure and what fraction of the exposed are likely to have been heavily exposed. c. Estimate the timing of the exposure history for U.S. cases. d. Calculate the number of workers now alive and exposed at different times in the past which would be required to account for the current observed mesothelioma incidence. e. Using actuarial techniques, calculate the size of the originally exposed worker population which would yield the estimated numbers of currently living, previously exposed workers. f. For exposure in the more recent past (which would give rise to no current disease), estimate exposure which could give rise to future disease. i. Use survey data to estimate exposure histories. ii. Use information on changing workplace environments to adjust crude exposure estimates. iii. Adjust all original estimates of the number of exposed workers so that the total number of cases of mesothelioma being observed is consistent with total estimates of the exposed workforce (distant past plus recent past). 2. Project mesothelioma incidence. a. Actuarially adjust the size of the exposed population to account for future mortality and calculate the incidence of new cases in the reduced exposed populations. b. Examine sensitivity of projections to component assumptions. 3. Project lung cancer incidence. a. As above, adjust the size of the at-risk populations to account for future mortality. Calcu late the future incidence of lung cancer as a function of future age, future elapsed time from first exposure, and the future size of the population at risk. b. Compare projected and observed lung cancer figures. 4. Estimate asbestosis prevalence now and in the future. a. i. 1980-1984 prevalence using mesothelioma mortality in asbestotics. --How many cases of mesothelioma with asbestosis are occurring? --Mesothelioma incidence as in (la). --Proportion of mesothelioma with asbestosis. --Estimate frequency of occurrence of mesothelioma in asbestotics. ii. Prevalence after 1984. --Age the 1980-1984 prevalence group using actuarial techniques. 2 A : 7774 b. 1980-1984 prevalence based on the equivalence of mesothelioma and asbestosis mortality. i. Derive expected number of asbestosis deaths from projected mesothelioma deaths. ii. Compare asbestosis deaths to death rates in asbestotics to derive an estimate of the number of asbestotics. c. "Quick and dirty" projections. i. U.K../U.S. equivalence. ii. Relative mortality data. iii. Extrapolation from x-ray surveys. d. Derive a general methodology for predicting lawsuits given propensity to sue and disease prevalence. 5. Estimate the amount of asbestos-related disease likely to occur in women. Task la: Determine the number of cases of mesothelioma in the United States, 1975-1979. The National Cancer Institute (NCI) has sponsored two major cancer incidence surveillance programs since 1969. The first, covering the period 1969-71, coordinated existing regional tumor registries and supported the establishment of new registries, permitting a direct assessment of cancer incidence for about 7% of the U.S. population; this was the Third National Cancer Survey (TNCS). In 1973, the NCI reconstituted the administrative structure of the TNCS to provide for an ongoing system of monitoring cancer incidence. This new program, named the Surveillance, Epidemiology and End Results program (SEER) has been in operation continuously since then. SEER reporting regions are Hawaii, Seattle, San Francisco-Oakland, New Mexico, New Orleans, Utah, Connecticut, Atlanta, Detroit, and Iowa. Collectively these represent about 10% of the population of the United States. In 1980, the NCI (Connelly 1980) released a detailed report of trends in mesothelioma incidence based on TNCS (1969-1971) and the first six years of SEER (1973-1978). The annual age-specific mesothe lioma incidence for males in SEER is given in Table 1 below, which also provides an estimate of the total number of men developing mesothelioma in the United States in 1977. The age-specific estimated counts were obtained by multiplying the SEER incidence rates times the Census Bureau's 1977 esti mates of the U.S. male population in the various age categories. TABLE 1 Annual Age-Specific Mesothelioma Incidence for Males Age Group 0-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 Incident Cases Per 100,000 Men Per Year 0 0.02 0.07 0.20 0.17 0.37 1.05 1.59 2.09 Implied U.S Total 1977 0 2.0 6.1 15.2 10.2 20.2 58.9 90.8 110.1 3 A ; 7 7 7 ro Age Group 60-64 65-69 70-74 75-79 Incident Cases Per 100,000 Men Per Year 3.37 3.84 5.72 6.34 Implied US. Total 1977 147.6 143.6 148.5 100.7 853.9 The implied total of about 854 new cases of mesothelioma in U.S. men in 1977 is probably inaccurate for a number of reasons and should be adjusted accordingly: 1. Comparison of the TNCS (1969-1971) and SEER (1973-1978) age-standardized male incidence figures shows an overall rise from 0.51 to 0.88 cases per 100,000 men per year, approximately a 10% annual increase from the midpoint of the TNCS to the midpoint of the reported SEER years (1975 16). To obtain a 1977 incidence figure, a further 15% inflation over the SEER incidence would be appropriate. (The 10% annual increase is virtually identical to the rate of increase in mesothelioma incidence predicted by the incidence model constructed for Task 2. That model predicts annual increases of 11.2% in 1970, declining steadily to 8.4% in 1975.) 2. Since the SEER regions contain proportionately more shipbuilding areas than the U.S. as a whole, the incidence rates are overstated to the extent that mesothelioma occurs more commonly in shipbuilding areas. SEER data suggest that this is indeed the case. The age-standardized white male rates for Seattle and San Francisco are 1.42 and 1.36 cases per 100,000 men annually, while those for Utah and Iowa are 0.82 and 0.54 cases per 100,000 per year. A reasonable estimate as to the overstatement of mesothelioma due to non-representative sampling of the U.S. in SEER overall would be 10-15%, with a best guess of about 12%. 3. Finally, the diagnosis of mesothelioma is by no means easy or clear cut. Selikoff (1980) maintains that a review of all medical evidence in the deaths of U.S. insulation workers from 1967 through 1976 indicates that of 175 mesothelioma deaths, only 104 (59%) had mesothelioma recorded on the death certificate. SEER diagnostic coding is generally felt to be of much higher quality than death certificate information, but since the SEER data are regional, they necessarily derive in part from smaller hospitals with less sophistication in diagnosis than is available in cancer referral centers. A reasonable estimate for the net underreporting in SEER is about 10%. Combining adjustments for time trends, non-representativeness, and underdiagnosis in the SEER data one arrives at a best guess for the number of mesothelioma cases occurring in U.S. men in 1977, not of 854, but of (853.9) (1.15)/(1.12)/(0.90) = 974 cases. The approximate number occurring in the 1975-1979 quinquennium would be 4870. Task lb. Calculate the fraction of mesothelioma cases which have a documented history of asbestos exposure, and estimate what fraction of the exposed are likely to have been heavily exposed. Not all mesothelioma occurs in men with a documentable asbestos exposure history. Assuming essentially no asbestos exposure in low incidence SEER regions, and that the entire difference in mesothelioma incidence is attributable to asbestos, the differences between the highest and lowest mesothelioma incidence would suggest that 38% of mesothelioma in the high risk areas might be "background" incidence. (The annual incidence per 100,000 white males ranges from 0.54 in Iowa to 1.42 in Seattle; 0.54/1.42 = 38%.) In low incidence areas, the fraction of disease attributable to background would be even higher. Inquiries undertaken among mesothelioma patients have yielded estimates of the percent with an asbestos exposure history running from 16% to 76%, depending principally on whether cases come from areas with heavy asbestos-using industries. Table 2 sum marizes the data from a number of sources. 4 A ; 7773 TABLE 2 Proportion of Mesothelioma Cases with Asbestos Exposure History Reference Region; Subjects Source of Data Proportion of Cues with Asbestos Exposure <%) Peto, et al., 1981............ Vianna, et al., 1981........ Vianna, et al, 1981 ........ Tagnon, et al., 1980 .... McDonald, et al., 1973 .. Newhouse and Thompson 1965 .............................. Los Angeles 1974-1978; males New York State excluding NY City; males, 1973-1978 New York State high incidence counties 1968-1978 Coastal Virgina 1972-1978; white males All Canada (emphasis on Quebec) 1968-1970 Patients dying at The London Hospital 1964 and earlier Case or close relative interview Death certificate Occupational record Patient or first degree relative interview Patient or next of kin interview Relatives and friends interview Surviving relatives, medical records 69/101 69/91* 67/193 (68%) (76%) (35%) 24/31 (77%) 17/31** (55%) 43/56 (77%) ll/69c (16%) 31/69d (45%) 40/76 (53%) 31/76** (41%) a Excludes those for whom interviewee was unsure of work history b Excludes indirect exposure (e.g. family member of an exposed person) c "Definite" or "probable" exposures d Includes "possible" exposure as well 5 A >777? While much of the variation in Table 2 must result from real differences in the asbestos exposure histories of mesothelioma patients from different regions, there may also be a substantial variation depending on the group interpreting the exposure history. Table 3 (from McDonald and McDonald, 1980) illustrates this phenomenon. Three hundred and forty-four male cases of mesothelioma were paired with cases matched for age, sex, hospital, and year of death. The comparison cases had died with lung metastases from a nonpulmonary malignant tumor. Job histories were obtained from interview of relatives, coded and submitted to four asbestos research centers for interpretation. While there was general agreement on the proportion with "definite" and "unlikely" exposure to asbestos, there was much less concordance in the interpretation of ambiguous exposure histories. One center, the Environmental Sciences Laboratory of the Mount Sinai School of Medicine, was very much more likely than the others to interpret ambiguous histories as "probable" asbestos exposure. The consensus United States figures (from centers 2, 3 and 4) in Table 3, when averaged, yield an overall figure for the proportion of male mesothelioma cases with a definite or probable history of asbestos exposure of 54%. This number is in the middle of the range of Table 2, and will be used as a best estimate. The proportion with definite exposure appears to be about 20%, averaging the figures from all four centers in Table 3. A separate line of observation and inference indicates that the proportion of mesothelioma cases attributable to relatively heavy occupational exposure to asbestos is considerably less than the 54% with a definite or probable asbestos exposure history. This is based on the observation of the relative frequencies of involvement of the two major sites of appearance of mesothelioma, the pleura and the peritoneum. As indicated in Table 4, in heavily exposed industrial cohorts, the peritoneum probably accounts for about half of all cases. The most important exception to this pattern arose in a group with what (for an industrial cohort) is an atypical exposure pattern: 75% of the observation time in the group of Australian crocidolite miners reported on by Hobbs et al (1980) was in men with a total duration of exposure of less than 12 months. There were no peritoneal tumors documented among these men. In groups less heavily exposed, the peritoneum is less frequently involved, as indicated in Table 5. The least exposed groups are the general population series of incident mesothelioma cases, and those cases determined by direct inquiry to have had no identifiable asbestos exposure. In these groups, the proportion with peritoneal disease is on the order of 10%. See Table 6. From Tables 4-6, it appears that substantial occupational exposure results in mesotheliomas which are at least 50% peritoneal, and that moderate exposure may result in about 20% peritoneal mesothe liomas. If this is true, then only a small fraction of the total contemporary U.S. mesothelioma incidence (9-13% peritoneal) can be attributable to heavy occupational exposure; at most, about 26% of U.S. cases (the 13% peritoneal and an equal number of pleurals) can be attributable to heavy exposure. In fact, the true proportion of heavily exposed mesothelioma cases must be less than 26%, since peritoneal tumors do arise in persons with little or no known asbestos exposure (see Table 6). The 20% with "definite" exposure in Table 3 may provide a more realistic upper limit to the heavily exposed proportion of persons who develop mesothelioma. 6 ft 1777 3 TABLE 3 Distribution (%) of 185 Canadian and 159 U5. Male Case-Control Pairs According to the Probability of Occupational Asbestos Exposure as Classified in Four Centers* Center* Canada (1960-72) 1 2 3 4 U.S.A. (1972) 1 2 3 4 Definite Cases Controls 11.9 1.1 11.9 1.1 13.5 2.7 11.9 1.1 18.2 3.8 18.9 3.8 24.5 6.3 18.9 3.8 Asbestos Exposure Probable Possible Cases Controls Cases Controls 43.8 31.4 4.3 2.7 20.5 14.6 21.1 15.7 29.7 21.1 18.9 20.5 22.7 13.0 22.7 27.6 51.6 35.2 3.8 1.9 31.4 18.7 20.8 18.6 34.0 19.5 16.4 19.5 33.3 15.1 22.6 27.0 Unlikely Cases Controls 40.0 64.9 46.5 68.6 37.8 55.7 42.7 58.4 26.4 59.1 28.9 58.9 25.1 54.1 25.2 54.1 Case-control difference for definite 8M probable exposures 23.2 16.7 19.4 20.5 30.8 27.8 32.7 33.3 Centers: (1) Environmental Sciences Laboratory, Mount Sinai School of Medicine, New York (2) Gesondsheitsorganisatie TNO, den Haag, the Netherlands (3) Department of Epidemiology & Health, McGill University, Montreal, Canada (4) TUC Centenary Institute of Occupational Health, London School of Hygiene and Tropical Medicine, London, England A>7772 7 TABLE 4 Relative Frequencies of Peritoneal and Pleural Mesothelioma in Heavily Asbestos-Exposed Groups Reference Source Population Selikoff 1980 .... Newhouse 1981 Newhouse 1981 Finkelstein 1981... McDonald & McDonald 1980. Hobbs et al. 1980 . Elmes & Simpson 1976. .. North American Insulation Workers 1/67-12/76 English Factory Workers 2 Years Exposure English Female Textile Workers Canadian Workers Receiving Asbestos Disability Benefits Insulation Workers, Asbestos Production and Manufacture US & Canada 1960-1972 Miners & Millers of Crocidolite W. Australia 1943-1966 Mesothelioma in UK 1/60-12/69 Classified from Medical Records as "Heavily Exposed to Asbestos" Peritoneal Pleural Mixed P-P Other 112 63 00 16 10 8 12 45 00 00 00 23 29 00 0 25 00 20 64 60 % Peritoneal or Mixed P-P 64% 62% 40% 44% 44% 0% 29% 8 A ; 773C TABLE 5 Relative Frequencies of Peritoneal and Pleural Mesothelioma in Less Heavily Asbestos-Exposed Groups Reference Source Population Peritoneal Pleural Mixed P-P Other Elmes Simpson 1976. .. Exposed, but not heavily 10 151 13 0 Chovil, et al. 1981. "Compensable" mesothelioma in Ontario, including exposure which was "relatively light, of short duration, in the distant past, or all of these" 9 23 00 McDonald & McDonald 1980. Heating trades, shipyards. 19 117 00 insulation workers), US and Canada 1960-1975 % Peritoneal or Mixed P-P 13% 28% 16% TABLE 6 Relative Frequencies of Peritoneal and Pleural Mesothelioma in General Populations and in Cases with No Identifiable Asbestos Exposure Reference Connelly 1980 Breslow 1982 Elmes & Simpson 1970 Baris, et al. 1979 McDonald & McDonald 1980 Source Population SEER 1973-1978 males TNCS 1969-1971 males U.S. Patients entering comprehensive cancer centers 1978-1980 No known exposure Peritoneal/Pleural/Mixed P-P/Otber 43 369 0 55 9 83 0 0 36 234 0 0 7 52 4 0 Turkish villagers in a hyperendemic area ' No occupational exposure, US and Canadian cases 19621975 1 37 20 119 0 0 0 0 % Peritoneal or Mixed P-P 9% 10% 13% 11% 5% 24% 9 A ; 778 1 Task lc: Estimate the timing of exposure in U.S. cases. Allegations in J-M mesothelioma case data provide information on the timing of asbestos exposure in those persons who develop mesothelioma. There are 278 J-M mesothelioma cases which give an analyzable asbestos exposure history: a plausible age at first exposure (between IS and 54 years of age), and a year of first exposure between 1930 and 1954. (See Task If for an explanation of the 1954 cutoff.) By five-year age groups we have identified the alleged year of first exposure to asbestos. Applying the percent alleging first exposure at various years in J-M data to the age-specific estimates of asbestos-exposed mesothelioma incidence obtained (after adjustments) from SEER, one can esti mate how many cases of mesothelioma occurred among U.S. workers who were first exposed to asbestos at any given age and any given year in the past. This procedure is carried out separately for the 20% of U.S. cases of mesothelioma with presumed heavy exposure, and for the 34% with presumed identifiable light exposure, on the assumption that the relative frequencies of heavy and light exposures did not change greatly until the introduction of dust controls in the work place in the 1960's and the 1970's. Qualitative review of the J-M case data indicates that most current J-M cases are derived from the smaller, more heavily exposed portion of the work force (insulation workers, asbestos factory workers, etc.). Task Id: Calculate the number of workers now alive and exposed at different times in the past which would be required to account for the current observed mesothelioma incidence. How many exposed workers are there? This question can be answered by combining the esti mated counts of mesothelioma cases having exposure history with data on the incidence of mesothe lioma in exposed workers. Peto (Peto et al, 1982) analyzed Selikoff et al's (1980) data on the incidence of mesothelioma in insulation workers and found that incidence could be very closely described by the equation: I = 4.37 x 10"8 x t<"> x P, where I is the number of cases of mesothelioma occurring per year, t is the elapsed time in years since first exposure to asbestos, and Pt is the size of the population of workers with first exposure t years previously. This incidence curve is graphed as a solid line on Figure 1 for a hypothetical population of 100,000 people. Having derived the above equation on the basis of SelikofTs data, Peto et al (1982) examined its generalizability by fitting similar equations, with the same exponent (3.2), to data obtained in a variety of settings. Table 7 provides, for five studies, the Peto estimate of the leading constant term in the above equation, the number of mesotheliomas observed and predicted at five-year intervals from first exposure, and the total number of cases observed. Observed incidence rates from the five studies are plotted on Figure 1, along with Peto's curve (with the constant term calculated from SelikofTs data). All show a sharp increase beginning 15 years from first exposure and continuing--so far as the data tell--indefinitely thereafter. 10 A ; 7782 Annual Incidence of Mesothelioma per 100,000 Workers FIGURE I Annual Incidence of Mesothelioma per 100,000 Workers By Years Elapsed Since First Exposure to Asbestos Years Elapsed Since First Exposure to Asbestos Legend: ----------- Peto incidence curve (Peto et al, 1982) .......... Breslow incidence curve (Breslow, 1982) Incidence rates reported by Peto et al (1982) for five studies (see Table 7): OSelikoff etal (1979) A Newhouse and Berry (1976) Peto (1980) Hobbs et al (1980) Seidman et al (1980) 11 k ;7783 TABLE 7 * 30 or more years; 32.5 assumed in calculating expected no. 20 or more years; 22.5 assumed in calculating expected no.t 8 <A<AOd<N SO -- 00 OO -- -- -- -- 00 8 in n v-> in 8 r-> r- m o s <n r- r** ^ ^ > s in r- os in in CA I04n8 Aft 0i8 f~ S ^C* -^"4ag** *C* fS" J_!oSwwi" 2 o O s s -- rs in v> r- 00 2 " V hm .5 g-V W.e * I BS" 2 *I -->n ffns r ^v --- ".g^ 3i#Esa 2Z " "S J 5 gr > 8 IS 2 u*z8 flu A A ~ .2 XIsE in -- m ^ o*. m Os N1 MA *c> .2 nc* ao- S5 a3 r tru--N- --n p(N- -- -- IN s0 r-* oe n O in -- -- sO 00 O' IN f*l n n ^ to 1 ^^ rs o 0000 N O N " "0 cdiN -- mr* O 00 t m in *n ^ o -------- m SO -- r- -- m m oe -- -- in in 00 -- -- IN -- o IN IN OOOOO IN IN 00 IN OOO -- T soo -- -- O-- n-- fwiuJ *0Bs8 a** M^ fa. 3 JA e- o e 5faA. Ea Z m -- -- rj- in OOOO0O m r-* in SO -- s g-> a. Z uj US > o. Z u Z a. Z u&>Z > a. Z t2 Z S-> SI Z uj Z Or-s Os C' c K ut CD 38 cl *2 ** V .2 v. o * 2 -- *3 o oII X t> ^ OO a* -- *E *< I* IS |R *1 V Jje 11 * -Clb w 8.2 K VK Crt l --. *" M o 00 Os -- w CO K w .3 X a-s ?si- M2i. -*Moy h41 "ItAfl <=s .5 # es |! 12 A ; 7784 Although Peto's equation has the theoretically attractive property that its mathematical form is derivable from current theories of carcinogenesis, it is not the only possible mathematical description of the U.S. insulation worker mesothelioma incidence. Breslow (1982) has used the same data to esti mate the components of a second formula which incorporates an estimated latent period as well as the exponential and constant terms used in Peto's formula. The form of this equation was proposed originally by Newhouse and Berry (1976), and in a statistical sense it fits the observations more closely than does the Peto equation. Using the symbols as above, that incidence equation is: I = 1.37 x 10"s x (t-15)IM4x P,. The better fit of this equation to SelikofTs data derives principally from the fact that it predicts no cases to occur in the first 1S years following first exposure; Peto's equation predicts a small number of cases. In fact, in SelikofTs data none were observed to occur. Breslow's function is plotted as a dashed curve on Figure 1. The two equations give very similar estimates. Throughout the ensuing analysis, Peto's equation has been used because it appears to provide estimates more consistent with the full range of reported values (as opposed to SelikofTs data alone). Nonetheless, projections carried out using the Peto equation have all been verified using the Breslow equation. Sources reporting to J-M's legal staff indicate that there may be some overstatement in SelikofTs data of the numbers of mesothelioma cases actually occurring. The overstatement certainly appears to be the case for asbestosis, and probably is negligible for lung cancer. For mesothelioma, we have provisionally adopted a correction factor of 0.8 in Peto's and Breslow's equations relating mesothe lioma incidence to asbestos exposure. This has the effect of expanding the size of the exposed popula tion required to account for currently observed mesothelioma cases. Given the number of cases which have occurred and knowing their alleged elapsed interval since first exposure, one can solve the equations above for Pt. We carried out this procedure for all five-year categories of age at diagnosis of mesothelioma in 1975-1979, and for all elapsed times since first exposure, giving a matrix of Pt's which corresponds to the distribution of asbestos workers alive in 1975-1979, which gave rise to mesothelioma, cross-tabulated by age and year of first exposure to asbestos. The size of the exposed population calculated in this step is an artificial figure corresponding to "insulation worker equivalents", that is, the number of asbestos workers as heavily exposed as the insulation workers studied by Selikoff who would be required to explain the current mesothelioma experience. As outlined in Task lb, it is likely that only some 37% of exposed mesothelioma cases (20% out of the 54%) actually arise in persons with heavy exposure. The remaining cases derive from larger populations with less intense exposure. For the purposes of mesothelioma incidence projections, there is no need to distinguish between the two situations. A population of 100,000 workers at one exposure level can be expected to give rise to the same number of cases as a population of 200,000 at half the exposure. Therefore the "worker-equivalent" has an interpretation which is independent of the actual distribution of intensities of exposure. For lung cancer, the distinction does make a difference; the larger population will have a larger "background" number of lung cancers to which the asbestosattributable cancers must be added. In either case, the impression of J-M's legal staff at present is that most lawsuits are coming from heavily exposed plaintiffs, indicating that the distinction may be rele vant for a person's propensity to sue. The cancer projections which follow (Tasks 2 and 3) therefore will be divided according to intensity of exposure. Task le: Using actuarial techniques, calculate the size of the originally exposed worker population which would yield the estimated numbers of currently living, previously exposed workers. Every group of workers now alive with first exposure at some time in the past corresponds to an originally exposed population, some members of which have died with the passage of time. We have used white male actuarial survival tables covering the years 1930-1979 to calculate the originally exposed population as follows: call the population alive in year y, of age a, with time t since first exposure, PyjM. Call the same population t years previously (when they were a-t years ld), Py_u_l-(>. If 13 A .'778b the actuarial probability for survival from age a-t to age a, starting in year y-t, is Sy_w_u, then: Py-i.a-t.0 -- Pyj.t/Sy-u-u* where Py_tJ_t0 is the number of workers entering the asbestos-exposed work force in year y-t at age a-t who would be required to produce in year y a surviving exposed population of size Py4,t. If that surviving exposed population has already been derived (Task Id) as one large enough to produce the estimated number of new exposed cases of mesothelioma of age a who report having been first exposed to asbestos t years previously, then Py-^-uo *s the size of the new workforce t years earlier which would be required to account for the current mesothelioma experience. A modification of the actuarial survival figures was necessary before carrying out the above calculations. As noted in Task lb, some 20/54 (37%) of the insulation worker equivalents estimated here derived from heavily exposed individuals. Selikoff has noted that insulation workers have a 37% higher overall mortality rate than the general population. As a result, a group consisting of 37% heavily exposed workers has approximately a (37%) x (37%) = 14% higher mortality rate than the general population. This correction factor has been introduced into all calculations of the survival of non-diseased, asbestos-exposed groups of insulation-worker equivalents. Task If: For exposure in the more recent past which would give rise to no current disease, estimate the quantity of exposure, since this could still give rise to future disease. The procedures of Task Id and le begin breaking down when first exposures less than 20 years prior to the 1975-1979 period are considered. Few cases of mesothelioma would have as yet arisen from this period, because of the delayed rise in the mesothelioma incidence curve; thus it is impossible to work backwards to the exposed population with any reliability. Instead, we have used the reported distribution of age at first entry into an asbestos-related industry from a survey of men aged 40 and above, conducted by Elrick and Lavidge. Although that survey does not allow us to calculate the absolute number of heavily exposed asbestos workers entering the workforce each year, it does allow us to calculate the relative number entering. Altogether, the Elrick and Lavidge survey identified 214 individuals who had worked in an asbestos-using industry, of whom 79 were aware of actual exposure to asbestos dust. The results of the survey were as shown in Table 8, from which it is apparent that the distribution of years of entry into asbestos-using industries is essentially identical for men who did recall and those who did not recall direct asbestos exposure. We have based our adjustments on the larger, statistically more stable numbers of all entrants into asbestos-using industries. TABLE 8 Year of Entry into the Asbestos-Exposed Workforce Year Workers Who Recalled Asbestos Exposure Number %_ Cumulative % All Workers In Asbestos-Using Industries Number _% Cumulative Before 1930 1930-1934 1935-1939 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 1975-1979 5 6.3 1 1.3 6 7.6 17 21.5 14 17.7 5 6.3 10 12.7 9 11.4 7 8.9 1 1.3 4 5.1 6.3 7.6 15.2 36.7 54.4 60.8 73.4 84.8 93.7 94.9 100.0 16 7.5 9 4.2 19 8.9 45 21.0 29 13.6 19 8.9 29 13.6 21 9.8 14 6.5 5 2.3 8 3.7 7.5 11.7 20.6 41.6 55.1 64.0 77.6 87.4 93.9 96.3 100.0 The work force was calculated directly from the mesothelioma incidence data through 1954. From 1955 on, we assigned numbers proportional to the annual and special sample number for men in the survey reporting first entry into the asbestos workforce in each quinqennium. Adding the post-1955 workers generally resulted in a small increase in the predicted numbers of mesotheliomas for 14 A i 778,3 1975-1979. Both pre- and post-1955 figures were then iteratively adjusted until: 1) the pre-1955 figures maintained the year of first exposure distribution implied by the J-M cases, 2) the post-1955 figures stood in the proportion to the pre-1955 that was dictated by the survey, and 3) the 1975-1979 mesothelioma predictions derived from the resulting distribution of exposed workers equalled the numbers actually thought to have occurred. From 1965 on, it is probable that exposed workers faced diminishing amounts of ambient asbestos fiber. We have accounted for this by recalculating the overall workforce size (as above) but with smaller fractions of the exposed workforce assigned to the post-1955 period. (In effect, this amounts to modeling the health effects of an unchanged number of workers exposed to less and less asbestos by assuming smaller and smaller numbers of workers exposed to essentially the same amount of asbestos.) Workforce discounts used to compensate for recent improvements in the workplace were: 1960-1964, 10%; 1965-1969, 50%; 1970-1974, 75%; and 1975-1979, 100%. Table 9 gives the number of insulation worker equivalents entering the workforce for each quin quennium 1930-1979, as estimated from J-M data directly, and with adjustment of the post-1955 figures to match the survey data in their calendar year distribution. Of particular interest is the general conformity of the worker distribution for earlier years (as inferred from J-M data. Table 9) to that actually observed in the Elrick and Lavidge survey. Table 9 also gives the number of insulation worker equivalents entering the workforce in each quinquennium from 1930 to 1979, according to intensity of exposure. For the heavily exposed group, one equivalent can be taken to equal (more or less) one worker. For the less heavily exposed group, there may be as many as five or ten actual workers making up each insulation worker equivalent. 1930-1934 1935-1939 1940-1944 1945-1949 1950-1954 1955-1959 1960-1964 1965-1969 1970-1974 TABLE 9 The Dissemination of Asbestos Exposure In Insulation Worker Equivalents Entering the U.S. Labor Force Heavily Exposed 1) Lightly Exposed(2) 2,700 13,800 55,900 35,400 38,100 34,900 22,700 8,600 1,400 3,700 18,100 74,900 49.900 59,400 49,300 32,100 12,200 2,100 TotaK3) 6,400 31,900 130,800 85,300 97,500 84,200 54,800 20,800 3,500 (1) In the heavily exposed, each insulation worker equivalent corresponds roughly to a single worker. (2) Many lightly exposed individuals are required to make up a single insulation worker equivalent (listed in the Table). (3) The total number of insulation worker equivalents, not the total of exposed workers. 15 A ; 7 7 8? Task 2a: Adjust exposed population and calculate future incidence of mesothelioma. For every five year period in the future, we calculated the number of mesothelioma cases that would be expected to arise out of each group of workers we estimated to have entered the workforce at each age-of-entry in each calendar-quinquennium in the past. Using the notation of previous sections, call a the age in 1975-1979, and t the number of years elapsed since first exposure in year y. Call p the number of years into the future for which a projection is being made. Then the age at first exposure is (a--t), age in the projected future year is (a+p), and elapsed time from first exposure in the future is (t+p). The components of the projection equation (Peto et al, 1982) are: Ia+P. i+p--The number of new cases of mesothelioma p years from 1975-1979 among persons then aged (a+p), with time elapsed since first exposure (t+p). Pa--t,o,y--The number of persons entering the asbestos workforce t years prior to 1975-1979, at age (a--p), in the year y. Sa_, a+Pi y--Actuarial survival from age (a-t) in year y to age (a+p), taking into account recorded actuarial survival for calender years prior to 1979, and assuming age-specific mortalities after 1979 to be unchanged from their late I970's values. K--A constant term derived from SelikofTs observation of insulation workers (Peto et al, 1981), 4.37xl0*8 for one-year incidence, 2.19 x 10-7 for five-year incidence, multiplied by a correction factor (0.8), for possible excess diagnosis in SelikofTs follow-up. The projection equation is: ^a+p, t+p Pa-t. 0. y X Sa _t> l+pi y X K X (t+p)^ The estimated populations of exposed workers presented at the end of Task 1 in Table 9 yield projections of mesothelioma incidence shown in the first column of Table 10 for the years 1980-2009. Use of Breslow's projection equation to derive the exposed population and the resultant mesothe lioma incidence leads to very similar projected counts, as seen in the second column of Table 10. Redefining K to equal 1.37 x 10'5 (multiplied again by 0.8), the Breslow projecting equation is: U+P.I+P = P-,o.y X sa_,,1+p,y X K x (t+p-15) l 846. 16 A ; 7738 TABLE 10 Projected Numbers of New Mesothelioma Cases 1980-2009 in Men With Plausible Asbestos Exposure Histories Using Two Models of Incidence No Latency Period (Peo) 1980-1984 1985-1989 1990-1994 1995-1999 2000-2004 2005-2009 3,200 3,500 3,600 3,400 2,900 2,100 Total 1980-2009 18,700 Latency Period (Breslow) 3,400 3,900 4,200 4,000 3,500 2,500 21,500 Task 2b: Estimate the sensitivity of mesothelioma projections to the assumptions involved. Assumptions underlying the mesothelioma analysis fall into two categories, depending on whether they are reflected in the final mesothelioma projections in a linear or non-linear fashion. The linear factors affect the height of the projected curve, whereas the non-linear factors affect its shape. The linear factors are those of Task la (corrections of trend in incidence, non-representativeness of the SEER populations, general underdiagnosis of mesothelioma), and Task lb (fraction of mesothe lioma cases exposed to asbestos). While the net effect of uncertainties in these elements may be as much as 30% either way, it is crucial to bear in mind that the ultimate projections of mesothelioma lawsuits must be tied to the number of suits currently occurring in relation to the number of cases of disease occurring ("propensity to sue"), so that these variations in absolute incidence have no final effect on projections for numbers of lawsuits. Non-linear effects, because they affect the shape of the future mesothelioma curve, are highly relevant to projections of suits. The shape of the curve is relatively robust to large variations in the non-linear parameters. Assuming that the reported years of first exposure to asbestos in the J-M files were off by five years in either direction gives very distorted year of entry distributions (with peak employment coming before or after the war), and only shifts the incidence peak forward or backward five years, but does not change its broad, relatively flat character. Changing the age distribution of current mesothelioma cases (in the SEER data) by assigning 30% more cases to the 70 and over, or to the under-50 year old age groups shows similar five year shifts in the incidence peak. Assuming that the workplace was entirely cleaned up by 1965 slightly increases the short term projections for mesothelioma and decreased the long term ones, since the effect of recent exposures (or non-exposure) can be expected to be felt only many years in the future. Assigning all worker-equivalents the higher mortality of insulation workers leads to slightly faster decline in the pool of exposed workers, and a 510% drop in late (post year 2000) estimated mesothelioma incidence. Assuming that mortality of exposed workers is that of the general population leads to a 5% increase in late mesothelioma incidence. Task 3a: Adjust the size of the exposed population and calculate future incidence of lung cancer. The incidence of lung cancer among insulation workers has been reasonably well documented by Selikoff et al (1980) and others (Berry 1980, McDonald et al 1980, Henderson and Enterline 1979) to be the product of two terms: the "underlying risk" for lung cancer that would hold in the absence of asbestos exposure, and a multiplication factor resulting from exposure. The underlying 17 A ; 7783 lung cancer risk is a sharply rising function of age. From the 1973-1978 National Cancer Institute report of the Surveillance, Epidemiology, and End Results program the underlying risk for white males is: TABLE 11 Incidence of Lung Cancer--Underlying Risk Age (yean) New Cues of Lung Cancer Per 100,000 While Men Per Year 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 0.2 0.6 2.6 7.4 22.8 58.3 106.8 181.7 284.2 400.9 481.5 519.0 The asbestos-multiplier calculated by Selikoff et al varies not with age, but with elapsed time since first exposure. Values for the multiplier are: TABLE 12 Selikoff Asbestos-Multiplier Time Since First Exposure (yean) 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 Multiplication Factor 2.55 3.40 3.48 5.00 6.08 5.68 4.93 3.89 Although the data which Selikoff has assembled on insulation workers represent the largest, statistically most reliable source of follow-up information, there have been a number of other studies of exposed cohorts. These are summarized on the following page (Table 13). Among all occupational groups, insulators are matched only by factory workers for their levels of lung cancer risks. Miners and, even more so, shipyard workers, have much lower risk multipliers, the former possibly because ambient fiber concentrations are lower in mining than in processing jobs, the latter almost certainly because the category "shipyard worker" includes many people with minimal exposure. 18 &-;770C TABLE 13 Risk Multipliers for Lung Cancer from Cohort Studies* Source Insulators Selikoff et al (1980) Factory workers Henderson & Enterline (1979) Newhouse & Berry (1979) Peto et al (1977) Cement workers Weill et al (1979) Shipyard workers (except insulators) Kolonel et al (1980) Miners McDonald et al (1980) Nicholson et al (1979) Hobbs et al (1980) Minimum (yean following ________ exposure)________ 0 10) 1.8 (!) 2.42 2)<J) 1.25 (10-14) 0.77 (10-15) 1.1 10) 0.80 -- 2.10 1)(J' -- (0-9) Maximum (yean following _________exposure) 6.08 (30-34) 7.78 (1) 5.38 (>2)<J) 1.71 (>20) 3.33 (30-35) 1.3 (2:10) 2.65 (>20)(l> 4.19 (40-49) 2.46 (2:15) (l> Time data not given separately from accummulated dust exposure (J) Time figures are for years of employment * In each study, if several exposure levels were given, the highest is reproduced here. The multipliers observed by Selikoff are unlikely to be purely the result of asbestos exposure. To the extent that the asbestos-exposed workers studied by Selikoff differed in their smoking habits from the general population, the multipliers include this effect as well. If part of the purpose of the projections was to estimate what fraction of cases in exposed workers were actually attributable to asbestos exposure, lack of smoking data from Selikoffs workers would represent a serious lack. In fact, the purpose of the projection is to decide how many cases of lung cancer occur in toto among exposed workers. For this purpose, it is sufficient to assume that the workers observed by Selikoff have approxi mately the same fraction of smokers as do asbestos workers in general. This appears to be a reasonable assumption, even for future projections. Although there has been a decrease in adult male smoking in the United States, the trend appears to include blue collar workers to a smaller extent than others (Surgeon General 1978). Selikoffs multipliers above describe the experience of a cohort of workers with lifetime exposure to asbestos, and may not correctly predict the future experience of workers, after occupational exposure to asbestos has been curtailed. Observations in chrysotile miners (Berry 1980; McDonald et al 1980) and asbestos factory workers (Henderson and Enterline 1979) have lent strong support to the idea that relative risk for lung cancer (i.e., the multiplier) may be nearly linearly related to accumulated asbestos exposure over a fairly wide range. SelikofTs multipliers can be interpreted as reflecting a reasonably steady rise through working life, with a decline beginning around the time of retirement, that is, about the time of cessation of asbestos exposure. We have incorporated this phenomenon into the projection equations by constraining the lung cancer multipliers for each exposure cohort to decline from their 1975-1979 value at a rate of 10% per quinquennium. It should be noted that while this discounting of later multipliers fits the Selikoff data most closely, there is no a priori reason to expect a decline, and mathematical modeling of other, smaller bodies of data does not suggest a decline. While the values used represent our "best estimate" for projections, experts who disagreed would probably choose somewhat higher multipliers and consequently would project somewhat larger numbers of cases. 19 A : 7791 The projection equation has the following components: Ia+Pit+P--The number of new cases of lung cancer p years from 1975-1979 among persons then aged (a+p) years, with time elapsed since first exposure (t+p). Pa-uo,y--The number of persons entering the asbestos workforce t years prior to 1975-1979, in year y at age (a-t). Sa-i4+p,y--Actuarial survival from age (a-t) to age (a+p), beginning in year y. L,+p--The underlying risk of lung cancer as determined from NCI tables for white males of age (a+p). Mt--Lung cancer risk multiplier for elapsed time since first exposure through 1975-1979, reduced by 10% for each subsequent quinquennium. The projecting equation is: !a+p,t+p ~ P-t,o,y ^ X L,+pX M,. The projected incidence is calculated for each worker group entering the workforce at every age in every past year, and the projected number of cases occurring in every five-year period are summed. The projection procedure described above is, by itself, correct only for a heavily exposed worker cohort (such as insulation workers or factory workers). For a less heavily exposed cohort, such as shipyard workers, the projection predicts only the asbestos-attributable disease, plus the background lung cancers which would be found in a population whose size corresponds to the number of insulation worker equivalents. When the insulation worker equivalent exposures are spread out over a larger number of persons, more background disease needs to be recognized in order to project the total burden of lung cancer (both spontaneous and asbestos-attributable) in asbestos exposed persons. Table 14 presents projections of lung cancer for a worker population, which, on the average, is about one-half as intensely exposed to asbestos as insulation workers. Since our best estimate is that 37% of the "insula tion worker equivalents" are indeed heavily exposed, this implies that the remaining 63% of equivalents arir- in workers whose exposure intensity is 2.6 times less than that of an insulation worker. TABLE 14 Projected Numbers of New Lung Cancer Cases 1980-2009 In U.S. Men Plausibly Exposed to Asbestos Year Numbers of New Cases 1980-1984 1985-1989 1990-1994 1995-1999 2000-2004 2005-2009 17,800 13,600 10,200 7,000 4,300 2,220 Task 3b: Compare projected and observed lung cancer figures. Since no part of the model used to predict lung cancer incidence is based on data from lawsuits coming in to J-M, it is possible to test partially the validity of the projections by comparing details of the J-M litigation file against the model's projections. Table 15 displays the distribution of alleged year of first exposure to asbestos among the 349 J-M cases aged 40-79 who filed suit claiming lung cancer in the years 1975-1981, and who further gave a sufficiently detailed exposure history to allow them to be classified. The percent distributions are further cross-classified by the age at which suit was filed. In parallel are the age-specific distributions of years of first exposure to asbestos for the 22,248 male lung cancer cases for the period 1975-1979 predicted by the model developed in Tasks 1 through 3. Lung cancer is less well connected with asbestos exposure than is mesothelioma, both in the medical 20 A ; 7792 and in the legal communities, and so there may be an increased element of serendipity in bringing cases of disease to litigation. Nonetheless, there appears to be a fair correlation between the observed and predicted distribution of years of first exposure, particularly in the overall figures. Although the lawsuits tabulated do not represent the total number coming in to J-M (most do not have detailed asbestos exposure data), their distribution over age categories reflects a plausible pattern of litigious ness when compared to the projected distribution of lung cancers. The highest propensity to sue appears to be in the 40-49 year age group, with a gradual tailing off to age 69, and a precipitous drop thereafter. TABLE 15 Age at Diagnosis /Lawsuit Lung Cancer Percent Distribution of Year of First Exposure to Asbestos Products Alleged in Lawsuits (1975-1981) and Predicted by Model (1975-1979) 1930-34 Total 1935-39 1940-44 1945-49 1950-54 1955-59 1960-64 1965-69 1970-74 % Count 40-49 L(l) 2.7 2.7. 13.5 8.1 35.1 18.9 8.1 8.1 2.7 100.0 37 M(2) 0 0 0 19.7 30.2 29.2 15.5 4.6 0.7 100.0 698 50-59 L 1.6 9.0 30.3 23.0 18.0 9.0 5.7 2.5 0.8 100.0 122 M 0 2.7 20.3 19.5 20.7 23.0 11.5 2.1 0.1 100.0 3770 60-69 L 4.2 15.5 33.1 22.3 10.1 6.8 4.7 1.4 1.4 100.0 148 M 3.9 11.0 35.1 16.3 14.5 15.4 3.6 0 0 100.0 8313 70-79 Total L 7.1 16.7 33.3 14.3 7.1 7.1 7.1 7.1 0 100.0 42 M 0.8 10.4 56.0 20.3 12.4 0 0 0 0 100.0 9443 L 3.7 12.0 30.1 20.1 15.2 8.9 5.7 3.2 1.1 100.0 349 M 1.8 9.0 40.3 18.7 15.1 10.6 3.9 0.5 0.0 100.0 22,248 (1)L Lawsuits with analyzable exposure data registered at JM (2)M Model projections of total number of lung cancer cases arising in asbestos-exposed men in the United States. Task 4: Predict future asbestosis prevalence. Unlike persons with mesothelioma or lung cancer, persons with asbestosis are likely to live many years after the onset of their disease. Much or all of that time they may be unaware that their symptoms are due to asbestosis. From the point of view of medical and legal awareness of the disease, then, the key event in the progression of a case of asbestosis is not the date of onset, but rather the date of diagnosis. For predicting the rates of diagnosis (and hence suit) the key underlying figure to examine is the prevalence of potentially diagnosable cases in the general population. There are no direct measures of the prevalence of diagnosable asbestosis in the United States. There are, however, methods of arriving at educated guesses. One depends on the occurrence of mesothelioma in persons with asbestosis; another depends on an equivalence between asbestosis mortal ity rates and mesothelioma mortality rates. Task 4a: Predict asbestosis using mesothelioma mortality rates in asbestotics. Elmes and Simpson (1976) have reviewed the clinical^ pathologic and radiographic records of 327 cases of mesothelioma occurring in the United Kingdom between 1960 and 1969. They found that 70/247 cases with chest radiographs (28%) had clear radiographic evidence of concurrent asbestosis. This figure is not a biological constant; rather it probably reflects the particluar distribution of intensi ties and durations of exposure to asbestos which U.K. mesothelioma cases had undergone in the 1960's. If the general historical pattern of asbestos exposure in the United States is similar to that in the U.K., then one may estimate that about 28% of the cases of mesothelioma in the U.S. in the late 1970's, or about 273 cases annually, had concurrent diagnosable asbestosis. Put another way, of all the people with diagnosable asbestosis in the United States, about 273 developed mesothelioma each year. Three studies give rates of occurrence of mesothelioma in persons with asbestosis which are of the same order of magnitude. Berry (1981) provides the most extensive data: 25 cases of mesothelioma 21 A ; 7793 occurred in 665 Englishmen with asbestosis certified for the purpose of disability insurance, followed for 4165 man-years of follow-up between 1952 and 1976. Thus, he found a rate of one case per 166 man-years. A total experience about three-fifths as large was reported by Edge (1979) who observed 7 cases of mesothelioma in 2637 man-years of observation in 429 men who had been identified by chest radiographs showing pleural plaques taken between 1964 and 1971 in an English shipyard community. His observed rate is one case per 377 man-years of observation. Least informative because of its small amount of observation is the report of Finkelstein et al (1981) who studied mortality among 172 workers receiving workman's compensation for asbestosis in Ontario between 1942 and 1979 followed for 733 man-years. There were three death certificate records of mesothelioma (one per 244 manyears) and six further cases identified by review of other records (total of one case per 81 man-years of observation). Totalling the experience recorded in the three studies, one obtains reports of 41 cases in 7535 man-years of observation, or one case in 184 man-years of observation. Exclusion of the six Finkelstein cases discovered only after record review would give an overall rate of one case of mesothe lioma per 215 man-years. We have chosen a figure of 1 per 200 man-years as a summary figure. As with the 28% figure for the fraction of mesothelioma cases with asbestosis, the one per 200 man-years estimate for mesothelioma in asbestosis should not be taken to be a biological constant. It too probably reflects the distribution of intensities and durations of exposure to asbestos holding rough ly over the three decades ending in 1975. If, however, we accept the rate of one case of mesothelioma per 200 asbestotics per year, and combine this with the expected number of mesothelioma-asbestosis cases derived before, i.e., 273, we arrive at an overall estimate of about 55,000 persons with diagnosable asbestosis in the United States in the late 1970's. Expressed algebraically, the line of reasoning above is as follows. Let I be the annual incidence of mesothelioma in asbestotics; let A be the number of asbestotics in the U.S.; let M be the annual number of new mesothelioma cases in the U.S.; and let P be the proportion of those with concurrent asbestosis, then A x I * M x P and A - M x P/I. Substituting known or estimated values: A = (974 )(0.28)/( 1/200) A = 54,544 men with asbestosis. We have gone through the above calculation separately for each age group in order to derive projected prevalent numbers of asbestosis cases at every age in the period 1975-1979. At present it is not known for how long new cases of asbestosis will continue to develop among currently healthy workers previously exposed to asbestos, assuming that workplace contamination has been essentially eliminated since 1975, and greatly reduced prior to that. J-M's worker experience indicates that there has been a precipitous decline in new cases of asbestosis over the last decade (Chase 1981). This would argue in favor of not projecting the occurrence of new cases beyond 1985. As a best estimate then, we have based asbestosis prevalence projections on an assumption of continued new occurrence through the first half of this decade. Starting from the projected mesothelioma incidence in 1980-1984, we have projected asbestosis prevalence in each age group for 1980-1984 using the projection equation described above. For projections beyond 1984, we have aged the populations using modified 1977 white male actuarial survival figures. The modification is based on strong evidence of very much higher mortality rates in men with asbestosis than in the general population. Berry (1981) reports on 283 deaths in asbestotics with only 108.6 expected; Finkelstein et al (1981) report 66 deaths in asbestotics with only 16.6 expected. Together these give an overall mortality for asbestotics 2.79 times that which would otherwise be expected. Table 16 provides our projections of annual diagnosable asbestosis prevalence for each quinquennium through the year 2009. Of particular importance in interpreting Table 16 (and Table 18 in the following task) is that the prevalence figures are for clinically diagnosable (not necessarily diagnosed) asbestosis which would qualify for worker's compensation in the U.K. or Canada. This is inescapable, because the only 22 779 detailed survival figures available on men with asbestosis derive from these registered and monitored groups. Depending on the criteria used, very much more "asbestosis" can be diagnosed on the basis of minimal radiologic changes. Table 17 (Selikoff 1976) illustrates the problem. Chest x-rays of 1117 men were graded according to the degree of asbestosis, classified on a four point scale ranging from 0 (no disease) to 3 (severe asbestosis). The readings were cross-classified by time since first exposure to asbestosis. It is very unlikely that workers with SelikofTs minimal (grade 1) asbestosis would qualify for worker's compensation. If such workers were to be included in prevalence estimates, however, the projections of Tables 16 or 18 (Task 4b) would have to be roughly tripled. TABLE 16 Projections of the Number of Prevalent Cases of Asbestosis In U.S. Males 1980-2009 Based on the Incidence of Mesothelioma in Asbestotics Years Number of Men Alive With Asbestosis 1980-1984 1985-1989 1990-1994 1995-1999 2000-2004 2005-2009 65,800 35,400 19,000 9,600 4,400 1,700 TABLE 17 Years Since Onset of Exposure X-ray Changes in Asbestos Insulation Workers Percent Distribution by Asbestosis Grade No. 0 i 2 3 40 + 30-39 20-29 10-19 0-9 121 5.8 28.9 42.1 23.1 194 12.9 52.6 25.3 9.3 77 27.2 45.5 22.1 5.2 379 55.9 41.7 2.4 0.0 346 89.6 11.4 0.0 0.0 Task 4b: Estimate asbestosis prevalence using the equivalence between asbestosis and mesothelioma mortality. A second line of reasoning about asbestosis prevalence can lead to an independent estimate by which to gauge the results of the previous task. This is based on the near perfect identity of the time course and magnitude of asbestosis mortality and mesothelioma mortality in SelikofTs (1980) insula tion worker data, combined with independent estimates of mortality in men with asbestosis. SelikofTs (1980) observations of mortality from asbestosis and mesothelioma are summarized in Figure 2. It is of particular importance that the asbestosis mortality recorded by Selikoff is not simply an estimate of mortality in men with asbestosis, but rather specifically of mortality due to asbestosis. Berpr (1981) found that 56 of 263 deaths (21.3%) in British men with asbestosis were actually attributed to asbestosis. Finkelstein et al (1981) found for the more inclusive category "non-malignant respiratory disease" 23 out of 61 deaths (37.7%) in Canadian men receiving workmen's compensation for asbestosis. This figure is consistent with Berry's, which is based on larger numbers and more specific reporting. Independent estimates of mortality rates among asbestosis sufferers place them at about 2.8 times the corresponding age-specific rates in the general population (see Task 4a). For any given age group, write the mesothelioma deaths among men exposed to asbestos as Ma, and the corresponding count of all deaths in men with asbestosis as Da. Then the Selikoff finding of an equality in the numbers of mesothelioma deaths and asbestosis deaths in insulation workers, combined with Berry's finding that 21.3% of all death is due to asbestosis, is Ma = 0.213 x Da. 23 A ; 770b FIGURE 2 Mesothelioma Mortality Death Rate Per 1000 Person-Years Number of Years Subsequent to Onset of Exposure Deaths per thousand person-years of experience of pleural, peritoneal, and total mesothe lioma among 17,800 asbestos insulation workers from 1967-1976, analyzed by duration from onset of employment in five-year periods. Ratios between observed and expected deaths cannot be computed since expected rates are not available for the general population. At least as of 4S years from onset, a decline in rates was not noted. This is consistent with the finding that cigarette smoking did not play a role in the risk of developing mesothelioma, in contrast to lung cancer. Asbestosis Mortality Death rates of asbestosis among 17,800 asbestos insulation workers 1967-1976, analyzed in five-year periods of duration from onset of employment. Some decline is seen after 45 years from onset, possibly related to the added burden of smoking-induced lung disease superim posed upon asbestosis (at least in some cases) with selective survival of nonsmokers. 24 iV ',7793 If the general death rate for men is G, and the number of asbestotics is A, then the total number of deaths in asbestotics is D, = 2.8 x G x A. A is the number of interest. Combining the above equations, A = M,/(0.2I3 x 2.8 x G). Ma in available for every age and future year from the mesothelioma projections (Task 2) and G is estimable for each age group from current vital statistics data. SelikofTs data apply to insulation workers exposed to asbestos essentially all their working lives, and cannot, therefore, not be expected to give a reasonable estimate of asbestosis mortality far into the future, after the workplace has been largely cleared of significant asbestos exposure. (Mesothelioma mortality, by contrast, is affected almost entirely by age at first heavy exposure, and is not changed greatly by workplace clean-up, at least as far as concerns already exposed workers.) The problem is that discussed at the end of Task 4a: new asbestosis probably stops occurring (with some lag, perhaps 10 years) after the cessation of asbestos exposure. Thereafter mesothelioma/asbestosis relations observed previously (under conditions of extended exposure) became inapplicable to future projection. As in Task 4a, we have handled this problem by estimating prevalence based on continued new incidence through the 1980-1984 quinquennium, and have estimated subsequent prevalence by aging the 1980-1984 population as described in Task 4a. Table 18 gives the projected asbestosis prevalence figures for U.S. males 1980-2009, using this second projection procedure. TABLE 18 Projections of the Number of Prevalent Cases of Asbestosis in U.S. Males 1980-2009 Based on the Equivalence of Asbestosis and Mesothelioma Mortality Rates Years Number of Men Alive With Asbestosis 1980-1984 1985-1989 1990-1994 1995-1999 2000-2004 2005-2009 64.000 45,300 31.000 19,700 11,400 5,700 Task 4c: Other methods of projecting asbestosis prevalence. There are several "quick and dirty" estimates of asbestosis prevalence which give estimates of the same order of magnitude as one another. 1. Berry (1981) reports that 133 workers were certified annually by U.K. pneumoconiosis panels in 1973-1976 and that the median survival of the least disabled certified workers was 15 years. This gives a maximum steady-state prevalence of 15 x 133 = 2000 workers in the U.K. with certified pneumoconiosis (essentially all asbestosis). The U.S. is'about four times the size of the U.K. Histori cal exposure patterns being equal, this implies the existence of about 8,000 asbestotic workers or former workers in the U.S. Apart from the looseness of the analogy between the U.K. and the U.S., this projection suffers from its dependence on the number of workers actually certified in the U.K. This is a lower limit to the number actually ill. 2. Burnham (1982) cites an unpublished estimate of the National Center for Health Statistics (NCHS) that there were 427,000 (range 248,000 to 606,000) pneumoconiosis sufferers in the U.S. in 1980. He also points out that the Mortality Statistics Branch of the NCHS noted 1422 deaths ascribed to pneumoconiosis in the U.S., of which 72 were ascribed specifically to asbestos. Applying the death proportionality to the pneumoconiosis prevalence gives an estimate of (72/1422)(427,000) = 21,000 asbestotics (range 12,000 to 30,000). The pneumoconiosis prevalence figure, however, was based on a 25 h ; 7797 questionnaire only and is therefore likely to be an underestimate, and the reporting of asbestosis on death certificates is notoriously low. 3. The prevalence of x-ray changes in workers listed in Table 17 can be multiplied by our estimates of the size of the heavily exposed work force (Table 9), to obtain estimates of asbestosis prevalence ranging from about 18,000 to 150,000 depending on whether radiologic grade 1, 2, or 3 is used as the minimal criterion for a diagnosis of asbestosis. Although the correlation between x-ray changes and symptomatology is imperfect, the low end of this projection (18,000 grade 3 cases) would certainly represent symptomatic individuals in every case. The upper end of the projection (150,000) would include many people with few or no symptoms, whose asbestosis would be detectable by physical or radiologic examination only. Task 4d: Derive a general methodology for predicting lawsuits as a function of asbestosis prevalence. Although it is not the purpose of the present work to derive estimates of a person's propensity to bring suit given that he has disease, the difference between asbestosis and the cancers insofar as diagnosability and survival times are concerned calls for some comment. Mesothelioma and lung cancer come to diagnosis fairly rapidly and reliably, and as a result the propensity to sue can be related directly to disease incidence in order to derive an expected number of lawsuits. Asbestosis is not diagnosed nearly as reliably or as quickly, so that the pathway leading from prevalent disease to a lawsuit involves two probabilistic steps: diagnosis and decision to sue. Men with asbestosis live for decades, and so may be diagnosed for the first time and sue years after the onset of diagnosable disease. Perhaps the best way to interpret overall propensity to sue for an asbestotic man is to calculate an annual probability of suing. The pool of prevalent, asbestotic men who are potential litigants can then be thought of as being diminished with the passage of time through two effects: first, through their own mortality, and second, through their bringing suit, thus removing themselves from the pool of potential litigants by becoming active litigants. Over years, then, a constant propensity to sue acts on a dimin ishing pool of potential litigants to produce a declining annual number of lawsuits. By way of illustra tion, Table 19 gives the expected number of lawsuits by quinquennium, assuming asbestosis prevalence pools as listed in Tables 16 and 18, with cases appearing (and being removed from the pools of potential litigants) at a rate corresponding to suits from 9% of all potential litigants appearing each year. TABLE 19 Projections of Asbestosis Lawsuits Assuming 9% of Prevalent Cases (Non-Litigants) Bring Suit Each Year Yore Method Of Projecting Asbestosis Mcso Incidence in Asbestotia Meso>Asbestosis Mortality Equivalence 1980-1984 1985-1989 1990-1994 1995-1999 2000-2004 2005-2009 24,800 8,300 2,800 900 200 100 24,100 10,600 4,500 1,800 700 200 The 9% figure was chosen for Table 19 so as to yield current lawsuit rates for the present quin quennium. Bear in mind that the rapid decline of lawsuits in Table 19 is the product of two factors, which hold true only for workers with symptomatic asbestosis. The first is the high mortality rate in these men, discussed in Task 4a; the second is the finite (though large) size of the pool of potential litigants. From Table 17 it should be apparent that the number of workers with minimal disease 26 A i 7703 (Grade I) is very much larger than the number of seriously affected workers. If in fact a large number of lawsuits derive from the relatively well, exposed population, neither of the conditions on which Table 19 is predicated would hold: mortality in the minimally diseased is not much elevated, and the number of minimally diseased persons is so large that current litigation rates will not result in any meaningful depletion of the pool of potential litigants. Limited J-M data suggest that, in fact, lawsuits from asbestotics do not decline as rapidly after last exposure as one would anticipate from Table 19. In effect. Table 19 represents minimal projection. If only a fraction of current cases are coming from symptomatic cases, then the symptomatic pool is being depleted more slowly than projected in Table 19. Symptomatic cases will come in over a longer period, and the remaining, minimally diseased cases will continue to flow in at a rate (determined by socio-legal factors) which is unlikely to be bounded by purely medical or epidemiological factors such as population mortality rates or depletion of a pool of injured workers. Taking all of these factors into consideration, a reasonable central projection of the number of lawsuits seen from 1982 on is likely to be about 45,000, with a reasonably Arm lower bound of 30,000 and a very indefinite upper bound on the order of 120,000. Task 5: Estimate the amount of asbestos-related disease occurring in women. Approximately 5% of the lawsuits being filed with J-M derive from women. This number is consistent with an estimate from a variety of sources of about 10% of the asbestos-exposed World War II workforce being female, with a diminished fraction after the war. Exposed female workers still alive can be expected to have an age-at-first-exposure distribution which is even more concentrated in the war years than that of men. The consequence of this is that asbestos-related disease will have reached its peak in women earlier than in men, and is probably past that point already. The number of female cases in J-M is too small to make a direct test of this hypothesis. Probably the most reasonable projection of female cases would involve accepting the current 5% figure and projecting the proportion of female cases to taper off gradually to a negligible number by the year 2000. 27 A 770J References Baris, Yl, Artvinli, M and Sahin, AA. (1979) Environmental mesothelioma in Turkey. Annals New York Academy of Sciences 339:423-432. Berry, G. (1981) Mortality of workers certified by pneumoconiosis medical panels as having asbestosis. British Journal of Medicine 35:130-137. Breslow, N. (1982) Unpublished review of NCI data. Burnham, CE. (1982) Unpublished letter to Norman Breslow, June 2, 1982. Chase, G. (1981) Preliminary results from a morbidity study of domestic J-M asbestos-using locations. Johns-Manville Corporation internal document, July 30, 1981. Chovil, AC, McCracken, WJ, Dowd, EC, et al. (1981) Occupational cancer: experience in Ontario. Canadian Medical Association Journal 723:1237-1241. Connelly, RR. (1980) Mesothelioma incidence in the United States. Presented at the Workshop on Opportunities for Research and Education in Asbestos-Related Disease, Bethesda, MD, June 17, 1980. Edge, JR. (1979) Incidence of bronchial carcinoma in shipyard workers with pleural plaques. Annals New York Academy of Science 330:289-294. Elmes, PC and Simpson, MJC. (1976) The clinical aspects of mesothelioma. Quarterly Journal of Medicine ALK:427-449. Finkelstein, M, Kusiak, R, and Suranyi, G. (1981) Mortality among workers receiving compensa tion for asbestosis in Ontario. Canadian Medical Association Journal 725:259-262. Henderson, VL and Enterline, P. (1979) Asbestos exposure: Factors associated with excess cancer and respiratory disease mortality. Annals New York Academy of Sciences 330:117-126. Hobbs, MST, Woodward, S, Murphy B, et al. (1980) The incidence of pneumoconiosis, mesothe lioma and other respiratory cancer in men engaged in mining and milling crocidolite in Western Australia. In: Biological Effects of Mineral Fibres, J. C. Wagner (ed.), vol. 2, Lyon IARC, pp. 615625. Hughes, J. and Weill, H. (1980) Lung cancer risk associated with manufacture of asbestoscement products. In: Biological Effects of Mineral Fibres, J. C. Wagner (ed.), IARC Scientific Publication No. 30, IARC, Lyon. Kolonel, LN, Hirohata, T, Chappell, BV, et al. (1980) Cancer mortality in a cohort of naval shipyard workers in Hawaii: Early findings. Journal of the National Cancer Institute 64:739-743. McDonald, AD and McDonald, JC. (1973) Epidemiologic surveillance of mesothelioma in Canada. Canadian Medical Association Journal 709:359-362. McDonald, AD and McDonald, JC. (1980) Malignant mesothelioma in North America. Cancer 46:1650-1656. McDonald, JC, Liddell, FDK, Gibbs, GW, et al (1980).Dust exposure and mortality in chrysotile mining, 1910-1975. British Journal of Industrial Medicine 37:11-24. Newhouse, M. (1981) Epidemiology of asbestos-related tumors. Seminars in Oncology 5:250257. Newhouse, ML and Berry, G. (1976) Predictions of mortality from mesothelial tumors in asbestos factory workers. British Journal of Industrial Medicine 33:147-151. Newhouse, ML and Berry, G. (1979) Patterns of mortality in asbestos factory workers in London. Annals New York Academy of Sciences 330:53-60. Newhouse, ML and Thompson, H. (1965) Mesothelioma of the pleura and peritoneum following exposure to asbestos in the London area. British Journal of Industrial Medicine 22:261-269. 28 h :7U0C Peto, J, Doll, R, Howard, SV, et al. (1977) A mortality study among workers in an English asbestos factory. British Journal of Industrial Medicine 54:169-173. Peto, J, Henderson, BE and Pike, MC. (1981) Trends in mesothelioma incidence in the United States and the forecast epidemic due to asbestos exposure during World War II. In: Quantification of Occupational Cancer, R. Peto and M. Schneiderman (eds.), Banbury Report 9, Cold Spring Harbor Laboratory, pp. 51-69. Peto, J, Seidman, H and SelikofT IJ. (1982) Mesothelioma mortality in asbestos workers: Implica tions for models of carcinogenesis and risk assessment. British Journal of Cancer 44. SelikofT, IJ, Hammond, EC and Seidman, H. (1980) Latency of asbestos disease among insultation workers in the United States and Canada. Cancer 46:2736-2740. Surgeon General (1978) Smoking and Health. U.S. Dept, of Health Education and Welfare. Tagnon, I, Blot, WJ, Stroube, RB, et al. (1980) Mesothelioma associated with the shipbuilding industry in coastal Virginia. Cancer Research 40:3875-3879. Vianna, NJ, Maslowsky, J, Roberts, S, et al. (April 1981) Malignant mesothelioma: Epidemio logic patterns in New York State. New York State Journal of Medicine pp. 735-738. 29 A : 780 1  -'t>nrE'y,i=. nz-s~'*% = s' g SQS.s.g o =