Document QXn1YjyYpXJZxjaJYMddQKom8

22640 Federal Register / Vol. 51, No. 119 / Friday. June 20, 1986 / Rules and Regulations Table 3.8ummarizes the results of the goodness-of-fit iests for OHSA's model and the multistage model with one, two or. three stages, fpr each of the data sets used by OSHA and for two additional sets of data. Consideration of the results in Table 3 show that, in fact, in four of the six cohorts, the best fitting model was linear with asbestos concentration (i.e..either the OSHA model or the multistage model with one stage showed the best fit. For the Finkelstein data, the multistage model with two stages Fit only slightly better than the linear model, P=0.99 versus P=0.97). For three of the six data sets, the OSHA model fit as well or better than.the multistage model! Although the fit of the OSHA model was adequate for the Finkelstein data, the OSHA model did not fit as well as the multistage model (P=0.39 (OSHA) versus P=0.99 [Crump]). And with regard to the Weill data, the fit of. . the OSHA model was inadequate (P=0.00i) and the three-stage multistage model provided an excellent fit to the data (P=0.90). Similarly, as reported by Dr. Crump, the fit of the OSHA model to the Selikoff et al. data was "marginal", and the multistage model with one doserelated stage provided a very good fit to the data (P=0.76). Implications of the goodness-of-fit tests on the selection of the individual estimates of Ku will be discussed in the next section. On the basis of these results, OSHA believes its choice of a risk model for mesothelioma is scientifically responsible. As discussed above, the model has received support from a large number of regulatory agencies, scientific bodies, and individual experts in risk assessment. Moreover, as will be seen in the next section, estimates of the individual Ku derived from this model are reasonable (and perhaps low), and represent the best estimate of the mesothelioma risk posed by exposure to asbestos. B. Data Usedfor the Calculation of Individual Ku-s. In the November proposal, OSHA used four studies judged by the Agency to have data adequate for the quantification of mesothelioma risk [Selikoff'et ail., Exs. 84-1701 84-90; Seidman et al.. Ex. 84-87, 84-170; Peto et al., Ex. 84-170, . and Finkelstein, Ex. 84-240). As Dr. Nicholson pointed out at the hearings; These were the four studies that did . . provide sufficient information that could be utilized. What is necessary is not simply the number of deaths in a particular study, but one has to know the time of those deaths; because the (fit) that was made involves the unexpected, particularly in light of the matching of the equation that's given there, higher average exposure found upon risk according to time per months of exposure, with data on mesothelioma risk at different times from onset of exposure in a defined population. We had to know the number or cases per person-years of risk (Tr. 6/19. p. 1-121-122). reexamination of the data. Dr. Crump'ssecond major objection to ' the use of these studies relateds to the issue of differential risk by fiber type. At the hearing, Dr. Crump noted that-- OSHA believed that these four studies were particularly appropriate studies for inclusion in the calculation of KM because of the large numbers of mesothelioma deaths observed in these * |T]urning to the risk specifically due to mesothelioma. I feel there is strong evidence that the risk in humans at least is less from chrysotilc exposure than from amphibole exposures. OSHA estimated risks from four studies, each of which involved either four studies (180.14, 7, and 11. exclusive or considerable exposures to respectively). It should be noted that omphiboles. these four studies are the same four studies employed by CHAP in its analysis of mesothelioma risk from asbestos exposure [Ex. 83-256.11--119-- 120). OSHA acknowledged in the preamble to the November proposal that its estimates of K* were derived from Although these estimates were adjusted downwards somewhat by comparing them with lung cancer estimates, they still are considerably larger than estimates made from populations exposed predominantly to chrysolite which I have made [Tr. 7/9, p. 84). In his written testimony, Dr. Crump elaborated on this position: studies with four of the five highest KL * * I believe there is considerable data to values. OSHA noted that there may be indicate that chrysotilc is less risky [thanthe "some bias in examining the value of KM independent of the KL in the same studies because it is likely that these Km would tend to be slightly higher than those derived from other studies, due to the demonstrated high power of these studies to'detect risk" [48 FR 51125). To account for this bias in its analysis, OSHA arrived at an average amphiboles|. OSHA has already omitted from its risk calculation data from mining and milling operations, on the grounds that these exposures are not representative of thoBC in the populations of workers OSHA has responsibility to protect. J believe this . principle should also be applied to the chrysotile-amphibole question, and that risk to modem day workers, who are exposed almost exclusively to chrysolite, should be Ku by examining the ratios of KM to Kt. estimated from studies in which chrysolite This gave an estimate of KH of 1 x 10'a exposures predominate [Ex. 237A, p. 47). rather than the higher central values of In an effort to expand the data from 4.98 x 10 ~s rather than the higher which to calculate an overall KM, Dr. central values of 4.98 x 10 ~8 (the Crump calculated KM's for two arithmetic mean) and 2.91 x 10 *8 (the additional studies "for which exposures geometric mean). OSHA believed this were predominantly to chrysolite. adjustment to the KM value to be Theses are the Dement el al. study; appropriate to avoid serious where exposures were to only overestimation of the risk of chrysotlle, and the Weill et al. study, in mesothelioma. Dr. Crump raised a number of issues regarding the calculation of KM from these studies. As he had for the calculation of KM, Dr. Crump noted that the Seidman et al. and Selikoff et al. studies are "particularly inappropriate for risk assessment because of the lack of exposure data!' [Ex. 237A, p. 39). OSHA's reasons for accepting the data from these two studies and the justification for their use in quantitative risk assessment have already been discussed in Section I. In light of the new data received from the Seidman cohort, OSHA has revised its estimates of Kh. Using the data in Table 1 [Ex. 267A] and four points of observation, the Km from the updated study is 2.4 x 10'8, somewhat lower than the value for which 77% of the workers were exposed exclusively to chrysotile" [Ex. 237A, p. 40). The mesothelioma data for these two studies are found in Tables D and E. The Ku calculations for various models are found in Table 3. For the Dement et al. data found in Table 4, the model used by OSHA provided a much better fit to the data (P=0.67) than any of the multistage models, and gave a KM of 2.2 x 10~9, approximately five times lower than the KM of 1 x 10-8 K given in the proposal. Of the multistage models, all of which allowed showed good fit, the three-stage model gave a KM of 3.1 x 10~8, more than 10 times larger than that estimated by the OSHA model and three times larger than OSHA's expressed preferred estimate of risk. Dr. Crump calculated Km put forth in the proposal for the the ratio of KM/KL for the Dement et al. original Seidman study. This is not study (KM/KL=2.2 x 10-/0.042 = 5.2 x GLEASON-000888