Tobacco Documents Online

I I I I I I I I I I I I I I I I I f' Ku /'7 ~ %frl~. ,pi i w~c~ cT.TMc c.11#'onNMEN`TAL PROTECTION AGENCY NATIONAL CENTER FOR ENVIRONMENTAL ASSESSMENT WASHINGTON, DC 20460 JAN 3 1 1996 Dan Mulhollan, Director Congressional Research Service Library of Congress Washington, DC 20540 Dear Dr. Mulhollan: OFFM OF aEss*acri uM oEVe.opMEKr I appreciate the opportunity we had to provide comments to the CRS Workshop Draft "Environmental Tobacco Smoke and Lung Cancer Risk" (June 5, 1995), and it appears that some of the comments provided by us and other reviewers were incorporated into the final report of November 14, 1995. We were also pleased to note that, using the nonthreshold model, CRS's best estimate of approximately 2,800 lung cancer deaths annually in U.S. nonsmokers from ETS exposure was similar to the estimates of b".h the EPA (1992) and the National Research Council (1986). Moreover, CRS's acknowledgement of the potentially large public health impact of ETS on heart disease in calling for additional research and assessment was a welcome addition. However, the final report does not redress some of our major concerns, and there are also new sections with which we do not concur. As a general comment, while the EPA risk assessment was a comprehensive analysis of the total weight of evidence, the CRS report selectively focuses on a few recent U.S. studies. We also believe that the CRS report repeatedly exaggerates uncertainties and is far too equivocal in drawing conclusions. This is especially true in the face of the consistency of 17 of 17 studies observing elevated lung cancer risks in the highest exposure groups and of the consistent and uncontroversial increases in lower respiratory tract infections observed in infants and young children exposed to ETS. Some specific concerns are outlined below. First, we believe the treatment of possible bias in the analysis of the epidemiology studies on ETS and lung cancer is grossly unbalanced. The CRS report overemphasizes uncertainties arising from hypothetical upward biases, while giving inadequate attention to the real downward bias created by exposure misclassification. ROCydedftcyolabN . PAnbd wtA VpetabM 09 Based WO ae 100% Re.Yded Papw (40% PoIConwnM)

- 2 - Notably, the new analysis of "consequences of exposure misclassification" (CRS-38) is seriously flawed. The analysis deals only with subjects "incorrectly counted as exposed" and conspicuously ignores the effects of exposure misclassification of "unexposed" subjects. Exposure to the "unexposed" subjects results from widespread sources of ETS other than spousal smoking, for example, workplace and social exposures. In fact, few people are truly unexposed to ETS. Both types of exposure misclassification exert a strong and very real downward bias on the observed relative risk estimates, because significant differences in exposures between comparison groups are necessary to detect a difference in risk. (That::is why EPA considers the results in the highest exposure groups, where the exposure differential is greatest, to be much more informative than the overall reported risks.) Second, as pointed out in our earlier comments, the threshold model proposed by CRS is not supportable. It relies on alleged nonlinearities in some of the exposure-response relationships at low exposures in the results of a few epidemiology studies. These studies are too insensitive to convey such conclusions, especially given the exposure misclassification problems described above. (The exposure differential is likely to be negligible between a low spousal exposure group and the "unexposed" group, and one would not expect to discern an increased risk from low spousal exposures.) In addition to the inability of the epidemiology studies to establish a threshold due to their lack of sensitivity, there is no demonstrated mechanistic basis for drawing such a conclusion. It is indefensible to presume a nontrivial threshold for a genotoxic carcinogen (i.e., a carcinogen that can irreversibly damage DNA, at conceivably minute exposures) that has been shown, despite the difficulties of exposure misclassification and low study power, to cause lung cancer at typical environmental exposure levels. Even if,one were to suppose a threshold, the methodology used by CRS is flawed, as well. As discussed in our earlier comments, the methodology incorporated no margin of safety to account for the lack of precision regarding cut-points, measurement errors, and variabilities in population susceptibilities to lung cancer. Furthermore, the assumption that there is no risk from "'background' exposures only" is not realistic, even for an erroneous threshold model. Some occupational exposures (e.g., bars, barbershops, and small, enclosed offices) are likely to be comparable to, or greater than, even high spousal smoking exposures and can be expected to convey a significant and preventable lung cancer risk. I I I I I I I I I I I I I I I I I

r r - 3 - Third, the new section on childhood respiratory effects (Appendix B) overemphasizes uncertainties and does not reflect the strength and consistency of results and the scientific consensus regarding the conclusions of health effects. For example,-the CRS report neglects to mention that the 1986 National Research Council and Surgeon General reports reached virtually the same conclusions as EPA. In summary, while the final report is an improvement over the draft, the overemphasis on uncertainties and the proposal of an unsupportable threshold model for environmental tobacco smoke risk severely limit its credibility. Thank you for your attention to these comments. Sincerely, Michael Callahan, Director National Center for Environmental Assessment, Washington Office I I I