Comprehensive Realism's Weight-of-Evidence Based Distributional Risk Characterizations for Cancer and Noncancer Health Effects. R. L. Sielken Jr., Sielken, Inc., 3833 Texas Avenue, Suite 230, Bryan, Texas 77802
Challenges to the use of over-simplified exposure characterizations, the use of low-dose linearity and other default assumptions in cancer dose-response assessment, and the use of default upper-bound values for the uncertainty factors along with lower-bound NOAELs, LOAELs, and Benchmark Doses in the evaluation of reference doses or margins of exposure for noncancer or cancer health effects have stimulated the continued evolution of risk assessment methodologies as have the increasing need for more realistic estimates of exposure and the dose-response relationship, better uncertainty characterization, and greater utilization of cost-benefit analyses. "Comprehensive realism" is an emerging quantitative weight-of-evidence based risk assessment methodology for both cancer and noncancer health effects which utilizes probability distributions and decision analysis techniques to reflect all of the available human exposure data, all of the available human and animal dose-response data, and the current state of knowledge about the relative plausibility of alternative exposure and dose-response analyses. Rather than using default constants for the components of exposure equations and models, distributional characterizations of exposure are developed using probability distributions to characterize the variability in exposure parameter values and the relative likelihood of different exposures. Separate distributional characterizations are developed and presented for each subpopulation and combined population. A tree (like a decision tree and a probability tree) is used to decompose the dose-response assessment into component factors and to provide a structure for explicitly considering multiple alternatives for each factor. The tree structure can explicitly incorporate multiple health effects, biological mechanisms, dose scales, dose-response models, data sets, interspecies extrapolation procedures, uncertainty factors, and distributional characterizations of uncertainty factors. The alternatives or branches in the tree can be weighted in order to explicitly incorporate the current state of knowledge about the relative plausibility of these alternatives and provide a basis for quantitative uncertainty analyses. Sensitivity analyses can be used to prioritize future research. Ground-breaking work has demonstrated the feasibility of weight-of-evidence based distributional characterizations and provided initial examples. Computer software implementations are available from Sielken, Inc.