Predictive Bayesian dose-response assessment for appraising absolute health risk from available information

Currently, no general measure of population health response to untestable doses of chemicals and microbes has been established that accounts for uncertainty quantitatively and indicates relative toxicity or virulence directly. Untestable doses include those corresponding to the 2.74 × 10^-7 illnesses per exposure expressed in goals of the U.S. Environmental Protection Agency's Surface Water Treatment Rule, and doses of human bioterror agents. For example, it is shown that relative Benchmark Dose (BMD) values depend upon the level of confidence assumed. Because of the lack of scientific basis for this level, BMDs are not comparable among health stressors for untestable doses and stressors. In this paper a new predictive Bayesian method is proposed for absolute and relative dose-response assessment based on available information. Information may include toxicological judgment, epidemiological statistics, genetic information, related data, and numeric dose-response data. Results for rotavirus indicate a "safe dose" of 6.3 × 10^-7 focus-forming units/exposure, approximately one-half log above the dose corresponding to the maximum risk for any pathogen assuming a 100% infection rate. The result further indicates the limited value of data in refining the assessment, due to the inability of data to reduce variability. The method is suggested for assessing risks of new and existing chemicals and pathogens, as a basis for prioritizing expenditures for protection against environmental and terrorist threats.