Scientific trends in risk assessment research.
The use of risk assessment approaches to evaluate the effects of toxic chemicals had its primary origin in 1976 when the U.S. Environmental Protection Agency (EPA) adopted the first federal guidelines to commit a major regulatory agency to risk assessment approaches for the evaluation of suspect carcinogens. The accompanying policy statement also adopted a risk management policy, which acknowledged that the agency would accept risk in making public health policy decisions; in essence, this represented a primary departure from the zero-risk goal that had dominated the first half of the environmental movement of the 70s. The approach adopted in 1976 was based on the experience of risk assessment approaches used for assessing low-dose-radiation effects on human health. To be certain that no public health risk be underestimated, particularly in light of the prior zero-risk goal, the practice of risk assessment for the first decade relied heavily on extremely protective assumptions in all aspects of the assessment process. For example, these assumptions included ranking the weight-of-evidence according to results in all related studies without regard for distinctions of tumor end point relevance to humans and the possibility that low-dose effects at environmental exposures might not be the same as high-dose effects. For low-dose-response characterization, the practice has been to characterize a plausible upper bound on risk by the use of a low-dose linear non-threshold dose-response curve. Exposure assessment likewise relied on maximum plausible assumptions to characterize exposure. More recently, more attention is being paid to developing accurate scientific databases that advance almost every aspect of the risk assessment process in the direction of more accurate risk characterization. This paper will discuss the recent trends in weight-of-evidence characterization, dose-response modeling, and exposure assessment and will compare the outcomes of these refined assessments to those evaluations that have relied on the earlier, conservative approaches. In essence, if the practices of the first decade for establishing plausible upper bounds on the risk were accurate, improved scientific data by and large should be expected to lower the overall theoretical risk. Indeed, this is the case when recent risk assessment research is applied but there are examples to the contrary. This paper represents a survey of recent trends and applications.