To be effective, EPA's proposed new guidelines must be integrated into a comprehensive framework for risk management.
Risk is the coin of the realm in environmental, health, and safety regulation, and the nation is currently caught up in a heated debate about the uses and limitations of risk assessment and the benefits and costs of risk-reduction options. The Environmental Protection Agency's (EPA) proposed new guidelines for risk, as summarized in the accompanying article by Huggett and Wiltse, take an important step in the right direction by moving to incorporate emerging scientific advances. I point out some areas in which the guidelines can be improved, but the more important consideration is how to use them to improve risk management.
The first EPA guidelines, issued in 1976, were updated in 1986, largely in response to the 1983 National Research Council (NRC) report Risk Assessment in the Federal Government: Managing the Process (popularly known as the Red Book). The current proposal incorporates many of the recommendations of the 1994 NRC report Science and Judgment in Risk Assessment. The Red Book as well as the preceding work of the Interagency Regulatory Liaison Group and the Office of Science and Technology Policy in the Carter administration sought to stimulate development of more scientific evidence about chemical mechanisms and exposure patterns and individual variation in susceptibility. It was hoped that standard default assumptions and conversion factors would be overridden by chemical-specific data and new knowledge about rodent/human similarities and differences. It has been a slow process, both in eliciting data sets with sufficiently firm conclusions and in changing regulatory agency risk assessments.
Scientists in the many fields on which risk assessments draw are confident that understanding modes of action, variation in metabolism, pathways of exposure, effects of genes and nutrition, and new testing strategies, such as use of transgenic mice, will substantially enhance our ability to evaluate the carcinogenicity of chemicals. The proposed guidelines would use such information in an attempt to override routine extrapolations to the extent justifiable, thus strengthening the science base for regulation. To help users understand the risk assessments, EPA is requiring that each assessment be presented as a two-page narrative that sums up the supporting science rather than as a simple classification.
The plan to provide detailed characterizations of the evidence at each step in the risk assessment is wise. The differentiation of the dose-response relationship into what is observable and what is an extrapolation is also laudable. We must find ways to help all interested persons realize that the scientific method of observation and experimentation is at play in the observable range, which means studies in which there is at least 10 percent incidence of cancers, birth defects, or other adverse consequences. This rate of incidence is necessary to detect a statistically significant increase in incidence in comparison with that in unexposed groups of people or animals. In contrast, the extrapolation to upper-bound, lifetime risk estimates of 1 in 100,000 or 1 in 1,000,000 persons goes far outside the observable range and rests on scientific inference, assumptions, models, and speculation. Even in occupational settings, where lifetime risk estimates from chronic exposures may be on the order of 1 in 1,000, the small number of workers actually exposed to particular chemicals and the limited periods of exposure and follow-up put heavy burdens on inferences used in the epidemiologic methods.
One of the more controversial aspects of the proposed guidelines will be the definition of the three risk categories into which all substances will be grouped. Despite EPA's emphasis on the need for narratives to adequately communicate the nature of risk and to …