Determining risk severity is one of a risk manager’s first tasks. On this basis, he or she must decide whether a given risk is unacceptable, hence must be reduced or prevented with monitoring, or can be accepted, at least for the present. The risks associated with possible alternatives must be considered, as well. For example, the alternative to a pesticide that kills x birds per year may be clean farming which destroys refugia for pests but also eliminates habitat for y birds. Even if y is much larger than x, clean farming may be more desirable because it is more readily embraced by environmental and animal rights groups. The integration of such considerations is nearly always an informal process.
Our single brief example is presented as if there is only one exposure model, one effects model, and one set of assumptions about how to parameterize them and combine them into a risk estimate. This approach is generally applicable to human health risk assessments where endpoints, models, and treatment of parameters are standardized. However, it is not appropriate for ERAs. One might estimate effects on fisheries with an ecosystem model, a population model, and a statistical model of an organism‐level toxicity test. Ecosystem‐level effects might be represented by results from a microcosm test, a lake ecosystem model, and a stream ecosystem model. Each risk estimate will have its own assumptions and associated uncertainties and those uncertainties may not be expressed equivalently. The separate lines of evidence must be evaluated, organized in some coherent fashion, and explained to the risk manager so that a weight‐of‐evidence evaluation can be made.
One approach to presenting alternate lines of evidence is to present them graphically of common axes. The axes should be chosen from the four dimensions of risk estimation so as to clarify the differences in the risk estimates that are most important to the decision. That is, does the decision depend primarily on the concentration at which an effect occurs, the time to recovery, the number of organisms dying, or some combination of axes? The output of the alternate models, assumptions, or data should be plotted on those axes.
Leave a Reply