Uncertainty vs. Interindividual variability

Distinct treatment of uncertainty and interindividual variability in variates used to model risk ensures that quantitative assessments of these attributes in modeled risk are maximally relevant to potential regulatory concerns. For example, such a distinction is required for quantitative characterization of uncertainty in population risk or in individual risk. Yet, most quantitative uncertainty analyses undertaken as part of environmental health risk assessments have failed to systematically maintain this distinction among modeled distributed input variates, and so have had limited relevance to reasonable concerns that regulators may have about how uncertainty and variability ought to relate to risk acceptability. The distinction is of course impossible if quantitative treatment of distributed input variates is rejected in favor of using single-point estimates due to the perceived impracticality of complex Monte Carlo analyses that might erroneously be thought of as being necessarily involved. Here, some practical methods are presented that facilitate implementation of the analytic framework for uncertainty and variability proposed by Bogen and Spear. Two types of methodology are discussed: one that facilitates the distinction between uncertainty and variability per se, and another that may be used to simplify quantitative analysis of distributed inputs representing either uncertainty or variability. A simple and a complex form for modeled increased risk are presented and then used to illustrate methods facilitating the distinction between uncertainty and variability in reference to characterization of both population and individual risk. Finally, a simple form of discrete probability calculus is proposed as an easily implemented, practical alternative to Monte-Carlo based procedures to quantitative integration of uncertainty and variability in risk assessment.