Going beyond the single number: Using probabilistic risk assessment to improve risk management

Recent efforts aimed at improving risk characterization emphasize the importance of uncertainty analyses and endorse probabilistic techniques as useful tools for performing them. One feature of explicitly including uncertainty and variability in risk assessments is that subsequent decision-making analyses (e.g., benefit-cost analyses) can use distributions of risk as inputs. This paper describes how probabilistic techniques can be used to characterize uncertainty and variability in risk throughout the spectrum of analytical tools used to support decisions. Different types of decision criteria determine the appropriate treatment of uncertainty and variability in the risk assessment, as well as the types of other analyses that may be used, and the value of performing them. Key issues discussed include the need to develop default distributions, decision criteria, and good examples for the probabilistic approach, and the extent to which risk assessors should be involved in decision analyses for risk management. The current lack of guidance for risk assessors and risk managers is a major barrier to implementation of probabilistic methods.