Avoiding Pitfalls in Probabilistic Health Risk Assessments. R. C. Lee and J. C. Simpson, Golder Associates Inc., 4104 148th Ave., NE, Redmond, WA 98052
Probabilistic risk modeling, typically using Monte Carlo simulation techniques, is an increasingly popular way of quantifying the uncertainty associated with exposures to hazardous substances in the environment. While it is generally agreed that probabilistic modeling is more scientifically valid and appropriate than the conservatively-biased deterministic modeling currently used by many environmental regulatory agencies, misuse and misinterpretation of probabilistic models can lead to misunderstandings and bad decisions. Risk analysts and risk managers should carefully consider these "probabilistic pitfalls". The stakes involved in risks resulting from exposures to hazardous substances in the environment are high in terms of costs associated with public health, remediation, alternative risks and benefits, and litigation; therefore, good information for decisions is essential. Probabilistic pitfalls can occur in the 3 main phases of risk assessment: problem formulation, analysis, and interpretation; and range from simple mathematical errors to gross misinterpretation of results. This paper explores examples of pitfalls that the authors have encountered in "real life" risk assessments, and possible solutions to these pitfalls.