Assessing Input Distributions for Monte Carlo Analyses in Quantitative Risk Assessment. B. Binkowitz and D. Wartenberg, EOHSI, UMDNJ-RW Johnson Medical School, 681 Frelinghuysen Road, Piscataway, NJ 08855-1179
Monte Carlo simulations are becoming commonplace in quantitative risk assessments (QRA). Designed to propagate the uncertainty associated with each individual exposure input parameter in a QRA, Monte Carlo methods statistically combine the individual parameter uncertainty distributions to yield a single, overall uncertainty distribution. Critical to such an assessment of this method are the accuracy, precision and representativeness of the individual distributions. In this paper, we review and compare the uncertainty distributions used in published QRAs. We selected those parameters used in Monte Carlo simulations in at least five published articles. These are: body weight, food consumption, soil ingestion rates, breathing rates, and fluid intake. To provide a basis of comparison, we evaluated all exposure parameters with respect to five properties: consistency, accuracy, precision, specificity, and the appropriateness of the distribution. Different investigators used different data sources and different distributional shapes which has important implications for comparability of QRAs.
Work supported by the New Jersey Department of Environmental Protection and CRESP, the Consortium for Risk Evaluation with Stakeholder Participation.