Using Monte Carlo simulations in public health risk assessments: estimating and presenting full distributions of risk

Abstract

With desktop computers as powerful as mainframes were just a few years ago, analysts can now use commercial software to estimate full probability distributions for--not just point estimates of--health risks experienced by people chronically exposed to toxic chemicals at or near hazardous waste sites. Even though probability is the central concept in risk assessment, and even though probabilistic methods offer strong advantages and insights as compared to the "deterministic" methods now required by U.S. Environmental Protection Agency's guidance manuals, analysts have only begun to use probabilistic methods at Superfund sites. In this paper, we examine a simplified case study using Monte Carlo methods to estimate full distributions of public health risk. We demonstrate the use of "toggles" to isolate the contributions of different inputs, and we also offer new graphical methods to communicate the results to risk managers and concerned citizens.