ESTIMATING EXPOSURE DISTRIBUTIONS - A CAUTION FOR MONTE-CARLO RISK ASSESSMENT

Monte Carlo studies of risk assessment commonly require estimates of e xposure distributions. The exposure distribution may be estimated by a ssuming the distribution follows a specified functional form (lognorma l), and estimating parameters of the assumed distribution from observe d sample exposure data. Alternatively, to avoid the distributional ass umption, the exposure distribution may be estimated directly from the observed exposures measured on a sample of subjects. We discuss proble ms with this second approach for estimating exposure distributions whe n exposures are measured with error. Specifically, we show that when t he true exposure varies from day to day, or the observed exposure diff ers from the true exposure due to measurement error, then the tails of the observed exposure distributions will be biased, with the magnitud e of the bias increasing toward the tails of the distribution. The bia s may be severe, and lead to overestimation of upper percentile exposu re. The size of the bias is directly related to the magnitude of respo nse error. Alternative estimators are discussed that frequently provid e closer estimates of a subject's true exposure. Issues regarding choi ce of estimator, and consequence for exposure distribution estimation are discussed in the context of estimating soil ingestion in children. The biases are illustrated via simulations.