Probabilistic prediction of exposures to arsenic contaminated residential soil

Probabilistic modelling using Monte Carlo simulation has been proposed as a more scientifically valid method of estimating soil contaminant exposures than conservative deterministic methods currently used by regulatory agencies. A retrospective application of probabilistic modelling to an exposure scenario involving arsenic-contaminated residential soil near the former ASARCO smelter near Tacoma, Washington is presented. The population of interest is children, aged 2–6 years, living within one-half mile (0.3 km) of the smelter site. Models that predict urinary arsenic levels based on unintentional soil ingestion and inhalation exposure pathways are used. Distributions of exposure variables are based on site-specific data and previous exposure studies. Simulated urinary arsenic levels are compared with data from two biomonitoring studies performed during the late 1980s. Arsenic distributions produced by simulation and biomonitoring are significantly different, and likely contributors to this difference are discussed. However the probabilistic model provides closer estimations of urinary arsenic levels than conservative deterministic models similar to those used by regulatory agencies, and provides useful information regarding parameter uncertainty. Soil ingestion rate was a driving variable in the probabilistic models. Further quantification of soil ingestion rates is warranted.