A Bayesian approach to probabilistic risk assessment in municipal playgrounds.

A Bayesian risk assessment of the exposure of children up to 6 years of age to trace elements in the sandy substrate of municipal playgrounds in Madrid, Spain was carried out. As a first step, exposure variables and parameters were borrowed from US Environmental Protection Agency (EPA) databases. However, the use of these non-site-specific exposure data introduced a high arbitrariness in the assumed distributions, which, in the Bayes approach, translates into a high uncertainty for those distributions that is also reflected in the risk outcome. In order to reduce this uncertainty, site-specific values for children's body weight and exposure frequency were determined from three surveys of 75, 56, and 34 individuals, respectively, carried out in randomly selected playgrounds in Madrid. This information was used in a Bayesian approach to modify the prior distributions of exposure frequency and body weight adapted from the literature. As a result, the predictive distributions of risk values for arsenic and mercury presented a reduced arbitrariness and a significantly lower uncertainty than those obtained either from the prior distributions borrowed from the US EPA or from the limited data gathered in Madrid separately.