Combining data from heterogeneous surveys for aggregate exposure: Application to children exposure to lead in France.

To assess human health risks related to the environment, it is necessary to aggregate exposure from multiple sources. The objective of this paper was to propose a relevant approach to combine data from heterogeneous populations and methodologies. Five different methods based on Monte-Carlo simulations were tested and compared. Differences were: taking into account or not stratification variable, timeline to assign exposure factors and concentration and way to account for concentration correlations. The methods were applied to estimate lead exposure from food, dust, soil, air, and tap water or French children aged between six months and three years old. Comparing results' uncertainty, it is recommended to 1) select a reference population representative of the target population, 2) select stratification variables to combine surveys, and 3) simulate a new population by randomly sampling individuals in the reference population and simultaneously assigning human exposure factors and environmental concentrations from other surveys in integrating correlations (MC1S). No difference was observed when taking into account correlations using vectors of determinist data from one survey or rank of correlations with the Iman-Conover method. Regardless the methods used to combine data, dust was the main exposure source, followed by soil and in a less extent by food. Exposures from air and tap water were found to be insignificant for most children. • Five calculation processes were tested to combine dietary and environmental surveys. • Resampling individuals and variables decrease the uncertainty. • Use stratification variables to combine surveys limits risk of error. • The tested methods to account for correlations between exposure factors gave similar results. • Dust and soil were main exposure sources of children in France. [ABSTRACT FROM AUTHOR]