Assessing multimedia/multipathway exposures to inorganic arsenic at population and individual level using MERLIN-Expo
Résumé
In this study, we report on model simulations performed using the newly developed exposure tool, MERLIN-Expo, in order to assess inorganic arsenic (iAs) exposure to adults resulting from past emissions by non-ferrous smelters in Belgium (Northern Campine area). Exposure scenarios were constructed to estimate external iAs exposure as well as the toxicologically relevant As (tAs, i.e., iAs, MMA and DMA) body burden in adults living in the vicinity of the former industrial sites as compared to adults living in adjacent areas and a reference area. Two scenarios are discussed: a first scenario studying exposure to iAs at the aggregated population level and a second scenario studying exposure at the individual level for a random sub-sample of subjects in each of the three different study areas. These two scenarios only differ in the type of human related input data (i.e., time-activity data, ingestion rates and consumption patterns) that were used, namely averages (incl. probability density functions, PDFs) in the simulation at population level and subject-specific values in the simulation at individual level. The model predictions are shown to be lower than the corresponding biomonitoring data from the monitoring campaign. Urinary tAs levels in adults, irrespective of the area they lived in, were under-predicted by MERLIN-Expo by 40% on average. The model predictions for individual adults, by contrast, under-predict the biomonitoring data by 7% on average, but with more important under-predictions for subjects at the upper end of exposure. Still, average predicted urinary tAs levels from the simulations at population level and at individual level overlap, and, at least for the current case, lead to similar conclusions. These results constitute a first and partial verification of the model performance of MERLIN-Expo when dealing with iAs in a complex site-specific exposure scenario, and demonstrate the robustness of the modelling tool for these situations.