Subgroups exposed to systematically different elemental compositions of PM2.5

Identification of exposure subgroups is important for both health-based assessments where health effects are linked to the elemental composition of PM2.5 mixture to which participants are exposed, and for development of population exposure models where population exposures to PM2.5 mass are modeled generally using fixed site ambient monitoring. Here we demonstrate that workplace sources dominate PM2.5 mass in the upper end of the distribution for EXPOLIS participants in Athens, Basel, Helsinki and Oxford, resulting in poor performance of models that use ambient concentrations to predict exposures when predicting higher exposures, where adverse health impacts would be more likely. Further, since different microenvironments reflect differing contributions from local PM2.5 sources, personal PM2.5 exposures for participants whose exposures are dominated by different microenvironments show systematically different elemental personal compositions. Perhaps a more significant complication for epidemiologic associations is that the proportion of participants whose exposures are dominated by each microenvironment varies across the exposure distribution to PM2.5. Participants exposed predominantly in the outdoor or personal microenvironments are a greater fraction of the lower end of the PM2.5 exposure distribution while participants with dominant workplace environments are a greater fraction of the upper end of the distribution, with corresponding differences in elemental compositions of PM2.5 exposures across the exposure distribution.