Abstract

Recent epidemiological cohort studies of the health effects of PM2.5 have developed exposure estimates from advanced exposure prediction models. Such models represent spatial variability across participant residential locations. However, few cohort studies have developed exposure predictions for PM2.5 components. We used two exposure modeling approaches to obtain long-term average predicted concentrations for four PM2.5 components: sulfur, silicon, and elemental and organic carbon (EC and OC). The models were specifically developed for the Multi-Ethnic Study of Atherosclerosis (MESA) cohort as a part of the National Particle Component and Toxicity (NPACT) study. The spatio-temporal model used 2-week average measurements from a monitoring campaign focusing on MESA participants, whereas the national spatial model relied on long-term means of daily measurements from the existing federally directed monitoring network. The spatio-temporal modeling framework consisted of long-term means, temporal trends, and spatio-temporal residuals. Spatial fields for long-term means and temporal trends were characterized in universal kriging with a land use regression component based on selected geographic covariates. The national spatial model was also constructed in a universal kriging approach with the mean model characterized by partial least squares scores instead of selected covariates. The cross-validation statistics of the two exposure models were 0.59 to 0.94 for sulfur, EC, and OC but 0.38 to 0.45 for silicon across the six study areas. Predicted long-term concentrations of PM2.5 components from the two models were fairly or highly correlated across cities within each of all four components except for OC, largely dominated by the between-city contrast. However, predictions were less correlated within each city than across cities. The national spatial model gave lower magnitude and less variable predictions than the spatio-temporal model. Different sources of monitoring data and modeling approaches between the two models contributed to these results. Predictions of long-term average concentrations for PM2.5 components for study subjects will allow us to investigate health effects of PM2.5 components and identify PM2.5 components responsible for the PM2.5 association.

Disciplines

Biostatistics

Download

Included in

Biostatistics Commons

Share

COinS