Abstract
BACKGROUND: Valid, high-resolution estimates of population-level exposure to air pollutants are necessary for accurate estimation of the association between air pollution and the occurrence or exacerbation of adverse health outcomes such as Chronic Obstructive Pulmonary Disease (COPD). OBJECTIVES: We produced fine-scale individual-level estimates of ambient concentrations of multiple air pollutants (fine particulate matter [PM(2.5)], NO(X), NO(2), and O(3)) at residences of participants in the Subpopulations and Intermediate Outcomes in COPD Air Pollution (SPIROMICS Air) study, located in seven regions in the US. For PM(2.5), we additionally integrated modeled estimates of particulate infiltration based on home characteristics and measured total indoor concentrations to provide comprehensive estimates of exposure levels. METHODS: To estimate ambient concentrations, we used a hierarchical high-resolution spatiotemporal model that integrates hundreds of geographic covariates and pollutant measurements from regulatory and study-specific monitors, including ones located at participant residences. We modeled infiltration efficiency based on data on house characteristics, home heating and cooling practices, indoor smoke and combustion sources, meteorological factors, and paired indoor-outdoor pollutant measurements, among other indicators. RESULTS: Cross-validated prediction accuracy (R(2)) for models of ambient concentrations was above 0.80 for most regions and pollutants. Particulate matter infiltration efficiency varied by region, from 0.51 in Winston-Salem to 0.72 in Los Angeles, and ambient-source particles constituted a substantial fraction of total indoor PM(2.5). CONCLUSION: Leveraging well-validated fine-scale approaches for estimating outdoor, ambient-source indoor, and total indoor pollutant concentrations, we can provide comprehensive estimates of short and long-term exposure levels for cohorts undergoing follow-up in multiple different regions.