Spatial and temporal interpolation of air pollutants in New York City: exposure assignment for use in a birth outcomes study

Background: Many studies of air pollution and birth outcomes are limited by spatially sparse monitoring to assess the importance of within-city exposure gradients. Aims: To develop spatio-temporal models of fine particles (PM2.5) and nitrogen dioxide (NO2) and characterize exposures during pregnancy in New York City. Methods: Data on emissions sources were used in a regression model to predict PM2.5 and NO2 concentrations collected at 150 locations in Year 1 of the NYC Community Air Survey (NYCCAS). Two-week average exposures were estimated by temporally adjusting city-wide pollutant time-series' derived from regulatory monitors. This method was validated by comparing model-based estimates to two-week measured concentrations during Year 2 of NYCCAS. The models were used to assign exposures to PM2.5 and NO2 for each trimester and the entire gestation period for 274,996 births 2008-2010 at maternal addresses and for 300m and 800m buffers around the maternal address. We then characterized the spatial/tempora...