Often spatiotemporal resolution/scale of environmental and health data do not align. Therefore, researchers compute exposure by interpolation or by aggregating data to coarse spatiotemporal scales. The latter is often preferred because of sparse geographic coverage of environmental monitoring, as interpolation method cannot reliably compute exposure using the small sample of sparse data points. This paper presents a methodology of diagnosing the levels of uncertainty in exposure at a given distance and time interval, and examines the effects of particulate matter (PM) ≤2.5 µm and ≤10 µm in diameter (PM2.5 and PM10, respectively) on birth weight (BW) and low birth weight (LBW), i.e., birth weight <2500 g in Chicago (IL, USA), accounting for exposure uncertainty. Two important findings emerge from this paper. First, uncertainty in PM exposure increases significantly with the increase in distance from the monitoring stations, e.g., 50.6% and 38.5% uncertainty in PM10 and PM2.5 exposure respectively for 0.058° (~6.4 km) distance from the monitoring stations. Second, BW was inversely associated with PM2.5 exposure, and PM2.5 exposure during the first trimester and entire gestation period showed a stronger association with BW than the exposure during the second and third trimesters. But PM10 did not show any significant association with BW and LBW. These findings suggest that distance and time intervals need to be chosen with care to compute exposure, and account for the uncertainty to reliably assess the adverse health risks of exposure.
Keywords: Chicago; air pollution epidemiology; coarse and fine particulates; exposure uncertainty; semivariance; spatiotemporal autocorrelation.