Abstract:
|
With the rapid growth of the economy, environmental pollution has become a serious issue. Previous studies reported the adverse impacts of air pollution exposure on preeclampsia. But few of them accounted for spatial and temporal autocorrelations. In this study, we compared general Bayesian hierarchical models with different spatially and temporally autocorrelated random effects at the county level using Florida birth record data from 2006 to 2015. The model with a single set of spatially and temporally autocorrelated random effects following a multivariate autoregressive process of order 1 performed best. After accounting for the spatial and temporal autocorrelation, borderline significant association between the first trimester PM2.5 exposure level and preeclampsia risk was detected. Conception during May to July was related to higher preeclampsia risk. Smoothed standardized incidence ratios (SIR) were mapped. Both the smoothed temporal trend and the spatial inequality increased over the first 4 years, then getting stable, with a peak in 2012. Clusters and hot spots with high SIRs were detected in north Florida, especially around the Wakulla County.
|