Spatial vulnerability of diabetes related to fine particulate matter in the United States

Recent research from laboratory and epidemiology studies supports the linkage between diabetes and fine particulate matter (≤ 2.5 µg in diameter; PM_2.5). However, advanced evidence relating location and diabetes vulnerability caused by PM_2.5 is still sparse, and the level of geographic disparities in diabetes attributed to PM_2.5 has not been evaluated. We conducted a spatiotemporal study to determine whether long-term exposure of PM_2.5 is spatially associated with diabetes prevalence. We extracted and analyzed diabetes prevalence data from the Behavioral Risk Factor Surveillance System, socioeconomic status data from the American Community Survey, and daily maximum 24-hour average of PM_2.5 concentration data from EPA's Air Quality System in 3,109 counties of the U.S. from 2005-2011. A Bayesian structured additive regression model was applied to estimate the interactive impact of PM_2.5 and county location on diabetes prevalence, adjusting for the county population’s socioeconomic status. We also conducted spatial comparison to further investigate vulnerable areas with different levels of PM_2.5 concentrations. The adjusted relative risk of PM_2.5 showed that 56.03% of counties had a significantly positive relative risk greater than 1, where the range was from 1.03 (95% CI = 1.00, 1.08) to 1.40 (95% CI = 1.35, 1.45). Vulnerable areas more likely appeared in counties in the eastern and southern U.S., where it also showed sensitivity on diabetes prevalence at a low level of PM_2.5 concentration. A clear cluster of significantly elevated relative risk was visible in Michigan counties when PM_2.5 reached a median-high level. This study identifies spatial vulnerability of diabetes prevalence due to PM_2.5, and provides the possibility of establishing targeted surveillance system to raise alertness of diabetes in areas with PM_2.5 pollution.