气候变化领域集成服务门户

Under the Clean Water Act, the US Environmental Protection Agency (EPA) collects information from states on intended use and impairment of each water body. We explore the feasibility of using these data, collected for regulatory purposes, for public health analyses. Combining EPA impairment data and stream hydrology information we estimated the percent of stream length impaired for any use, recreational use, or drinking water use per county in the US as exposure variables. For health outcomes we abstracted county-level hospitalization rates of gastrointestinal infections, GI (ICD-9CM 001-009 excluding 008.45) and gastrointestinal symptoms, GS (ICD-9CM 558.9, 787) among US adults aged 65 years and older from the Center for Medicare and Medicaid Services (1991-2004). Linear mixed-effects models were used to assess county-level associations between percent impaired waters and hospitalization rates adjusted for population density, a proxy for person-to-person transmission. Contrary to expectation, both GI and GS were negatively associated with any water impairment in adjusted models (GI: -0.052, 95 % CI: -0.077, -0.028; GS: -0.438, 95 % CI: -0.702, -0.174). GI was also negatively associated with recreational water impairment (-0.079, 95 % CI: -0.123, -0.036 after adjustment). Neither outcome was associated with drinking water impairment. Limited state data were reported to the EPA for specific recreational (27 states) and drinking (13 states) water impairment, thus limiting the power of the study. Though limited, this analysis demonstrates the feasibility of utilizing regulatory data for public health analyses.