Drinking Water Quality and Emergency Department Visits for Gastrointestinal Illness, 1993–2004

Abstract

P-237 Introduction: Evidence suggests that drinking water quality contributes to endemic gastrointestinal (GI) illness in the US, but the determinants of this relationship and the proportion of illness attributable to drinking water are poorly understood. Contaminated drinking water can reach the end user through inadequate treatment and through intrusion of pathogens into the distribution system. These sources of contamination are considered in this set of analyses. Methods: We collected emergency department data from 41 of 42 hospitals serving a major US city from 1993 through 2004. A total of 7,952,826 visits are being analyzed, 446,559 (5.63%) for GI illness. Six major drinking water utilities serving three million residents are participating in this study, providing water quality and distribution system network data. To better understand the role of inadequate water treatment in contributing to GI illness we are considering temporally variable water quality measurements, including turbidity and disinfection efficiency, taken at each plant and the differences in raw water sources and treatment techniques among plants. We are evaluating the role of the distribution system by considering temporally variable water quality measurements, including chlorine levels and detection of indicator bacteria, in samples taken from the distribution system. We are also evaluating the potential for distribution system contamination, as quantified by the estimated distance from treatment plant to end user and the time the water travels, or water age. Water age is being estimated using hydraulic models. We are examining the association between these water quality factors in each zip code and emergency visits for GI illness among people residing in that zip code. Poisson generalized linear models are being fit, controlling for potential confounders. Results: An average of 102 emergency visits for GI illness occurred per day in the study area, out of an average of 1810 total emergency visits per day. The maximum turbidity of the source waters serving the treatment plants, including both lakes and rivers, ranged from 0.6 to 472 nephelometric turbidity units (NTU). The average filtered water turbidity ranged from 0.03 to 0.166 NTU. The average distance from plant to end user was estimated to be 9.50 miles (range: 0.96–23.6 miles). The described epidemiological analyses are ongoing and will be presented. Discussion: Results from this study, the most comprehensive of its kind to date, will contribute important new evidence regarding the role of drinking water quality in endemic GI illness.