Effects of time and point-of-use devices on arsenic levels in Southeastern Michigan drinking water, USA

Health effects associated with chronic, low-level exposures to arsenic in drinking water (< 100 μg/L) remain unclear, in part due to uncertainties in assessing exposure. Drinking water concentrations have been used to assess past exposure to arsenic in epidemiological studies, under the assumption that a single measurement can be used to estimate historical exposure. This study aims to better understand (1) temporal variability in arsenic concentrations in drinking water and (2) the impact of point-of-use (POU) treatment devices on arsenic exposure measurements, and on reliability of the exposure measurement for population-level studies. Multiple drinking water samples were collected at two points in time (an average of fourteen months apart) for 261 individuals enrolled in a case-control study of arsenic exposure and bladder cancer in Michigan. Sources of drinking water included private wells (n = 221), public water supplies (n = 33), and bottled water (n = 7); mean arsenic concentration was highest in private wells (7.28 μg/L) and lowest in bottled water samples (0.28 μg/L). Arsenic concentrations in primary drinking water samples were highly correlated (r = 0.88, p < 0.0001, n = 196), with 3% of the water sources exceeding the United States Environmental Protection Agency's Maximum Contaminant Level (MCL) in one sample but not in the other sample. Measurement reproducibility did not vary by type of POU device (e.g., softener, filter, reverse osmosis system). Arsenic concentrations did differ, however, between samples treated with POU devices and untreated samples taken on the same day. Substantial differences in arsenic concentrations were consistently observed for reverse osmosis systems; other POU devices had variable effects on arsenic concentrations. These results indicate that while a single residential arsenic measurement may be used to represent exposure in this region, researchers must obtain information on changes in water source and POU treatment devices to better characterize population exposures over time.