BACKGROUND: Established causes of diabetes do not fully explain the epidemic. High level arsenic exposure has been implicated in diabetes risk but the effect of low-level arsenic exposure in drinking water remains unclear.

OBJECTIVE: To determine if long-term exposure to low-level arsenic in drinking water in Denmark is associated with increased risk of diabetes using a large prospective cohort.

METHODS: During 1993-1997 we recruited 57,053 persons. We followed each cohort member for diabetes occurrence from enrollment until 31 December 2006. We traced and geocoded residential addresses of the cohort members and used a geographic information system to link addresses with water supply areas. We estimated individual exposure to arsenic using all addresses from 1 January 1971 until the censoring date. Cox proportional hazards models were used to model the association between arsenic exposure and diabetes incidence, separately for two definitions of diabetes: all cases and a more strict definition, where cases of diabetes based solely on blood glucose results were excluded.

RESULTS: Over a mean follow-up of 9.7 years of 52,931 eligible subjects, there were 4,304 (8.1%) diabetes cases in total, and 3,035 (5.8%) cases of diabetes based on a stricter definition. The adjusted incidence rate ratio's per 1 µg/L increment in arsenic levels in drinking water were (IRR = 1.03; 95% CI: 1.01, 1.06) and (IRR = 1.02; 95% CI: 0.99, 1.05) for all and strict diabetes cases, respectively.

CONCLUSIONS: Long-term exposure to low-level arsenic in drinking water may contribute to development of diabetes.