Association between arsenic and diabetes mellitus in National Health and Nutrition Examination Survey 2007-8

Abstract

OBJECTIVE: Whether low-level exposure to arsenic in the environment is associated with diabetes is controversial. We therefore studied this association in the National Health and Nutrition Examination Survey (NHANES) 2007-8. METHODS: In NHANES 2007-8, urine arsenic was measured in around one third of participants. After excluding children and those with missing fasting blood glucose or urine arsenic data, 591 participants were included in the analysis. There were 109 (18.4%) participants with diabetes, defined as glycosylated haemoglobin of ≥6.5%, fasting serum glucose of ≥126 mg/dL, self-reported previous physician diagnosis of diabetes, or self-reported use of anti-diabetic medication. RESULTS: Comparing the highest and lowest quintiles of total urine arsenic (>13.94 vs 3.38 μg/L), the odds ratio for diabetes was 1.71 (0.78-3.72) in the unadjusted model, but after adjustment for age, sex, ethnicity, education, body mass index, cotinine, blood mercury, urine creatinine and anti-hypertensive medication use, the odds ratio was 3.22 (1.21-8.56) [P=0.022]. The result remained significant after subtracting arsenobetaine concentration. There was a weak relationship between fasting blood glucose and urinary total arsenic (r=0.07, P=0.035). CONCLUSIONS: High arsenic exposure (total urine arsenic >13.94 μg/L) in one fifth of Americans is associated with diabetes, and fasting blood glucose is weakly related to urinary arsenic. Our results do not prove a causal link between arsenic and diabetes. However, as arsenic is present in drinking water and foods such as rice, it would be prudent to monitor the exposure of the general population to arsenic and keep it low.