In Reply: To address the concern raised by Dr Tseng, we further adjusted our analyses for family history of diabetes, use of hormone therapy, and use of dietary supplements. The results did not change appreciably. The ratios of the geometric means comparing participants with and without diabetes were 1.27 (95% confidence interval [CI], 1.04-1.55) for total arsenic, 1.10 (95% CI, 0.93-1.29) for dimethylarsinate, and 0.92 (95% CI, 0.68-1.24) for arsenobetaine.
Tseng also questioned the adequacy of urine arsenic levels as a surrogate marker for dosage of arsenic exposure. Urine arsenic is an established biomarker of arsenic exposure in drinking water1 that integrates multiple sources to serve as a marker of overall arsenic exposure.2 In addition, urine speciation can be used to evaluate arsenic metabolism and to differentiate inorganic from organic arsenic exposure.3 This is important because organic arsenicals, mostly found in seafood, seem to have little or no toxicity compared with inorganic arsenic. In our study, the high detection limits for arsenite, arsenate, and methylarsonate precluded the evaluation of the role of inorganic arsenic metabolism.
Navas-Acien A, Guallar E. Arsenic Exposure and Diabetes Mellitus in the United States—Reply. JAMA. 2008;300(23):2728-2729. doi:10.1001/jama.2008.813