Dose-Response Modeling for Nitrate and Chloral Hydrate: Use of Data from Sensitive Subpopulations. L. T. Haber, TERA, Cincinnati, OH; B. C. Allen, ICF Kaiser, RTP, NC; and A. Bathija, US EPA, OW, Washington, DC.
Human exposure to nitrate in drinking can result from agricultural runoff. Chloral hydrate is a drinking water disinfectant byproduct and is used as a sedative for children for medical imaging. We used categorical regression, a mathematical tool that can be used to combine disparate data sets into an overall dose-response curve, to analyze the human data for nitrate and chloral hydrate. In this work, a modeling approach was used that allowed incidence and group-level data to be combined. Nitrate toxicity results from its conversion to nitrite by bacteria in the gut. Infants have an elevated sensitivity to nitrate, because the lower acidity of their gastric environment allows the growth of nitrate-reducing bacteria, and because nitrite more easily oxidizes infant hemoglobin than adult hemoglobin to methemoglobin. We considered the appropriate choice of response level (EDx value) when data from a sensitive population is used for dose-response modeling. The implications of the identification of subsets of sensitive subpopulations (e.g., infants exposed to bacteria-contaminated water) were also considered. The reliability of the nitrate modeling was limited by the limited amount of data available. For chloral hydrate, categorical regression modeling had the advantage over the NOAEL/LOAEL approach of being much less dependent on assumptions regarding the background rate of response.
Supported by EPA contract 68-C7-0002. This abstract does not represent EPA policy.
Go to . . .