Benchmark Dose Estimation Using SAS®

Toxicologically-based quantitative risk assessment is concerned with estimating human risks based upon experimental data linking an environmental agent to a known outcome (tumor incidence, acute toxicity, etc.). For dichotomous outcomes dose-response curves are modeled as complex functions of dose which often require specialized software to estimate. The SAS procedure NLMIXED readily allows for maximum likelihood estimation of binomial response data to any non-linear function. This power gives the SAS system the ability to fit dichotomous response curves that have traditionally been modeled using specialized software. For this paper excess risk is estimated within SAS using NLMIXED and the SAS MACRO programming language. Nine dose-response curves are fit examining the excess risk associated with renal tubular degeneration in ethylene glycol exposed rats. The results are compared with those given using the USEPA Benchmark Dose Software. The results show that these models can be reliably implemented within the SAS system producing very similar results as the USEPA software.