Analysis of Responses from Developmental Toxicity Experiments in Non-Homogeneous Populations. M. Razzaghi, Bloomsburg University, Department of Mathematics, 400 E. Second Street, Bloomsburg, PA 17815
In developmental toxicity experiments with laboratory animals, pregnant female dams are usually exposed to one of several doses of the test agent during a critical time of the gestation period. The animals are generally sacrificed before term and the uterine contents of each animal are examined for a variety of teratological effects. A conceptus may either be dead or alive, and a live fetus may exhibit one or more malformations. In such experiments, however, it is often observed that some subjects are less susceptible to the treatment. In order to account for this non-homogeneity in the population a mixture dose-response model is proposed. It is shown that the advantage of using a mixture model, besides accounting for the population non-homogeneity, is that the model is more flexible and thus provides a better fit for the data. Here, the quasi-likelihood approach is used to analyze the quantal-response teratology data incorporating the induced over dispersion relative to the multinomial variation due to correlation among litters. Generalized estimating equations are derived and solved iteratively in order to estimate the model parameters and the proposed procedure is illustrated through the analysis of data from a recent large scale study of the developmental toxicity of the herbicide 2, 4, 5 - trichlorophenoxyacetic acid.