Dose-Response Models for Teratological Experiments

In a recent paper, hereinafter referred to as RVR, Rai and Van Ryzin (1985) propose a dose-response model for use in low-dose extrapolation from teratological experiments involving quantal responses. Suppose n litters are divided into m + 1 dose groups, the ith group consisting of ni litters exposed to dose di (O = do < di < ...< dn), the group i = 0 being a zero-dose control. Let Sij denote the size of the jth litter in the ith group (j = 1, . . ., ni), and let Rij denote the number of foetuses or pups in the litter exhibiting the toxic response of interest. The model proposed by RVR assumes that the conditional distribution of Rij, given Sij = sij, is Binomial(sij, p1j), where