The Impact of Litter Effects on Dose-Response Modeling in Teratology

Abstract

The fitting of dose-response models to teratology data involving littermates in order to generate estimates of teratogenic risk is receiving increasing attention as a potential alternative to the "safety-factor" approach to risk estimation. In this paper, we utilize the beta-binomial distribution to introduce varying degrees of intralitter correlation, and, for purposes of illustration, consider a logistic dose-response model that describes the logit of risk as a straight-line function of ln(dose). The biases and (exact and asymptotic) variances of the maximum likelihood estimators of the intercept and slope are studied by simulation as a function of the intralitter correlation structure.

A Linear <em>B</em>-Spline Threshold Dose-Response Model with Dose-Specific Response Variation Applied to Developmental Toxicity Studies

“…They also conducted simulations and found that the dosespecific model with multiple variation parameters led to unbiased estimation of all model parameters, whether the true variance structure was that of single or multiple parameters, whereas the single-parameter model was less robust. These results are similar to those of Kupper, et al (1986), who assumed the beta-binomial distribution for the response number for each litter and a logistic doseresponse model and found that the model with the multiple-intralitter correlation structure produced less biased results than the model that assumed a single-intralitter correlation.…”

“…This more accurately specified model improves the estimation of important parameters such as the threshold or, in the case of the spline model, the change point. As illustrated in Kupper, et al (1986) and in Hunt and Rai (2007), the model assuming multiple parameters to model response variation leads to negligible bias, whereas under conditions of major differences in dose-specific variation the model with single parameter may lead to extremely biased estimates. Thus, the model that has multiple variation parameters is the general model that should be used.…”

7. Categorical Data Analysis

“…When varying q (0.05-0.2), the Dirichletmultinomial MLE caused biased estimates. This situation was also noted for the binomial response cases (Kupper et al, 1986;Yamamoto and Yanagimoto, 1988;Car and Portier, 1993).…”

Alternatives To Using Human Experience In Assessing Health Risks

“…Unfortunately, the former approach ignores the litter, while the latter ignores the contribu tions of the individual fetuses within the litters. More recently, investigators have begun to look at methods for taking litter effects into account in the modeling process (42,63).…”