Dose-Response Modeling with Summary Data from Developmental Toxicity Studies
ABSTRACT: Dose response and benchmark dose analysis of binary developmental data (e.g., implant loss, fetal abnormalities) is best done using individual fetus data within litters (here, “litter-specific” data) in order to accurately estimate fetal risk and its variability. However, such data are not often available to risk assessors as scientific articles usually present only summary statistics. When litter-specific data are not reported, responses can be adjusted for “litter effects” (i.e., intralitter correlation, design effect) using a default value for the design effect or by estimating the design effects from historical data. In the present study, we show that summary data on fetal malformations can be adjusted satisfactorily using estimates of the design effect. When adjusted data are then analyzed with models designed for binomial responses, the resulting benchmark doses are similar to those obtained from analyzing litter level data with nested dichotomous models. By extension, when litter-specific data are available, adjusting the response data by estimated design effects and using a model for binomial data is also a reasonable alternative to using a nested dose response model.