Statistical re-analysis of the Biel maze data of the Stump et al (2010) study: "Developmental neurotoxicity study of dietary bisphenol A in Sprague-Dawley rats"

Bisphenol A, Biel water maze, time to event, right censoring, equivalence, neurotoxicity
First published in the EFSA Journal
30. September 2010
30. September 2010
Scientific Report


The Biel maze data presented by Stump et al. (2010) were reanalysed by EFSA using statistical methods that took the censoring into consideration along with key experimental and design features. The results showed that it cannot be concluded that there is equivalence between the control group and BPA for neurotoxicity variables (i.e. time to escape and number of errors). Hence, statistically, the possibility of an effect of BPA on the neurotoxicity variables can neither be confirmed nor ruled out. After accounting for key design elements in the modelling there is still large variation in the hazard ratios for the dose effect. This indicates that the Biel maze test as performed by Stump et al. (2010) does not have the potential to demonstrate equivalence of BPA compared to a control. Therefore this study should be considered as being inconclusive.

European Food Safety Authority
amu [at]
EFSA Journal 2010;8(9):1836
Question Number
On request from
European Commission