Sven Schnaidt

Sven Schnaidt

University of Heidelberg, Germany

Title: Efficiency of two-stage designs for crossover bioequivalence trials


Sven Schnaidt has completed his Diploma in Biology and Master in Medical Biometry/Biostatistics at the Universities of Tuebingen and Bremen. He wrote his Master’s thesis at Boehringer Ingelheim. Since 2013, he is a Scientist in the Department of Medical Biometry and Informatics at the University of Heidelberg. His research focus lies on methodological research in the field of bioequivalence trials.


Bioequivalence studies aim at showing in vivo that two pharmaceutical products are comparable with respect to their bioavailability. Usually, the assumptions about the true effect and variability are uncertain, which makes it particularly difficult to calculate the required sample size assuring a predefined power. Recent regulatory guidelines on bioequivalence studies acknowledge these difficulties and accept application of two-stage designs in this context. Specifically, two approaches are described in these guidelines: group sequential designs with pre-specified fixed sample sizes in each of the two stages, and adaptive designs where the sample size of stage 2 is calculated based on the information observed in the interim analysis. The Canadian Guideline even gives recommendations for a specific design, namely to use a method proposed by Potvin et al. when applying an adaptive design. However, a number of alternative two-stage designs have been proposed that allow an adaption in sample size based on interim results. Up to now, these approaches have only been discussed for superiority and non-inferiority trials. We systematically compare the method described by Potvin et al. to alternative adaptive and group-sequential methods in the situation of two-way crossover bioequivalence trials. The performance characteristics in terms of type I error rate, power, and expected sample size are investigated. Results are discussed to give recommendations for future planning of two-stage bioequivalence trials.