Four methods for sample size re-estimation in a two-stage 2x2 cross-over trial for testing for Average Bioequivalence (ABE) were presented in Potvin et al. (2008). However, none of these methods formally controls the Type I error rate of falsely claiming ABE. In fact, the assessment of any inflation in the error rate of these methods has to be done in an ad hoc way using simulation.
We describe an alternative method of sample size re-estimation that is exact and guaranteed to control the Type I error rate. This method uses a new and robust version of the weighted combination of p-values test in conjunction with standard group sequential techniques. The sample size re-estimation step is based on significance levels and power requirements that are conditional on the first-stage results. We compare the operating characteristics of the new method with those of the Potvin et al. (2008).
Reference: Potvin, D., et al. (2008).Sequential design approaches for bioequivalence studies with crossover designs. Pharmaceutical Statistics, 7, 245-262.
|