Bioequivalence and Bioavailability Forum

Dear Joy,

I'm afraid your sample size has to be calculated for the one drug with higher variability. If there's a large difference in variability between drugs, I would suggest to analyze only a neccessary subsample for the drug with lower variability (let's say the first 24 out of a total of 64). In analyzing just a subsample you avoid chances for a significant treatment effect (the stupid '100%-not-included-in-confidence-interval-question')... If the study is to be performed in more than one group - which is likely - the subsample should consist of subjects coming from all groups. Of course the planned procedure must be described in detail in the protocol, otherwise the impression of cherry picking may arise.