Bioequivalence and Bioavailability Forum

❝ 3. A significant treatment effect was observed from the PROC GLM model output.

❝ Obviously the significant treatment effect was caused by B-R, but how do you defend it?

1. Nothing to defend. A significant treatment effect is irrelevant in BE. Furthermore, from your data it is clear that it is caused by T2 in the analysis of the complete data set. Since the 90% CI of T1/R included 100% this treatment was not significantly different from R (only T2: the CI did not contain 100%).
   
2. Drop Mr Schuirmann an and ask for his 2004 presentation. Maybe it gives you an idea which kind of evaluation FDA prefers (complete model or dropping T2 in your case)