Bioequivalence and Bioavailability Forum

Dear Boonchai,

I see your point. I also wouldn't be afraid of a significant group effect. I imagine your local regulators are perhaps seeing this in the same fashion as multicenter studies (??) where one or two centers stand out. That could indicate center lapses. But with multiple groups in XO BE studies, why bother? After all, in XO studies it is all about the intra-subject comparison. I can't see how a group effect in its own right can affect a conclusion of BE. Groups, as HS said, is a between-subject factor just like a harmless sequence effect. Subjects are nested in sequences nested in groups, sort of (but screw the nesting thing; number all subjects uniquely).
In your example, the three separate analyses would probably not show inequivalence but just inconclusivity re. equivalence. This would be important to you.

I would not consider #4 too seriously. Aim at balancing the subject dimension across groups and that should be it. Everyone knows that balance is a nice thing, but heck we cannot even guarantee balanced sequences ever.

Perhaps drugs related to treatment of disease with seasonal variations could be an exception here (pulmonary steroids for example) since groups are separated in time. There is to the best of my limited knowledge no literature covering this area.