Bioequivalence and Bioavailability Forum

Hello Derya and Hi to everybody.

I would like to continue that discussion and adding my 2 cents here. My opinion has a good chance to be wrong, but then the corrections will be highly appreciated.

❝ There are 2 different dosages of the same test drug (T1, T2) and one dosage of reference (R).

You mean dosage forms or different formulations, right? Like in two candidates for the new product

❝ -Should I compare "T1 vs R" and "T2 vs R" in 2 anova tables?

From my understanding of the current thinking in the EU and because I personally do not understand why an ANOVA including also unrelated treatments should be beneficial at all I would not include that treatment not being compared. In other words, I would always prefer a pairwise comparison, so "Yes".

❝ -If I make 2 anova should I adjust the alpha?

I think that is a great question. My safe answer would be "Yes" and I know of some colleagues that would support this, because this is a classical situation of testing multiple times/formulations and being happy with 1 positive result.
At least when this is your pivotal study for submission. If it is "only" an exploratory pilot adjustment is of less relevance.

❝ -And if I adjust the alpha, does the CI change?

❝ Not 90% anymore...?

Sure, it is 1-2*ALPHA

However, my problem without an answer I understand would be: ALPHA-adjustment is requested in cases where I test multiple formulations in one study. However, nobody seems to expect that when I simply test multiple formulations in separate studies.
Is there any rationale to not consider these studies (of which I even do not need to tell anybody as they do not concern "the new product") multiple testing?

Best regards,

Steven.