Bioequivalence and Bioavailability Forum

Dear Stefano!

❝ the AUCs are compared based on an ANOVA model...

Right. PE, confidence interval, end of story.

❝ However, in several papers I found absolute BA calculated for each subject

❝ (usually reported as F) and summarized using descriptive statistics.

IMHO this would only make sense for a balanced study to have a 'quick look' on the geometric mean. If the study was imbalanced this value may be misleading.

❝ In my opinion this second approach may be useful to provide some support to

❝ the result obtained using the ANOVA, but it can not be considered as the

❝ analysis of primary interest.

I do not see how this would support the result of a statistical model. I would say it's redundant at the best.