Bioequivalence and Bioavailability Forum

Dear Dan!

I agree essentially with all your points, but I would strongly suggest to analyze all samples of dropouts (and even samples of subjects withdrawing consent) as well. I wouldn’t restrict analysis to subjects experiencing AEs. See this thread (especially the end of Ohlbe’s post). If any regu­la­tory authority asks you for PK data of these samples – and you don’t have them “handy”, you will be in a bad position (long-term stability exhausted or the samples are even already discarded).
Another often forgotten point: If a subject drops out at t > t_max in the last period, this subject may serve well in the comparison of C_max – only AUC is not evaluable (I’ve seen people excluding all data of such subjects from the evaluation – decreasing the power of the study).