Bioequivalence and Bioavailability Forum

Dear HS,

❝ Yeah – was in your nice country more than once… Took me ages to pronounce even simple words like ชา – almost – correctly.

Ha Ha Ha, it is usual for foreigners and thanks for the admiration my country.

Refer to that announcement, let me add more 1 thing.

4. All groups must have equal number of subjects.

BTW, refer to #3, why I said stupid? I have an example.

If you are going to run BE study with high CV drug on 2x2 standard design and the number of subject, 80 to reach the desired power and your max capacity of the facility is 34. You decide to separate them into 3 groups, 27, 27 and 26 (because of #1). During the study, you have dropout 1, 2 and 4 subjects in each group, respectively. So you analyze 26, 25 and 22 in each and 73 subjects in total (Is it in conflict with #4?). However, you pool them to evaluate BE and the result shows equivalence with 84% power but the ANOVA found the significant group effect (It is possible to occur because the group effect is nested with subject effect, you never know which subject is good, poor, fast or slow metabolizer, assuming you bad luck in this case, the first group was filled with the slow metabolizers and the second group was filled with the fast metabolizers). So you can not pool them according to #3. The results from the separated analyses show inequivalence for all 3 evaluations with 60, 62 and 58% power.

My question: Is it reasonable?

Regards,
Boonchai_l