Bioequivalence and Bioavailability Forum

Dear Matz!

❝ One doubt regarding the analysis of a BE study. In a 2x4 replicated BE study with 24 subject. Finally only 22 subjects are completed all period and two of them are completed only first two period.

So in this two subjects you have no information from a third treatment which may be used in a mixed-model approach (SAS PROC MIXED). Therefore the results of PROC MIXED (n=24) will be identical to PROC GLM (n=22).

❝ Is there any problem if I analyse the study using the data of that 22 subjects, who completed all periods?.....

No (see above).

❝ Is it required to specify in the protocol that the statistical analysis will be done only using the data from subjects that complete all periods?......

You should specify the method in the protocol. Quoting Senn (2002):

“In the context of drug development […] it will be important to specify beforehand which approach (fixed or random) will be used.”

For the entire quote and the reference see this post.