Bioequivalence and Bioavailability Forum

Dear Matz!

❝ For a 2 x 2 BE study of total 24 subjects with two sequences AB and BA. […] one subject in the sequence AB is dropped out. So 11 subjects with seq AB and 12 subjects with seq BA.

❝ Is there any problem if I analyse this unbalanced data using a usual GLM procedure in SAS.?

No.

❝ or go for a MIXED procedure in SAS?.

Why? PROC MIXED makes only sense for designs with >2 periods (Williams’ designs, replicate designs).

❝ Any difference in the output?

Generally no; but for a 2×2 design I would stick to PROC GLM, because PROC MIXED is an iterative procedure (i.e., very small differences may exist).
A final remark: the more unbalanced any study gets, the more biased the PE (point estimate) based on the geometric mean ratio gets (the confidence interval remains unaffected). I would recommend using the MVUE (Minimum-variance unbiased estimator) instead - for SAS-code see:

S-C Chow and J-p Liu
Design and Analysis of Bioavailability and Bioequivalence Studies
Marcel Dekker, New York (2nd ed. 2000)

The PE is of importance in sample size planning for follow-up studies - the BE assessment is based on the CI only.