Bioequivalence and Bioavailability Forum

Dear Jo!

❝ [...] intraCV for Cmax of 34%.

❝ Sample size [...] after statistical calculation was of 36 subjects.

❝ The sponsor [...] showed [...] different software, obtaining [...] 48 subjects.

❝ Who is right in this case?

For ±5% deviation of test from reference and 80% power using exact methods I'm getting 50 subjects (because the calculated sample size is slightly larger than 48 and must be rounded up to the next even number) according to the R code given in this post (with this modification; also implemented in package bear), in StudySize 2.0.1, and in FARTSSIE 1.6.
Approximations (Hauschke et al. (1992), Chow & Wang (2001); references) come up with 50 subjects also.

How did you get 36?
I'm getting 36 for T/R=100% and 75% power. 100% is too optimistic, and 75% power leads to 25% producer's risk...