Bioequivalence and Bioavailability Forum

Hi John,

❝ world record (?): 180 subjects, full replicate.

❝

❝ What would the sample size be if it was run as a 2-way study

IIRC what the guy planning the study told me, B-I expected a high GMR. Don’t forget that the study was designed for FDA’s RSABE. For a GMR of 1.2, CV 0.55, and 85% power that would transfer into 172 subjects. Add an expected droput-rate of 5% and you end up with 180 dosed.
No scaling in a 2×2×2 study. Therefore, 2,282 subjects… To fulfill the FDA’s new requirements (4-period full replicate, assuming CV_T=CV_R) according to Detlew’s new function sampleN.HVNTID() in the pre-release PowerTOST 1.3-01 1,146 subjects. Forget it.
If dabigatran would be classified by the EMA as an NTID, 90.00–111.11% would be required. Sample size estimation not possible for a PE of 1.2.
Anyhow, the PEs in the “world record” study were >1.25. Not even n=∞ would have helped – regardless which fancy method ever one would apply.

This one on mesalamine was a 2×2×4 in 238 subjects. AFAIK, the current world record is held by another innovator, Pfizer (two fully replicated 4-period studies in hundreds [sic] of subjects; personal communication Scott Patterson). No idea which Wonder-Pill economically justified such studies.