Bioequivalence and Bioavailability Forum

Hi Joystar,

adding some points to what ElMaestro already said.
Sample size is estimated – not calculated – because it is based on assumptions (of the T/R-ratio, the expected drop-out rate, the risk of failure the producer is willing to accept). See this presentation;¹ especially the section about sensitivity analysis. Many guidelines suggest a power of 80–90%. If you plan a study with a power of <80% it might be possible that the ethics committee rejects the study (too high risk of failure). I know of only one guideline (from New Zealand) suggesting a lower power in case of high variability:²

7.5.5 Number of subjects
[…] the number should be sufficient […] to provide the necessary discriminatory power (normally ≥80%) to detect the maximum allowable difference (usually ±20%) in C_max, AUC etc.
If the calculated number of subjects appears to be higher than is ethically justifiable, it may be necessary to accept a statistical power which is less than desirable. Normally it is not practical to use more than about 40 subjects in a bioavailability study.

Personally I have serious problems understanding the rationale behind. More than 40 are considered unethical. What if the CV is 70%? Power in a full replicate would be just 38.8%³ – so one has a 61.1% chance to fail and repeat the study in another 40 subjects hoping for a better outcome this time? That’s ethical? I call it ignorant.

---

1. Sorry for its title. In the meantime I have learned better.
   Should have been “Sample Size Estimation”.
   
2. Maximum CV (40 subjects, T/R 0.95, 80% power):
   2x2x2 cross-over:  30.6%
partial replicate: 36.0%
full replicate:    44.9%
   
3. That’s like rolling two dice and expecting ten dots in total (p₁ = 11/36) or at least one die shows two dots (p₂ = 3/6); p_1|2 = p₁+p₂ = 0.389… I do like gambling but not if humans are pawns in a game.