Bioequivalence and Bioavailability Forum

Hi Siva Krishna,

some hints:

• I’m not sure whether I understand your subject line. For reference-scaling you need a replicate design and software which can handle that. If the software was designed for 2×2×2 only, it should throw an error (since the design is not supported).
  
• The intention to apply reference-scaling should have been stated in the protocol (approved by the OGD).
  
• If your friend planned to used ANOVA (i.e., fixed effects in suitable software) if s_wR <0.294, that’s a deviation from the guidance (requiring a mixed-effects model). Your friend should have contacted the OGD before the study. I doubt that the OGD would have accepted that.
  
• By not doing so, your friend actually risked a Refuse-to-Receive (Section B) and was lucky that the OGD asked for a second evaluation according to the guidance.
  
• In many cases the mixed model is more conservative than ANOVA. Hence, a wider CI with the former is not unusual.

❝ How can we justify this failure. please help to my friend.

Justify cherry-picking? Imagine you are an assessor. You have two evaluations on the desk. One deviating from the guidance – passing – and another one acc. to the guidance – failing. What would you do? Let’s reverse this game. You have another study passing acc. to the guidance and plan to submit it. Just for fun you assess it by an ANOVA and it fails. Would you present that? I doubt it.

P.S.: I suggest that you friend registers here. Possibly we have more questions and it’s more efficient to get background information directly than playing Chinese Whispers.