Bioequivalence and Bioavailability Forum

Dear all!

I have a rather philosophical questions regarding adaptive BE studies. IMHO there are two different concepts available.

Concept 1: Plan your study with a type I error of 5% and a target power of >80%. An interim analysis is performed after 50% of subjects were treated. This can lead to the following decisions:

• Early stopping (not really expected)
  
• Increase sample size for 2nd stage leading to a larger total sample size compared to the a-priori planned study (try to rescue the study when the a-priori made assumptions were wrong)
  
• Decrease sample size for 2nd stage leading to a smaller n compared to the a-priori planned study (that’s what we are implicitly hoping for)

Concept 2: Go for early stopping by planning a study with a type I error of 2.5% and a target power of >80% leading to n subjects in the first stage:

• Early stopping (best case as expected)
  
• Calculate sample size for 2nd stage (try to rescue the study when the a-priori made assumptions were wrong)

I think that these are the most extreme cases and I would be grateful for arguments why one should go for concept 1 or concept 2 in the field of BE. I would also be grateful for discussions and suggestions around the early stopping bound (e.g. 2.5%). For concept 2, I think a early stopping bound of around 4% makes also sense.

best regards

martin

---

Edit: Category changed. [Helmut]