Bioequivalence and Bioavailability Forum

Hi sedhosen,

❝ Is there any guideline (or even an article) for determining sample size in comparability studies?

What do you mean by “comparability”? The most common ones:

• Testing for equivalence. This is a two-sided test, generally performed at a nominal level α 0.05.
  
• Testing for non-inferioriy (e.g., verum vs. placebo, where you assume that higher responses are better). This is a one-sided test, generally performed at a nominal level α 0.025.
  
• Testing for non-superiority (e.g., in safety studies comparing a new treatment to an established one and you assume that lower responses are better). A one-sided test as well.

❝ How should we calculate it?

Not at all. Since the difference between treatments and its variability are assumptions, you can only estimate the sample size.