Bioequivalence and Bioavailability Forum

Dear Bala and Neethu,

can you please give any reason - besides money! - why you would not run two separate studies (one fasting, one fed)?
Just my two cents, you will run into touble for various reasons!
With an assumed drop-out rate of 5% / wash-out 3-period studies will end with 21 subjects, but your 6-period study with just 18. Are you really prepared in terms of power to loose 25% of subjects?
I'm afraid, designing a study needs a little bit more preparation than writing down some random blocks.
Think about the model first! In a BE study the main effect of interest is 'treatment' (>2 different formulations, but either in fasting or in fed state). In a food effect study it's 'food' (using the same treatment).
One of the main assumptions in the usual (nonreplicate) cross-over model is an Independent Identically Distribution (IDD) of effects. This assumption simply may not hold. If e.g., the variability of the reference is higher than the one of the test, we will obtain a high common variance and the test will be penalized for the reference performing badly. That's the reason why some people advocate RSABE. In a food study we see the same story. For most MR formulations we yet would expect different variabilities in fasting and fed state. Even for IR formulations food will change liver blood flow > hepatic clearance > not only the absorption, but also the elimination may be altered.
Since the study is of a nonreplicate design with 2 effects (2 levels: fasting/fed, 3 levels: T₁/T₂/R) the assumption of a common variance IMHO is downright absurd. Metaphorically speaking you are running out of degrees of freedom... and your effects are massively confounded (no unbiased estimates)!
Though it's a nice phrase 'to kill two birds with one stone' - I wouldn't call that a scientific concept.
So the only chance you have is to modify Bala's last suggestion (@Bala: why didn't you use Latin squares? You randomized no T₂ in P₁ and no T₁ in P₃.) ...
       fasting     fed
      P1 P2 P3   P4 P5 P6
  S1  T1 T2 R    T1 T2 R
  S2  T2 R  T1   T2 R  T1
  S3  R  T1 T2   R  T1 T2
... and evaluate P₁-P₃ and P₄-P₆ separately (I would use a 6-sequence Williams' design instead - but that's another story).
Can you please give any reason - besides money! - why you would not run two separate studies (one fasting, one fed)?