Bioequivalence and Bioavailability Forum

Originally we have 300 subject in control group and 600 subject in treatment group.


We found in a study that the nutritional supplement has become available outside of the study and determined that 50% of the mothers who are randomized to the control group are going to find a way to obtain this supplement. We have decided do an Intent to Treat Analysis and include all subjects in the Control Group in the analysis as untreated even if they in fact obtained study treatment. We restrict to Δ = 0.05, 0.10 and 0.15 and the only rate of event in untreated subjects is 0.50. Also assume that:

a) The 50% of controls who will acquire the supplement are a random subset of all controls and by acquiring the treatment outside the study, their babies will benefit as much as if the mother had been randomized to the treatment group.

b) Those assigned to the real intervention (Arm A) will not be influenced (i.e., the weight of their babies will not change) by the treatment being available outside of the study

c) While clearly there is some randomness to how many of the controls will obtain the study supplement, for the purposes of this assignment assume that Exactly 50% of Controls get the True supplement

How to define the final n and m in PS the power and sample size program for incorporating this new information and its consequences?
Bassically the main question is how many control and experimental patients we have to use at the end.