Bioequivalence and Bioavailability Forum

Hi Tina,

❝ Is it possible to have a two-stage design in a 4-way crossover study (involving fasting and fed of both R and T) if variability observed between the two conditions differ pointing to 2 different sample sizes for fasting and fed.

1. In a completely randomized study (beman’s Approach 1 above) you’ll get one (pooled) estimated variance. Nothing is published yet. You have to perform simulations in oder to demonstrate that the patient’s risk is maintained. Good luck.
   
2. In beman’s Approach 2 you’ll get separate estimates of variances in fasting and fed conditions. Therefore, second stages can be of different sizes.

❝ Suppose fasting reuires 60 subjects and fed requires 100 subjects, do we have to dose 100 subjects intoto or to separate the sample sizes?

Example for #2: First stage 48 subjects, CV_fasting 34%, CV_fed 45%, Method B (T/R 0.95, 80% power). Perform the second stage in 60–48=12 subjects fasting and 100–48=52 subjects fed.