Odds ratio based two-stage group sequential analysis for two-treatment two-period crossover trials

Abstract

We consider a two-treatment and two-period crossover design where the treatment responses are binary. The outcome from each patient under such trial can be represented by a two-component vector which follows a bivariate binomial distribution. We develop a two-stage group sequential test procedure for equivalence based on the estimates of odds ratio. Various error spending functions are studied in this context. Simulation studies are carried out. An illustration with some real data example is also given.

Keywords:

• Correlated binomial model
• odds ratio
• type I error spending function