Blinded Tests in 2 x 2 Cross-Over Trial

We propose tests for equality of treatment effects in the context of a double-blind 2 x 2 cross-over trial which do not require treatment codes to be unblinded and thus can be used as a part of the interim analyses. We first discuss the set-up of a 2 x 2 cross-over trial and state the important assumptions necessary for our procedure. Section 2 discusses implications of blinded interim data analysis under the givensetting and review the test proposed in Millder et al. [27] while pointing out its limitations. Section 3 introduces two preliminary (blinded) tests for equality of means when the data are collected from a 2 x 2 cross-over trial and study their power properties. Section 4 proposes a procedure to improve upon the performance of the tests proposed in Section 3 which require the assumption of equality of variances for the two treatments. The resulting methodology is a two-stage test and is introduced in Section 5. Section 6 discusses the performance of the two-stage test against the benchmark test which would have been effective had the data been unblinded. At the end we add a few concluding remarks.