Crossover trials are only useful when there is a positive correlation between the response to different treatment modalities

Abstract

1. Trials that do not allow us to reject the null hypothesis of no treatment effect may have had an inappropriate design. Trials are virtually never assessed for correlation between responses to different treatment modalities. 2. The level of correlation between responses to different treatment modalities is a major determinant of the power of crossover trials. 3. It is relevant to assess correlation levels between responses to the different treatment modalities a priori. With a negative correlation a crossover design is likely to lack power. 4. In 1991 the British Journal of Clinical Pharmacology published eight crossover trials which would have been more efficient had they been performed with a parallel groups design.