Power and sample size computations for the simultaneous assessment of the consistency of treatment effects

In this study power and sample size calculations for the analysis of treatment-by-subgroup interactions are proposed. Therefore, the interaction effect is either defined via product-type interaction contrasts or as ratio of treatment differences. The latter formulation allows besides the detection of interactions also the assessment of the consistency/heterogeneity of the subgroup-specific treatment effect by the definition of an inconsistency margin.

You can find the pdf of a detailed article here. Furthermore, the slides from a corresponding talk are given here.