A confirmatory strategy for therapeutic equivalence trials.

In the planning phase of a therapeutic equivalence trial, a range has to be specified that defines the parameter region within which the treatments to be compared are considered equal for practical purposes. It is current practice in equivalence assessment to fix some value for limiting the equivalence range and to test the corresponding null hypothesis in confirmatory analysis. This procedure is unsatisfactory, if, for example, the drug to be investigated is not only equivalent but turns out to be even superior to standard. In this situation it would be desirable to test not only for equivalence but also for superiority. The paper presents an alternative approach that allows for a flexible handling of the problem. If therapeutic equivalence can be established the method additionally allows, under maintenance of the multiple level, for testing all null hypotheses that correspond to acceptable equivalence ranges and even of superiority. Because of the special structure of the hypotheses no alpha-adjustment is needed for these additional confirmatory analyses.