An equivalence trial is appropriate when it is desired to demonstrate equivalence between two treatments, regimens or interventions (methods) or non-inferiority of a new one compared to a standard one. The conduct of an equivalence trial requires different techniques during design and analysis compared to a superiority trial. The existing formulae for sample size calculation to demonstrate equivalence between two methods using the confidence interval approach are reviewed. The establishment of the margin of equivalence and the choice of the type of test are discussed. Plots of sample sizes required to demonstrate equivalence in the case of binary outcomes are presented for values of proportions and margins of equivalence common in the reproductive health field. Examples are given of method comparisons in the reproductive health field in which the relevant question is to demonstrate non-inferiority. The approach to equivalence is described in the trials included in three published systematic reviews in which these comparisons were conducted, addressing the statement of hypotheses, sample size calculation and the interpretation of results. The use of the conventional superiority approach to design equivalence trials has led to underpowered trials to show equivalence within clinical relevant margins. The analysis and interpretation of results from such trials has resulted in conclusions of equivalence based on lack of significance. We draw attention to the lack of awareness of the appropriate techniques for equivalence trials among researchers in the field of reproductive health. Finally, the issue of interim analyses and stopping rules in equivalence trials is addressed.
Copyright 2001 John Wiley & Sons, Ltd.