Mini-computer programs for bioequivalence testing of pharmaceutical drug formulations in two-way cross-over studies

For bioequivalence testing of pharmaceutical formulations of the same drug entity, it is not sufficient to carry out an analysis of variance on the characteristic to be evaluated (e.g., area under the plasma level vs time curve, half-life of elimination, time to plasma-peak level, plasma peak level) and to establish ‘classical’ 95% confidence intervals for the difference or the ratio of the characteristic concerned. In the past 10 years, several approaches have been proposed as an aid in decision-making: Westlake's 95% intervals, Rodda and Davis' probabilities, Fluehler's posterior probability histograms and the evaluation of the residual variation coefficient. A survey of these approaches is given, together with a discussion of their merits, their differences and their similarities. It is recommended that the final evaluation should be supported by probability density plots, which facilitate easy understanding of the differences and similarities between the various approaches. A bioequivalence study with two types of oral tablets containing bepridil, a new anti-anginal drug, is used as an example. Computer programs are presented, which enable the user to easily apply the various approaches in order to meet requirements of regulatory agencies.