Statistical aspects of comparative bioavailability trials.

The role of comparative bioavailability trials in testing for the bioequivalence of different formulations of a drug is discussed and the statistical aspects of the design and analysis of such trials are reviewed. It is suggested that the design of such trials presents no special problem but that the customary method of analysis, which tests the null hypothesis of no difference between the formulations, is irrelevant to the central purpose of such trials, which is to determine whether the formulations have essentially equivalent therapeutic effects. It is proposed that only those characteristics of the data that possess a meaningful relation to the therapeutic use of the drug should be analysed and also that estimation procedures rather than hypothesis testing techniques should be employed. Several aspects of the statistics of bioavailability trials which require further investigation are listed.