On the assessment of bioequivalence in a two-period cross-over design.

The results obtained with Meineke and De Mey's algorithm for posterior probability distributions in the nonparametric evaluation of two-period cross-over bioequivalence studies are critically discussed. Suggestions for improvement of their program NEWPARM are given. It is shown that this program in its present form cannot handle the results from bioequivalence studies in which more than 15 subjects participated. For larger study sizes, conversion to a more powerful programming language capable of handling large three-dimensional arrays, is mandatory. An alternative algorithm, allowing large sample sizes to be analysed nonparametrically without program conversion, is offered. A general bioequivalence program written by the author, into which this alternative algorithm is implemented, has been in use by a number of pharmaceutical companies and drug regulatory agencies since February 1990; its most recent update is Version 3.7 (BIOEQV37.EXE) of October 1990.