A practical approach for the assessment of bioequivalence under selected higher-order cross-over designs

The two-period cross-over design with two sequences of drug administration is a standard experimental design when bioequivalence of one test formulation is to be assessed in comparison with a reference formulation. Previously, an approach based on Fieller's confidence interval has been presented for the assessment of average bioequivalence under this particular design. However, the two-sequence two-period cross-over design is not very useful in the presence of unequal carry-over effects. Besides, this basic design does not provide independent estimators for the intra-subject variabilities. To overcome these limitations, it might be of interest to consider a higher-order cross-over design in which the number of periods and/or the number of sequences is greater than the number of formulations to be compared. Because of this, the present communication will concentrate on the generalization of Fieller's confidence interval concept for a particular group of higher-order cross-over designs. In addition to this, since the evaluation of simple average bioequivalence does not guarantee that the two products can be used interchangeably, the assessment of population and individual bioequivalence is addressed through the application of a comprehensible three-step decision rule. An example study with a two-sequence four-period design is also analysed to illustrate the use of the proposed methods.