On assessment of bioequivalence under a higher-order crossover design

In bioavailability studies of two formulations of a drug, the standard two-sequence, two-period crossover design is usually considered to assess bioequivalence. The standard two-sequence, two-period crossover design, however, may not be useful when differential carryover effects are present. In addition, it does not provide independent estimates of intrasubject variabilities for the two formulations. To overcome these problems, alternatively, a higher-order crossover design may be considered. In this paper, we derive statistical methods based on Schuirmann's two one-sided tests procedure for assessing bioequivalence for some commonly used higher-order crossover designs. Four designs, including Balaam's design, the two-sequence dual design, and two four-period designs (with two and four sequences), are considered. The relative merits of these designs as compared to the standard two-sequence, two-period design are discussed. Two examples concerning bioequivalence are used to illustrate the use of these methods.