Abstract:
|
The FDA has implemented new bioequivalence regulations that introduce two new criteria: population bioequivalence (PBE) and individual bioequivalence (IBE). The FDA recommends studies with a four-period crossover design to assess these criteria, and also recommends a testing procedure developed by Hyslop, the Howe small-sample confidence interval (HSCI). An alternative testing procedure has been developed by McNally et al. based on the generalized p-value (GPV). This investigation compares the HSCI and GPV testing procedures for evaluating both IBE and PBE in three-period crossover studies. We also compare the efficiency of the two designs we consider, the three-period design given in the FDA guidance and an alternative design in which only the reference formulation is replicated, for both IBE and PBE. We show that the two tests are very similar for IBE, and that the GPV is significantly more powerful for PBE. We also show that a major shortcoming of the HSCI test, the discontinuity in the test criteria when changing from constant- to reference-scaling, is not present in the GPV test. We also find that the alternative three-period design is more efficient for both IBE and PBE.
|