All Times EDT
Keywords: Outliers, Bioequivalence Studies, Replicated Crossover Design
In bioequivalence studies outliers are defined as “subject data for one or more BA measures that are discordant with corresponding data for that subject and/or for the rest of the subjects in a study”, and hence their characterization relates to detecting either extreme instances or extreme subjects. Furthermore, an outlier can lead to a different conclusion concerning bioequivalence. All existing tests for outlier detection are based on different criteria for assessing the relative magnitude of the residuals from a given model.
Although there has been research related to outlier detection in a standard crossover 2x2 trial, limited work has been done about the fully replicated 2x4 (ABAB, BABA) and the partially replicated 3x3 (BAA, ABA, AAB) designs. Outlier characterization in replicated crossover designs is critical since in that context we can distinguish outliers caused by product failure from those due to subject-by-formulation interaction. Additionally, a large proportion of bioequivalence studies received by FDA are replicated crossover studies.
We consider the adaptation of existing methodology to the replicated crossover design, as well as the development of an enhanced version of such tests appropriate for these designs. Furthermore, the tests’ performance is assessed with respect to statistical power and type-I error probability.