The multivariate assessment of bioequivalence.

We propose an approach for comparing directly the reference and test blood concentration x time profiles of subjects in a bioequivalence trial under a crossover design. We compare the test and reference profiles via the construction of two regions, namely, an acceptable region based on the reference profile and another region based on reflecting the test profile about the reference profile. Also, we describe how to construct these regions if the assay for measuring the concentration has a lower limit of detection and how to adjust the reference and test concentrations of the n subjects participating in a bioequivalence trial. If the test and reference formulations are bioequivalent, then the area displaced by the latter region should be less than the area displaced by the acceptable region. Therefore, we construct a ratio of these areas for each subject. Then we calculate upper confidence limits for quantiles, either under theassumption of lognormality or based on a bootstrapping algorithm, to investigate population and/or individual bioequivalence.