Mean difference vs. variability reduction: tradeoffs in aggregate measures for individual bioequivalence. FDA...

Aggregate criteria for individual bioequivalence allow a tradeoff between difference in average bioavailability and reduction in within-subject variability. That is, a large difference in the average bioavailability between a test and a reference formulation can be offset by a sufficient reduction in variability of the test formulation. This offset could allow the test formulation to pass many individual bioequivalence criteria. We have identified 4 possible approaches for dealing with this tradeoff issue: say "No problem," since a reduction in variability is desirable; use disaggregate criteria; use general weighted forms of the individual bioequivalence criteria that weight the variance terms; and change the acceptable upper limits to reduce the impact of scaling to the reference formulation's within-subject variability. A dataset with a 14% increase in average bioavailability and a 48% reduction in within-subject standard deviation is used as an example of these issues.