The use of three partial areas for establishing bioequivalence and estimation of sample size for equivalence studies

Methods of adjusting for multiple comparisons and for sample size estimation for equivalence testing are discussed. In paper 1, a method using multiple areas under the concentration curve as a surrogate for absorption rate in generic drug evaluation is proposed. Additionally in Paper 1, since multiple comparisons are recommended, a method that adjusts the equivalence criteria by using the percentage points from the correlated maximum multivariate Student's t distribution is proposed. In Paper 2, assuming α = 0.10 and β = 0.20, the similarity between the currently recommended method for sample size estimation for equivalence testing and the recommended method for difference testing is demonstrated. It is also demonstrated that a non-iterative method of sample size estimation is superior to the iterative method. Additionally, a process using multiple pilot studies and one simulated study for each pilot study is shown to be a superior evaluation method for determining the appropriateness of sample size. In paper 3, two new methods of sample size estimation for use in equivalence testing are proposed. The method requiring the solution of a nonlinear equation tends to overestimate required sample size. In the second process, sample sizes are estimated for 1000 pilot studies, results of equivalence testing of simulated studies based on the estimated sample sizes are fit to a probit model and the sample size is selected to provide an appropriately chosen chance of establishing equivalence. This second process provides stable and adequate estimates of sample sizes for future experiments and is the method recommended.