Statistical Assessment of Mean Differences between Two Dissolution Data Sets *

Abstract

When comparing the dissolution data of a postapproval change product and a reference approval product, the goal is to assess the similarity between the mean dissolution values at the observed sample time points. The decision on accepting or rejecting the hypothesis that the two batches have similar dissolution is based on the evidence regarding whether the difference in mean dissolution values between the test and reference products is no larger than the maximum expected difference between any two batches of the approval product. When dissolution value is measured at a single time point, the confidence interval of the true difference between the two batches is compared with the prespecified similarity limits. When dissolution values are measured at multiple time points, a multivariate statistical procedure for difference assessment can be a generalized form of the t-statistic procedure. The proposed procedure is a modification and generalization of the regular bioequivalence test concept. The application of the proposed multivariate analysis procedure is illustrated using an example.