Statistical Inference of the Probability of Passing the USP Dissolution Test

Abstract

In pharmaceutical manufacturing, dissolution test for immediate-release dosage forms is routinely performed to ensure the drug products conform to the quality standard specified in the United States Pharmacopeia and National Formulary (USP/NF). The USP dissolution test is a complex three-stage dissolution test involving the inference of individual dissolution measurement or/and sample mean of dissolution measurement of all dosage units. Passing the USP dissolution test once does not necessarily assure that the same batch of drug products would pass again if tested later. For the purpose of inferring the dissolution quality of a batch of drug products, a statistical inference based on acceptance limit is considered. The statistical inference assures a high probability of passing the USP dissolution test with a prescribed level of confidence. There is no closed-form expression of the true probability of passing the USP dissolution test. In the literature, the true probabilities of passing the USP dissolution test are usually approximated based on the lower probability bound. The available statistical inferential procedures based on the lower probability bound give an approximate and conservative inference. In this article, we propose the derivation of the true probability of passing the USP dissolution test based on the numerical computation. The probabilities of passing the USP dissolution test derived from the proposed numerical computation are no appreciable difference from the simulated true probabilities. The statistical inferential procedure via acceptance limits for various sample sizes of dissolution measurements and probabilities of passing the USP dissolution test is also provided.