A Generalized Pivotal Quantity Approach to the Parametric Tolerance Interval Test for Dose Content Uniformity Batch Testing

Abstract

The Food and Drug Administration (FDA) has proposed a parametric tolerance interval test (PTIT) for batch-release testing of inhalation devices. The proposed test examines dose uniformity based on several inhalation units from a batch, with two observations per unit. An underlying assumption is that the observations are a random sample from a univariate normal distribution. Because there are two observations per unit, it may be more appropriate to model the data as stemming from a bivariate normal distribution. We take a bivariate approach and use generalized confidence interval methodology to derive a parametric tolerance interval for the distribution of doses within a batch. We then use Monte Carlo simulation to compare results based on this bivariate approach with those based on the FDA-proposed PTIT.