464 – Novel Approaches for Complex Biomedical Data
Dissolution Acceptance Limits Based on Parametric Tolerance Intervals
Richard Lewis
GlaxoSmithKline
ASTM Standard E2709 provides an approach for establishing confidence in passing the USP <711> dissolution test procedure for immediate-release dosage forms. An ISPE Good Practice Guide on process validation uses this approach to calculate acceptance limits for sample standard deviations. The resulting acceptance limits are extremely conservative, however, due to a poorly-shaped joint confidence region for the lot mean and standard deviation. A parametric tolerance interval approach provides wider acceptance limits with much better statistical properties by using a better partition of the parameter space. The improvement in acceptance limits can be considerable. As an example, consider establishing 90% confidence of 95% probability of passing USP <711>. For a sample size of 6 dosage units, the tolerance interval approach provides acceptance limits that are 1.5 to 2.1 times greater than those in the ISPE Good Practice Guide. For a sample size of 90 dosage units, the tolerance interval approach provides acceptance limits that are 1.13 to 1.35 times greater than those in the ISPE Good Practice Guide. An example is provided to show that the improved acceptance limits provide a meaningful reduction in decision errors.