Statistics

Abstract

Preparation, manufacturing, quality control and dispensing of medicinal products have always been associated with the pharmacist. Traditionally the pharmacist has therefore been trained in pharmaceutical analysis, focusing on analytical measurement of quality characteristics (identity, strength and purity) but the pharmacist was marginally trained in statistical quality control that is related to manufacturing processes.

The routine measurement of a characteristic or quantity in a dosage form by an accurate and highly precise analytical method will reduce the risk of rejecting batches when these truly comply or alternatively reduce the risk of falsely accepting batches, when batches do not comply.

The collection of analytical data or data with format go no-go is meaningless without a conclusion on rejection or acceptance of a batch. In addition, a decision is worthless when not properly based on sound scientific statistical principles: a carefully conceived sampling plan that considers issues such as sample size and variability.

The up scaling of production (batch size, numerous products in a single facility) and by consequence the up scaling of complexity within industry as well as within larger production facilities e.g. hospital pharmacies have led to the application of statistical quality control. Such an approach has been in use since the middle of the past century and is nowadays easy accessible by statistical software programs.

This chapter discusses several statistical principles that are used in pharmaceutical quality decisions, such as: normal distribution, rounding, confidence interval, standard deviation, outliers, operating characteristic curves, acceptance sampling. Examples have been embedded in a pharmaceutical context.