ABSTRACT
The advent of PAT, Process Analytical Technologies, offers the possibility of large scale monitoring of tablets as they come off the press. The most rapid techniques allowing the largest sample sizes are based on reflectance spectroscopy. As these techniques sample only a portion of the tablet, it is critical to prove that sampling a portion has a larger (and, therefore, more conservative) statistical variance than sampling the entire. Once demonstrated, a partial sampling technique, for example, a Near Infrared (NIR) sensor, should be able to provide a safe bracket for content uniformity evaluation. It is the purpose of this report to support the claim that the coefficient of variance (CV) from sampling a part of a dosage form cannot be smaller than the CV from sampling the whole dosage form. This hypothesis will be supported in this study by both a statistical proof and experimental data acquired from a model tablet system.
