BIS repetita non placent -- a statistical analysis of the number of sample preparations and number of injections required for chromatographic analyses.

This article attempts to answer the question of how many replicate sample preparations and replicate chromatographic injections must be done to provide accurate results in chromatographic analyses of pharmaceuticals. Using a random selection of chromatographic runs obtained with 1-3 replicate preparations and duplicate injections, the variance associated with preparation-to-preparation and injection-to-injection variability were estimated by a mixed-model statistical analysis. The analysis also predicted the probability that two injections of the same sample preparation are not in agreement with each other. Results indicated that, with modern chromatographic equipment, duplicate injections do not improve the precision. The number of replicate preparations needed to provide accurate results for various types of analysis depends on the type of sample and the desired tightness of the specification limits.