Quality control of drug assays

Previously published work relating to quality control of drug assays has dealt mainly with interlaboratory comparisons. It is suggested that participants in external quality assessment schemes wil derive the maximum benefit from their participation if a suitable intralaboratory quality control scheme is established. Such a scheme should be capable of providing estimates of within- and between-batch imprecision over the concentration range for which the assay is used and give an indication of the development of systematic error (relative to previous performance). A scheme using serum pools at 3 concentrations, each to be analysed at least in duplicate, is recommended. Adaptations of this scheme, to be used when the number of patient samples per batch does not justify this number of control samples, are discussed in relation to various analytical situations. A ratio of 1 control sample to 10 patient samples is considered reasonable. The interpretation of results from external quality assessment schemes using performance indices, graphical assessment of 'bias' and standard deviation interval (SDI = bias divided by standard deviation) and the statistical evaluation of proportional and additive error by linear regression (least squares) analysis is discussed. In some cases it has been possible, from the results of interlaboratory studies, to show that a particular method will not generally give satisfactory results when compared either with other currently available methods, e.g. UV - spectrophotometric methods for phenobarbitone, phenytoin, carbamazepine and theophylline, or on the grounds of clinical requirements, e.g. doubling dilution broth techniques for gentamicin. Improvement in intralaboratory reproducibility would facilitate the identification of technical factors contributing to interlaboratory variation.