Critical appraisal of the CLSI guideline EP09c "measurement procedure comparison and bias estimation using patient samples".

Abstract

In laboratory setting evaluating the agreement between two measurement methods is a very frequent practice. Unfortunately, the guidelines to refer to are not free from criticisms from a statistical methodological point of view. We reviewed the Clinical and Laboratory Standards Institute guideline EP09c, 3rd ed. pointing out some drawbacks and some aspects that have not been well defined, leaving situations of uncertainty and/or of excessive subjectivity in the judgement. We have stressed the need of having replicates to estimate the systematic and the proportional biases of the measurement methods to be compared. Indeed, unequal variance of the two measurement methods gives a slope and intercept of the regression between the difference and the mean of the two values of the measurement methods to be compared that can be absolutely calculated from their means, their variances and their correlation coefficient. So,it is not possible to disentangle true from spurious biases. For laboratory professionals we have developed a worked exemplification of an agreement assessment. We have stressed the need of other approaches than the classic Bland and Altman method to calculate the systematic and proportional biases of two measurement methods compared for their agreement in a study with replicates.