A new statistical method for evaluating long-term analytical performance of laboratories applied to an external quality assessment scheme for flow cytometry

Abstract

The Belgian External Quality Assessment Scheme for Flow Cytometry evaluates the long-term analytical performance of participating laboratories by calculating a regression line between the target and reported values of each parameter for each laboratory during the past 3 years. This study aims to develop a method to find laboratories with aberrant variability or bias using robust techniques and to obtain robust estimates of the variability. A method is proposed to find outliers with respect to the individual regression line, followed by a step to find regression lines with excessive variability and finally a step to find regression lines with high bias. The model was applied to the results obtained by 52 laboratories for CD4%. From the 1340 data points, 35 were determined to be regression outliers. The second step revealed one regression line with excessive variability; the third step detected three regression lines with exceeding bias. The methodology allows assessment of the long-term performance of laboratories, taking into account samples with different target values. Outliers in the first step indicate accidental mistakes, outliers in the second and third step point to high analytical variability or bias.