Construction and application of the robust Youden plot in an EQA program

Abstract

The Youden plot is a method of graphical data analysis used in inter-laboratory comparisons. The confidence ellipse of a traditional Youden plot can be distorted by non-normally distributed data. In this study, a robust Youden plot is constructed based on robust statistical parameters since these are scarcely affected by non-normally distributed data and this approach is applied in an external quality assessment (EQA) program. We consider the robust between-laboratory z-scores (z _b,i ) and within-laboratory z-scores (z _w,i ) obtained by the participants of EQA schemes regarding glucose concentration measurement results and their locations relative to the confidence ellipse in a robust Youden plot. It turned out that 98.1 % of the acceptable results (|z _b,i | ≤ 2 as well as |z _w,i | ≤ 2) are located inside the ellipse and that 92.9 % of questionable results with z-scores between 2 and 3 are located on or near the ellipse; while all the unacceptable results (at least one z-score equal to or larger than 3) are located outside the ellipse. These results indicate that, in an EQA program, it is reasonable to choose robust statistical techniques to construct the Youden plot and robust z-scores, in order to illustrate and assess the EQA data.