Improving Youden plots by including analytical performance specifications

Abstract

Historically, Youden plots are made of a scatter plot of the results of two EQA samples with an elliptical confidence area around them. Because Youden plots compare individual laboratory results with respect to other laboratories, they intrinsically evaluate with state of the art. In the field of medical laboratory science, analytical performance specifications have been proposed and various EQA providers show these limits on Youden plots by horizontal and vertical lines near the limits of total error. This study shows the weaknesses of these limits of total error and proposes a new approach for drawing acceptance areas on Youden plots.The classical Youden plot is extended with two acceptance areas: A first acceptance area evaluates the reported results with respect to maximum allowed variability, a second acceptance area evaluates the results with respect to maximum allowed variability and bias. The ellipses are calculated using quantiles of the Chi-square and noncentral Chi-square distribution.Results show that the new approach allows evaluating results from individual laboratories and from measurement procedures by comparing the position and overlap of the confidence area and the acceptance areas. It is demonstrated that his approach controls the type I error better than the rectangular limits.