Estimation of the Youden Index and its Associated Cutoff Point

• Published in: Biometrical journal Vol. 47; no. 4; pp. 458 - 472
• Main Authors: Fluss, Ronen, Faraggi, David, Reiser, Benjamin
• Format: Journal Article
• Language: English
• Published: Berlin WILEY-VCH Verlag 01-08-2005; WILEY‐VCH Verlag; Wiley-VCH
• Subjects: Algorithms; Applications; Biology, psychology, social sciences; Biomarkers - blood; Computer Simulation; Confidence Intervals; Coronary Artery Disease - blood; Coronary Artery Disease - diagnosis; Coronary Artery Disease - enzymology; Creatine Kinase - analysis; Data Interpretation, Statistical; Diagnosis, Computer-Assisted - methods; Diagnostic markers; Exact sciences and technology; Female; Humans; Kernel smoothing; Male; Mathematics; Models, Biological; Models, Statistical; Nonparametric inference; Parametric inference; Power transformation; Probability and statistics; Reproducibility of Results; Research Design; ROC Curve; Sciences and techniques of general use; Sensitivity and Specificity; Sensitivity Specificity; Severity of Illness Index; Statistics; Biometrics; Smoothing methods; Sensitivity Specificity; Error estimation; Diagnostic markers; Bias; Life science; Empirical method; Roc curve; Empirical distribution function; Power transformation; Optimal estimation; Statistical simulation; Mean square error; Kernel method; Statistical method; Diagnostic marker; Selection problem; Sensitivity; Specificity; Receiver operating characteristic curves; Kernel smoothing; Biased estimation
• ISSN: 0323-3847; 1521-4036
• DOI: 10.1002/bimj.200410135

The Youden Index is a frequently used summary measure of the ROC (Receiver Operating Characteristic) curve. It both, measures the effectiveness of a diagnostic marker and enables the selection of an optimal threshold value (cutoff point) for the marker. In this paper we compare several estimation procedures for the Youden Index and its associated cutoff point. These are based on (1) normal assumptions; (2) transformations to normality; (3) the empirical distribution function; (4) kernel smoothing. These are compared in terms of bias and root mean square error in a large variety of scenarios by means of an extensive simulation study. We find that the empirical method which is the most commonly used has the overall worst performance. In the estimation of the Youden Index the kernel is generally the best unless the data can be well transformed to achieve normality whereas in estimation of the optimal threshold value results are more variable. (© 2005 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)