Youden Index

The Youden index, also called Youden's J statistic, is a single summary measure of a diagnostic test's overall discriminative ability. It is calculated as sensitivity plus specificity minus one. It can range from −1 to 1, and for a useful test it lies between 0 and 1: 1 indicates a perfect test with no false positives or false negatives, 0 indicates a test that performs no better than chance, and a negative value means the test points in the wrong direction more often than chance.

Diagnostic accuracy studies are central to clinical medicine, public health screening programs, and health technology assessment. The Youden index provides a convenient way to compare tests or to identify the optimal cutoff point on a receiver operating characteristic (ROC) curve. It weights sensitivity and specificity equally and takes no account of prevalence or of the costs of each kind of error, so its optimal cutoff suits a setting where missed cases and false alarms matter about equally once the numbers of people with and without the condition are taken into account. Systematic reviewers evaluating diagnostic test accuracy frequently encounter the Youden index in primary studies and meta analyses, so understanding its calculation and limitations is essential for critical appraisal.

The Youden index is calculated as J = sensitivity + specificity - 1.

Sensitivity is the proportion of people with the condition whom the test correctly identifies as positive, and specificity is the proportion of people without the condition whom the test correctly identifies as negative. To find the cutoff that maximizes J, the researcher calculates J at every possible cutoff point along the ROC curve and selects the largest; on the plot, this is the point farthest above the diagonal chance line.

For example, if a biomarker test has sensitivity of 0.88 and specificity of 0.82 at a particular cutoff, J = 0.88 + 0.82 - 1 = 0.70. Comparing J values across multiple cutoffs or across competing tests identifies the threshold or test with the best overall discriminative performance.

Use the Youden index when comparing the overall accuracy of two or more diagnostic or screening tests, when identifying the optimal cutoff point for a continuous biomarker, and when conducting or appraising diagnostic test accuracy systematic reviews. Meta analyses of diagnostic studies usually pool sensitivity and specificity jointly with the bivariate or HSROC model rather than pooling J, but primary studies often report a Youden optimal cutoff, which reviewers need to appraise.

The Youden index weights sensitivity and specificity equally, which may not be appropriate when the clinical consequences of false negatives and false positives differ substantially. A test with J = 0.70 achieved through sensitivity of 0.95 and specificity of 0.75 has very different clinical implications than one with sensitivity of 0.75 and specificity of 0.95, even though both produce the same index value. Reporting J without the underlying sensitivity and specificity values deprives readers of essential clinical context. When the cutoff that maximizes J is chosen and then evaluated in the same sample, the reported sensitivity, specificity and J tend to be optimistic, because the cutoff is fitted partly to chance features of that sample; they should be checked in independent data.

A research team compares two rapid diagnostic tests for detecting tuberculosis in a low resource setting. Test A achieves sensitivity of 0.92 and specificity of 0.78, yielding J = 0.70. Test B achieves sensitivity of 0.85 and specificity of 0.91, yielding J = 0.76. While Test B has a higher Youden index, the public health team selects Test A for initial screening because its higher sensitivity misses fewer infections in a high prevalence population, then uses Test B as a confirmatory test.