Specificity

In diagnostic testing, specificity is the ability of a test to correctly identify individuals who do not have the condition, calculated as the number of true negatives divided by the total number of individuals who truly do not have the condition (true negatives plus false positives).^[4][1] It is also called the true negative rate. High specificity means fewer false positive results.

Specificity determines how effectively a test avoids false positive results in people who do not have the condition. A highly specific test produces few false positives, so a positive result from a highly specific test helps rule in the condition (the "SpPin" mnemonic: Specificity Positive rules In).^[3] The mnemonic is a simplification: how far a positive result raises the probability of the condition also depends on sensitivity and on prevalence, and in the example below a positive result from a test with 99.5% specificity is still usually a false positive. This property is especially important when the consequences of a false positive diagnosis are severe, such as unnecessary surgery, toxic treatment, or psychological harm. In screening programs, specificity determines the volume of false positives that require follow up evaluation, directly affecting the program's cost and acceptability.

Specificity is calculated from a two by two table comparing test results against a reference standard. The formula is: true negatives divided by the sum of true negatives and false positives. A specificity of 98% means the test correctly identifies 98 out of 100 people without the condition as negative, while 2 out of 100 receive false positive results.^[4]

Like sensitivity, specificity is calculated within one group, here the people without the condition, so prevalence does not enter its formula; it can still vary between settings with the spectrum of people tested, such as which other conditions they have. However, the practical implications of a given level of specificity depend heavily on prevalence. In a low prevalence population, even a small false positive rate translates to a large number of false positives relative to true positives, reducing the positive predictive value.^[2]

The tradeoff between sensitivity and specificity is governed by the threshold for a positive result. Raising the threshold (requiring a more extreme result to call the test positive) increases specificity, or at least does not lower it, and decreases sensitivity, or at least does not raise it. The ROC curve plots this tradeoff, and the optimal threshold depends on the clinical consequences of false positives versus false negatives.

Prioritize specificity when the consequences of a false positive are serious (such as invasive follow up procedures, toxic treatment, or significant psychological burden) and when a positive result will be used to initiate treatment. Report specificity with confidence intervals alongside sensitivity, predictive values, and likelihood ratios to give a complete picture of test performance. In sequential testing strategies, a highly specific confirmatory test is often paired with a highly sensitive screening test.

A blood biomarker test for early detection of ovarian cancer achieves a specificity of 99.5% and a sensitivity of 80%. In a screening population of 100,000 women where the prevalence of undiagnosed ovarian cancer is 0.04% (40 cases), the test would correctly identify 32 of the 40 cases (true positives) but would also produce approximately 500 false positives (0.5% of 99,960 women without cancer). The PPV would be 32 divided by 532, or about 6%, meaning that only about 1 in 17 positive results would indicate true cancer.