KONSİL

Glossary · Study designs

What is a diagnostic accuracy study?

A diagnostic accuracy study measures how well a test under evaluation, the index test, detects a disease or condition by comparing it with a reference standard in the same people. STARD 2015 notes that results typically take the form of sensitivity and specificity estimates.

STARD 2015 describes diagnostic accuracy studies as ones in which the test under study is set against a clinical reference standard, with results typically reported as estimates of sensitivity and specificity. Its glossary calls the index test the test under evaluation, the target condition the disease or condition the test is expected to detect, and the clinical reference standard the best available method for establishing whether the target condition is present; an error-free reference standard would be a gold standard. Sensitivity is the share of people who truly have the target condition and test positive; specificity is the share of those without it who test negative.

STARD stresses that a test has no single fixed sensitivity and specificity: the values vary with the setting, with how patients present and with the tests they have already had. The explanation paper adds that differences in how eligible patients are identified can change the disease spectrum and the prevalence in the study sample, and so the accuracy estimates, so STARD asks reports to say how, where and when participants were recruited.

STARD 2015 is a list of 30 essential items (BMJ 2015;351:h5527). Item 5 asks whether the plan for data collection preceded the tests (prospective) or followed them (retrospective).

Example

In a hypothetical study, a rapid antigen test (the index test) and a PCR test (the reference standard) are both given to the same patients. If 90 of 100 PCR-positive patients also test positive on the antigen test, sensitivity is 90%; if 190 of 200 PCR-negative patients test negative, specificity is 95%.

A point to watch

STARD stresses that a test has no single fixed sensitivity and specificity; the values change with the setting and the patient spectrum. STARD’s glossary also defines the clinical reference standard as the best available method and a gold standard as an error-free reference standard, so by those definitions the two terms are not the same.

Related terms

Guides and tools

Sources

  1. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies — BMJ, 2015
  2. STARD 2015 guidelines for reporting diagnostic accuracy studies: explanation and elaboration — BMJ Open, 2016
  3. MeSH: Sensitivity and Specificity — U.S. National Library of Medicine, 2018
  4. STARD 2015 kriterlerinin Türkçe uyarlaması — Türkiye Klinikleri Biyoistatistik Dergisi (EQUATOR Network), 2019