What Is a Diagnostic Accuracy Study?

A diagnostic accuracy study compares the results of a test (the index test) with a reference standard in the same patients to estimate how well the test identifies people with and without the condition. Typical measures are sensitivity, specificity, predictive values and likelihood ratios.

Core design elements

• Index test: the test being evaluated, with defined thresholds.
• Reference standard: the best available method to determine true disease status.
• Participants: a representative sample of those in whom the test would be used in practice.
• Blinding: those interpreting the index test and the reference standard should not know the other result.
• Analysis: a 2×2 table and confidence intervals.

The 2×2 table

Sensitivity = TP/(TP+FN); specificity = TN/(TN+FP). Predictive values depend on prevalence. See Sensitivity vs Specificity.

Illustrative example

Invented for teaching. A rapid antigen test is compared with PCR in 500 symptomatic patients. PCR is positive in 100. The rapid test is positive in 80 of those 100 (sensitivity 80%) and negative in 380 of the 400 PCR-negative patients (specificity 95%). The test misses 20% of true cases, which matters if used to rule out infection.

Common biases

• Spectrum bias: including only clear cases and obvious controls inflates accuracy.
• Verification bias: only some patients get the reference standard.
• Incorporation bias: the index test forms part of the reference standard.

Reporting

Use the STARD 2015 checklist to report these studies completely.

Common pitfalls

• Reporting accuracy without confidence intervals.
• Using a poor reference standard.

Key takeaways

• Choose a good reference standard and a representative sample.
• Report sensitivity and specificity with confidence intervals and follow STARD.

This page is general educational information about research methods. It is not medical, legal or regulatory advice; follow your institution's ethics board and your target journal's instructions.