QUADAS-2 Explained

If your systematic review is about how well a test detects a condition — not whether a treatment works — RoB 2 and ROBINS-I are the wrong tools. QUADAS-2 is built specifically for diagnostic and screening test accuracy studies, and using the wrong tool here is a common, avoidable error.

The four domains

1. Patient selection — was the sample selected in a way that avoids bias (e.g., consecutive or random sampling, appropriate exclusions)?
2. Index test — was the test being evaluated conducted and interpreted without knowledge of the reference standard result?
3. Reference standard — is the reference standard likely to correctly classify the target condition, and was it interpreted blind to the index test?
4. Flow and timing — did all patients receive the same reference standard, and was the interval between index test and reference standard short enough to avoid the condition changing in between?

Signalling questions and applicability

Each domain has specific "signalling questions" (yes/no/unclear) that guide the risk-of-bias judgment for that domain — this structured approach is what distinguishes QUADAS-2 from a general impression of study quality. The first three domains are also rated separately for applicability — whether the study's patients, test, or reference standard match your review's specific research question, which is a distinct concern from bias.

Tailoring QUADAS-2 to your review

Unlike RoB 2, QUADAS-2 is meant to be tailored: you define review-specific guidance for what counts as a "yes" or "no" for each signalling question before you start rating studies, based on your specific target condition and tests. Skipping this customization step is one of the most common ways QUADAS-2 gets misapplied.

For a comparison of QUADAS-2 against the tools used for intervention studies, see our guide to RoB 2 vs ROBINS-I vs Newcastle-Ottawa Scale.