Adaptation of GRADE Definitions for Diagnostic Tests
| Quality of evidence | Study design | Decrease if… | Increase if… |
|---|---|---|---|
| High | RCT or prospective cross-sectional (or cohort) studies of patients with diagnostic uncertainty and direct comparison | Study limitations exist* | Test has low level of false-positive and false-negative results, and good data exist on beneficial consequences of avoiding false-positive and false-negative results (clinical outcomes) |
| Serious limitations (−1) | |||
| Very serious limitations (−2) | |||
| Important inconsistency is present (−1) | |||
| Directness (generalizability) is uncertain | |||
| Some uncertainty (−1) | |||
| Major uncertainty (−2) | |||
| Data are sparse or imprecise (−1) | |||
| Probability of reporting bias is high (−1) | |||
| Moderate | |||
| Low | Anything else | ||
| Very low |
↵* Such as nonrepresentative population, presence of selection bias, poor choice of gold standard, incomplete performance of tests (verification bias), lack of independent test interpretation (detection bias), and loss to follow-up (attrition bias).