Table 5.1.
GRADE criteria for rating certainty of evidence
| Reasons for rating down a | Reasons for rating up 195,b |
|---|---|
| Risk of bias196 | Large magnitude of effect |
| Imprecision197 | Dose-response gradient |
| Inconsistency198 | All residual confounding would decrease magnitude of effect (in situations with an effect) |
| Indirectness199 | |
| Publication bias200 |
Applies to randomized studies.
Applies to non-randomized studies.