For authors
Risk of Bias & Certainty
Use a validated, design-appropriate tool. Two assessors should normally work independently, with a documented process for resolving disagreements.
| Evidence type | Expected instrument |
|---|---|
| Randomized trials | RoB 2 |
| Non-randomized intervention studies | ROBINS-I |
| Diagnostic accuracy studies | QUADAS-2 or its current validated successor where applicable |
| Prediction models | PROBAST |
| Systematic reviews in umbrella reviews | ROBIS and/or AMSTAR 2 |
| Other designs | Appropriate JBI or other validated design-specific instrument |
| Quantitative certainty | GRADE or a justified alternative |
| Qualitative findings | GRADE-CERQual |
Submission requirements
- State the tool and version.
- Report assessor independence and conflict resolution.
- Do not create unvalidated summary scores unless the instrument permits them.
- Provide item-level judgements and supporting quotations or reasons.
- Explain any modification of a published tool.