eviStreams workflow guide
Risk-of-bias assessment with RoB 2 in eviStreams
Updated
Which risk-of-bias assessments does eviStreams support?
eviStreams supports RoB 2 assessment for parallel-group randomized trials. Reviewers work through the assessment with supporting evidence and resolve differences through consensus. AI assistance is available to support review; reviewers make the recorded judgments.
Check that the tool matches the study design
Start by confirming the study design and the result you are assessing. The eviStreams workflow described here is for parallel-group randomized trials. Do not assume that it covers nonrandomized studies, cluster-randomized trials, or crossover trials.
Use the official RoB 2 guidance to define the assessment and interpret its questions. This product guide explains how to organize work in eviStreams; it does not replace the tool’s methodological guidance.
Prepare the assessment with your team
Agree on the assessment scope before reviewers begin. Make the relevant study reports and supporting materials available, and identify the outcome and result under consideration.
- Set up the assessment protocol and the results to be assessed.
- Review the available evidence for each signaling question.
- Check AI suggestions against the source before accepting or changing them.
- Complete independent reviewer assessments using the agreed approach.
- Compare differences and record the consensus judgment and supporting rationale.
Separate missing information from a supported judgment
A paper may not describe every methodological detail needed for assessment. Record what the available evidence supports and follow the official guidance for missing information. A plausible AI explanation is not a substitute for a passage in the study reports.
Source excerpts should be checked in context. A statement about one outcome or analysis may not support a judgment about another. When reviewers disagree, compare both the evidence and their interpretation of the assessment question.
Use the recorded assessment in your review
Keep the final judgments and their rationale available alongside the study data. Before using them in synthesis or reporting, check which assessments are complete and which still require a decision.
A risk-of-bias judgment and an extraction accuracy check answer different questions. Correctly transcribing a number does not establish that the study result has low risk of bias. Review both as part of the evidence workflow.
| Decision | What to document |
|---|---|
| Scope | The study design, outcome, and result being assessed. |
| Evidence | The study report or supporting material used for the response. |
| Consensus | The final judgment and the rationale for resolving reviewer differences. |