Risk of Bias Assessment and Figures

Choose a framework

Click Risk of bias in the sidebar (Team plan, on a cloud-saved review) and pick the framework that matches your study designs. The built-in frameworks are: Cochrane RoB 2 for randomized trials, ROBINS-I for non-randomized studies of interventions, QUADAS-2 for diagnostic accuracy studies, the Newcastle-Ottawa Scale in both cohort and case-control variants, the JBI checklists for randomized trials, cohort studies, case-control studies, analytical cross-sectional studies, and qualitative research, and AMSTAR 2 for appraising other systematic reviews.

If none fits, build your own: the custom framework builder lets you define your own domains and signaling questions, and custom frameworks behave exactly like the built-in ones, including in the figures.

Record judgments

Assessment works study by study and domain by domain. For each domain you record a judgment (Low risk, Some concerns, or High risk in RoB 2, with each framework using its own judgment scale) and free-text support for the judgment, the quotes and reasoning a reader would need to check your call. An overall judgment per study is derived and can be reviewed. The whole assessment exports as a CSV table for your supplement.

RoB 2, ROBINS-I, and QUADAS-2 also offer the full signaling-question mode: switch from Domain judgements to Signaling questions and answer each published signaling question on the Yes / Probably yes / Probably no / No / No information scale. Your answers suggest a judgment for each domain (the suggestion follows the direction of the published decision algorithms and stays fully editable), and each domain gets its own supporting-quote field so the audit trail sits next to the answer that justifies it.

Because RoB 2 is assessed per result rather than per study, any study can carry additional named per-outcome assessments: use the Outcome button next to Study level to add one (for example, Pain at 12 weeks). Outcome assessments appear as extra rows in the summary table and the CSV export.

Dual (blinded two-reviewer) assessment

The owner or an arbiter can switch on Dual assessment in the toolbar. Each reviewer then records an independent assessment that the other reviewer cannot see. When both assessments agree, the consensus is recorded automatically. When they disagree on any domain or on the overall judgment, the study is flagged as a conflict: the owner or an arbiter compares both blinded assessments side by side, adopts one as the consensus, or edits the consensus directly. Figures, summary tables, and exports always read the consensus assessment, never the individual votes.

Publication-ready figures

Two standard figures are generated from your judgments, ready for the manuscript. The traffic light plot shows the study-by-domain grid with one colored circle per judgment, the figure reviewers expect for RoB 2 and ROBINS-I. The stacked judgment bar plot shows judgment proportions per domain across all studies, the weighted summary figure that shows where your evidence base is weakest.

1. Complete judgments for every included study in your chosen framework.
2. Open the figures view and check both plots.
3. Tick Color-blind-safe palette if your journal requires accessible figures or you prefer the safer colors.
4. Download each figure with the SVG button (vector, best for journals) or the PNG button (raster, best for slides and documents).