How to assess risk of bias with RoB 2 and ROBINS-I
Short answer
Use RoB 2 for randomized trials and ROBINS-I for non-randomized studies of interventions. Both work per result rather than per study, through domains of signalling questions that lead to a judgment for each domain and an overall judgment. Two reviewers assess independently, justify each answer with text from the report, and agree on the final judgments.
Step by step
- 1
Pick the tool that matches the study design
RoB 2 is for randomized trials, with variants for cluster-randomized and crossover trials. ROBINS-I is for non-randomized studies of interventions, such as cohort studies comparing treatments. Other designs need other tools, for example QUADAS-2 for diagnostic accuracy studies.
- 2
Specify what you are assessing
Risk of bias is assessed for a specific result: an outcome, at a time point, for a comparison. In RoB 2, also decide whether you are interested in the effect of assignment to the intervention (intention to treat) or the effect of adhering to it, because the questions in the deviations domain depend on it.
- 3
Work through the RoB 2 domains
RoB 2 has five domains: bias arising from the randomization process, due to deviations from intended interventions, due to missing outcome data, in measurement of the outcome, and in selection of the reported result. Signalling questions are answered yes, probably yes, probably no, no or no information, and an algorithm suggests low risk, some concerns or high risk for each domain.
- 4
Or work through the ROBINS-I domains
ROBINS-I compares the study with a hypothetical target trial and covers seven domains: confounding, selection of participants, classification of interventions, deviations from intended interventions, missing data, measurement of outcomes and selection of the reported result. Judgments are low, moderate, serious or critical risk of bias, or no information. Before starting, list the important confounders for your question.
- 5
Reach an overall judgment
The overall risk of bias for a result is generally at least as severe as the worst domain. In RoB 2, several domains with some concerns can also justify an overall high risk. Write a short justification for every judgment, quoting the report.
- 6
Assess in duplicate and present the results
Two reviewers assess independently and resolve disagreements. Present the judgments per domain, for example in a traffic-light table, and use them in the analysis and in the GRADE assessment of certainty.
Check you are using the current version
The risk-of-bias tools and their guidance documents are maintained by their developers and are revised from time to time. Download the current version and guidance from the official site (riskofbias.info) before you start, and cite the version you used.
How Prolytiq does this step
Prolytiq's risk-of-bias workspace lays out the signalling questions of RoB 2, ROBINS-I and other tools for each study, next to the PDF, and records each answer with its supporting quote and a per-domain and overall judgment. AI pre-fill, where it is enabled, proposes answers with quotes for reviewers to confirm or change; reviewers make every final judgment.
See the workflow →Frequently asked questions
What happened to the original Cochrane risk of bias tool?
RoB 2 (2019) replaced the original Cochrane risk-of-bias tool for randomized trials in Cochrane reviews. The main differences are the focus on specific results, signalling questions and algorithms that map answers to judgments.
Can I use RoB 2 for observational studies?
No. RoB 2 is designed for randomized trials. For non-randomized studies of interventions, use ROBINS-I; for observational studies of exposures, consider ROBINS-E.
Should I calculate a quality score?
No. The Cochrane Handbook advises against summary quality scores. Report domain-level judgments and justify them.
Methodological sources
- Sterne JAC et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ 2019
- Sterne JAC et al. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions. BMJ 2016
- Cochrane Handbook for Systematic Reviews of Interventions, chapters 8 and 25
Run your next systematic review in Prolytiq
From research question to publishable chart, with every decision logged and exportable for peer review.