RoB 2: Risk of Bias Assessment for Randomized Trial Results

What is assessed?

RoB 2 assesses potential systematic distortion of a particular result in a randomized trial. One trial may require different assessments for different outcomes or time points.

Domain 1: randomization

• Check sequence generation and allocation concealment.
• Inspect baseline differences for possible randomization problems.

Domain 2: deviations

• Specify whether the effect of assignment or adherence is of interest.
• Examine trial-context deviations and whether the analysis addresses them.

Domain 3: missing data

• Review how much data are missing and why.
• There is no universally safe dropout percentage; consider whether missingness depends on the outcome.

Domain 4: outcome measurement

• Assess method suitability and consistency.
• Consider whether knowledge of intervention could influence outcome assessment.

Domain 5: selective reporting

• Compare registry, protocol and statistical plan with the published outcome and time point.
• Look for selection among multiple eligible analyses.

Judgment and documentation

• Use signalling questions and record evidence for every decision.
• Judgments are low risk, some concerns or high risk; overall risk follows the least favorable relevant domain.
• RoB 2 is not a numerical quality score or a substitute for GRADE certainty.

Common mistakes and clinical pearl

• Do not assume randomization prevents all bias.
• Do not label an entire paper low risk without specifying the result.
• Document exact supporting pages or quotations for reproducibility.

Summary

A reproducible assessment names the result, clarifies the target effect, evaluates all five domains and explains each judgment.