Nonrandomized studies are an important method for evaluating the effects of exposures (including environmental, occupational, and behavioral exposures) on human health. Risk of bias in nonrandomized studies of exposures (ROBINS-E) is used to evaluate the risk of bias in natural or occupational exposure observational studies. This paper introduces the main contents of ROBINS-E 2022, including backgrounds, seven domains, signal questions and the operation process.
ObjectiveTo systematically interpret the updated Risk of Bias in Non-randomized Studies of Interventions Version 2 (ROBINS-I V2) in 2024, summarizing its key improvements, operational procedures, and clinical application value. MethodsThrough literature review and case studies, the improvements of ROBINS-I V2 were compared with the 2016 version, including the expansion of bias domains, refinement of signaling questions, and optimization of decision flowcharts. A retrospective study in stomatology was used to demonstrate the practical application of the tool. ResultsThe ROBINS-I V2 tool has restructured the hierarchy and refined the definitions of bias domains, optimized the evaluation processes across seven risk-of-bias dimensions, and minimized subjective judgment errors through standardized decision flowcharts. ConclusionROBINS-I V2 significantly improves the rigor of bias assessment in non-randomized intervention studies through its scientific design and standardized workflow. It is recommended for evidence quality grading and decision-making support in clinical research.