Article type
Year
Abstract
Background: Systematic reviews should include rigorous risk-of-bias assessments of eligible studies. Although a large majority of Cochrane reviews address the effects of interventions, there is increasing interest in systematic reviews of other types of problem. One important category is systematic reviews on the effects of environmental and occupational exposures, for which eligible studies are almost always observational in nature. The ROBINS-I tool (BMJ 2016; 355: i4919) has become widely used to assess the risk of bias in non-randomized studies of interventions but is not appropriate for studies of exposures.
Objectives: To develop a new tool for assessing risk of bias in non-randomized studies of exposures, for application to studies of environmental and occupational exposure.
Methods: Between 2017 and 2020 a workgroup of methodologists, content experts and systematic reviewers with experience of relevant topic areas adapted ROBINS-I to create a new tool, ROBINS-E, for assessing risk of bias in non-randomized studies of exposures other than interventions, including environmental and occupational exposures. The preliminary instrument underwent piloting during a meeting in Bristol, UK, in October 2019.
Results: ROBINS-E features preliminary considerations of risk of bias within the review protocol; specification of a ‘target experiment’; use of signaling questions leading to algorithm-guided risk of bias judgments; assessment of bias within seven domains (confounding, measurement of exposure, selection of participants into the study, post-exposure interventions, missing data, measurement of outcomes, selection of reported results); and derivation of an overall risk of bias judgment for the assessed result. Pilot testing identified areas requiring additional refinement, including signaling questions within the domains of confounding, exposure assessment, and selection of participants into the study, to expand on the concept of ‘post-exposure interventions’, and modifications to the algorithm-guided risk of bias judgements. Piloting served as a mechanism to collect examples of instrument application across a wide range of environmental and occupational health topics.
Conclusions: Proposed changes to ROBINS-I will make ROBINS-E suitable for assessing studies of exposures. Further revisions will incorporate feedback from pilot testing and application across a spectrum of environmental and occupational health topics.
Patient or healthcare consumer involvement: No
Objectives: To develop a new tool for assessing risk of bias in non-randomized studies of exposures, for application to studies of environmental and occupational exposure.
Methods: Between 2017 and 2020 a workgroup of methodologists, content experts and systematic reviewers with experience of relevant topic areas adapted ROBINS-I to create a new tool, ROBINS-E, for assessing risk of bias in non-randomized studies of exposures other than interventions, including environmental and occupational exposures. The preliminary instrument underwent piloting during a meeting in Bristol, UK, in October 2019.
Results: ROBINS-E features preliminary considerations of risk of bias within the review protocol; specification of a ‘target experiment’; use of signaling questions leading to algorithm-guided risk of bias judgments; assessment of bias within seven domains (confounding, measurement of exposure, selection of participants into the study, post-exposure interventions, missing data, measurement of outcomes, selection of reported results); and derivation of an overall risk of bias judgment for the assessed result. Pilot testing identified areas requiring additional refinement, including signaling questions within the domains of confounding, exposure assessment, and selection of participants into the study, to expand on the concept of ‘post-exposure interventions’, and modifications to the algorithm-guided risk of bias judgements. Piloting served as a mechanism to collect examples of instrument application across a wide range of environmental and occupational health topics.
Conclusions: Proposed changes to ROBINS-I will make ROBINS-E suitable for assessing studies of exposures. Further revisions will incorporate feedback from pilot testing and application across a spectrum of environmental and occupational health topics.
Patient or healthcare consumer involvement: No