Summary: | OBJECTIVES: To develop a framework for the assessment of the risk of bias and confounding against causality from a body of observational evidence, and to refine a tool to aid in identifying risk of bias, confounding, and precision in individual studies. METHODS: In conjunction with a Working Group, we sought to develop an overarching approach to assess the effect of confounding across the body of observational study evidence and within individual studies. We sought feedback from Working Group members on critical sources of bias most common to each observational study design type. We then refined and reduced the set of "core" questions that would most likely be necessary for evaluating risk of bias and confounding concerns for each design and refined the instructions provided to users to improve clarity and usefulness. RESULTS: We developed a framework that identifies additional steps necessary to evaluate the validity of causal claims in observational studies of benefits and harms from interventions. With the help of the Working Group, we narrowed the list of RTI Item Bank questions for evaluating risk of bias and precision from 29 to 16. Working Group members also provided their opinion of the most important questions for assessing risk of bias for four common observational study design types. CONCLUSIONS: Attributing causality to interventions from such evidence requires prespecification of anticipated sources of confounding prior to the review, followed by appraisal of potential confounders at three levels: outcomes, studies, and the body of evidence. We propose a substantial expansion in the critical appraisal of confounding when systematic reviews include observational studies for evaluation of benefits or harms of interventions. Questions about burden, reliability, and validity remain to be answered. Consensus around specific items necessary for evaluating risk of bias for different types of observational study designs does not yet exist
|