Browse the corpus
Walk the Even Hospital Database by book and chapter — the raw source passages that ground Ask, DDx, and the rest.
1 passage
New GRADE Evidence-to-Decision Framework for Pairwise and Multiple Comparisons (GRADE Guidance 45). Evidence-based decision making in health often requires comparison of multiple options for a given condition. The GRADE (Grading of Recommendations Assessment, Development and Evaluation) evidence-to-decision (EtD) framework provides a structured approach for moving from evidence to decisions but was originally designed for pairwise comparisons. Hence, there is a need to accommodate decision making based on multiple comparisons, especially with the increasing use of systematic reviews and network meta-analyses in guideline development. Furthermore, since the original EtD framework was developed, further relevant GRADE guidance has been developed. The aim of this work was to develop a new EtD framework to accommodate multiple comparisons and reflect current GRADE guidance. The new EtD framework was revised and developed through iterative discussion, feedback, and refinement by the GRADE EtD Project Group and the GRADE Working Group. Experiences and examples from guideline developers, methodological experts, and other stakeholders informed improvements in its structure and usability for multiple comparisons and were subsequently approved by the GRADE Working Group. This article describes the new EtD framework, which now includes 2 corresponding parts for reviews of pairwise and multiple comparisons. The authors describe application to a review with multiple comparisons for the different parts of the EtD framework: the question definition, which now includes the presentation of values of health outcomes and decision thresholds; the assessment section, where the new "net effect" criterion has been included; and the conclusion section, which includes an adaptation for multiple comparisons. The article provides examples and suggestions for presentation of findings. The framework does have limitations, in that its usability has not been tested across a broad spectrum of guideline development contexts.