Skip to main content


Optimal use of GRADE in Cochrane reviews to inform clinical practice guidelines

Date and Location




Saturday 21 September 2013 - 10:30 - 12:00


Presenting author and contact person

Presenting author

Ignacio Neumann

Contact person

Ignacio Neumann
Abstract text
Background: The GRADE approach (short for Grading of Recommendations Assessment, Development and Evaluation) offers a transparent, explicit and systematic way to summarize evidence, rate its quality, and move from evidence to recommendations. Cochrane reviews are a valuable resource to inform trustworthy clinical practice guidelines. However, systematic reviews are not always conducted or reported in a way that facilitates their use for guideline development. An optimal use of GRADE in Cochrane reviews could help to enhance their usefulness. Methods: Through an iterative process, we developed a set of criteria to evaluate the optimal use of GRADE. In this report, we focus on the criteria concerning the synthesis and evaluation of the evidence, which are relevant for systematic reviewers. Results: We developed six criteria for the optimal use of GRADE. Three are relevant for Cochrane reviewers (Table1): 1. The outcomes selection should be driven by what it is important to patients instead of what it is reported in primary trials. 2. Reviewers should provide effect-estimates for all the patient important outcomes relevant for decision making, including relative and absolute measures. 3. Finally, the reasons to rate down and up the confidence in effect-estimates should be explicit and described in detail, ideally providing an Evidence Profile in addition to the Summary of Finding table. Conclusions: The adherence to a simple set of criteria could enhance the usefulness of Cochrane reviews for the development of clinical practice guidelines. Table 1 – Optimal use of GRADE in systematic reviews 1. Were all the patient important outcomes relevant for decision making considered? 2. Did the systematic review provide explicit relative and absolute effect estimates for all the relevant outcomes? 3. Were the judgements about the confidence in the effect-estimates explicit?