The GRADE System for Rating Clinical Guidelines

Professional medical groups commonly issue clinical practice guidelines. Such guidelines are traditionally the result of consensus conferences or expert panels and represent attempts to synthesize— from the best available evidence and expertise—practical guidance on the best possible care. Beyond issuing a guideline, many organizations have felt the need to provide a grading of each guideline’s quality, thereby conveying to the reader a sense of the confidence that might be placed in it. This article addresses only the grading of guidelines, not their use or development. The idea that evidence in the medical literature should be graded was initially proposed in publications from McMaster University [1–3], with the idea of categorizing individual studies into grades of reliability ranging from randomized controlled trials (most reliable) to case reports with expert opinion (least reliable). Grading of guidelines followed, but this has been besieged with problems. To give one example, a guideline by Ferraris and colleagues gave the use of aprotonin during high-risk cardiac surgery a ‘‘highgrade’’ recommendation [4], but this intervention was subsequently shown to increase mortality [5]. The pursuit of better approaches to grading guidelines has resulted in GRADE (Grades of Recommendation Assessment, Development and Evaluation), introduced in 2004 [6]. GRADE has been adopted ‘‘unchanged or with only minor modifications’’ by national and international professional medical societies, health-related branches of government, health care regulatory bodies, and UpToDate, an on-line medical resource that is accessed by trainees and physicians in most US academic medical centers (Box 1) [7,8]. The developers of the GRADE system emphasized consistency in the rating of guidelines, as well as a wish to incorporate, and distinguish between, the ‘‘strength’’ of each guideline and the ‘‘quality’’ of the underlying studies (i.e., evidence) upon which it is based. Yet there is a central paradox: while GRADE has evolved through the evidence-based medicine movement, there is no evidence that GRADE itself is reliable.