Discordant and inappropriate discordant recommendations in consensus and evidence based guidelines: empirical analysis. (25th November 2021)

• Record Type: Journal Article
• Authors: Yao, Liang; Ahmed, Muhammad Muneeb; Guyatt, Gordon H; Yan, Peijing; Hui, Xu; Wang, Qi; Yang, Kehu; Tian, Jinhui; Djulbegovic, Benjamin
• Is Part Of: BMJ. Volume 375(2021)
• Publication Date: 2021-11-25
• Subjects: Medicine -- Periodicals; Medicine -- Periodicals; Medicine; Periodicals; 610
• DOI: 10.1136/bmj-2021-066045
• Languages: English
• ISSNs: 0007-1447

Abstract

Abstract: Objective: To investigate whether alignment of strength of recommendations with quality of evidence differs in consensus based versus evidence based guidelines. Design: Empirical analysis. Data source: Guidelines developed by the American College of Cardiology and the American Heart Association (ACC/AHA) and the American Society of Clinical Oncology (ASCO) up to 27 March 2021. Study selection: Recommendations were clearly categorised as consensus or evidence based, were separated from the remainder of the text, and included both the quality of evidence and the strength of the recommendations. Data extraction: Paired authors independently extracted the recommendation characteristics, including type of recommendation (consensus or evidence based), grading system used for developing recommendations, strength of the recommendation, and quality of evidence. The study team also calculated the number of discordant recommendations (strong recommendations with low quality evidence) and inappropriate discordant recommendations (those that did not meet grading of recommendations assessment, development, and evaluation criteria of appropriateness). Results: The study included 12 ACC/AHA guidelines that generated 1434 recommendations and 69 ASCO guidelines that generated 1094 recommendations. Of the 504 ACC/AHA recommendations based on low quality evidence, 200 (40%) proved to be consensus based versus 304 (60%) evidence based; of the 404 ASCO recommendations based on low quality evidence, 292 (72%) were consensus based versus 112 (28%) that were evidence based. In both ACC/AHA and ASCO guidelines, the consensus approach yielded more discordant recommendations (ACC/AHA: odds ratio 2.1, 95% confidence interval 1.5 to 3.1; ASCO: 2.9, 1.1 to 7.8) and inappropriate discordant recommendations (ACC/AHA: 2.6, 1.7 to 3.7; ASCO: 5.1, 1.6 to 16.0) than the evidence based approach. Conclusion: Consensus based guidelines produce more recommendations violating the evidence based medicine principles than evidence based guidelines. Ensuring appropriate alignment of quality of evidence with the strength of recommendations is key to the development of "trustworthy" guidelines.