Health Policy$ense

Guidelines Change, and Some Guidelines Change More than Others

New Study Reviews the 'Durability' of Clinical Recommendations

Lest we forget that guidelines are (and should be) living, breathing documents, LDI Senior Fellows Mark Neuman and Sandy Schwartz provide an excellent reminder in a new study in JAMA that reviews the “durability” of recommendations across serial versions of the same cardiology guidelines.

Neuman-Schwartz
Penn's Mark Neuman, MD, and J. Sanford (Sandy) Schwartz, MD

In a review of 11 American Heart Association/American College of Cardiology guidelines issued from 1998-2007, and revised from 2006-2013, the authors find that 80% of 619 index recommendations stayed the same, 9% were downgraded or reversed, and 11% were omitted from subsequent versions.

Not surprisingly, the likelihood that recommendations changed varied by the strength of the initial evidence. Ninety per cent of recommendations based on multiple randomized studies were retained, compared to 81% of those based on one randomized trial or observational data and 74% of those based on expert opinion. After accounting for guideline-level factors, recommendations based on opinion were 3 times as likely to be downgraded, reversed, or omitted as those based on multiple randomized studies.

The authors note the importance of these findings:

[O]ur results may have important implications for health policy and medical practice. The categorization of medical evidence, through guidelines, into stronger and weaker recommendations, influences definitions of good medical practice and informs efforts to measure the quality of care on a large scale. Our findings stress the need for frequent reevaluation of practices and policies based on guideline recommendations, particularly in cases where such recommendations rely primarily on expert opinion or limited clinical evidence. Moreover, our results suggest that the effectiveness of clinical practice guidelines as a mechanism for quality improvement may be aided by systematically identifying and reducing unwarranted variability in recommendations. Finally, our work emphasizes the importance of greater efforts on the part of guideline-producing organizations to communicate the reasons that specific recommendations are downgraded, reversed, or omitted over time.

Oh, and there’s an excellent editorial accompanying the article.