Expert opinion and low-quality evidence still drive strong recommendations in cardiology

You are here

Guidelines heavily impact clinical practice, standards of care and quality measures. Intuitively the expectation is that Class I recommendations would be supported by level A evidence. Instead, it appears that although the literature has grown, guideline authors continue to base many strong recommendations on expert opinion alone.

Clinicians often consult clinical practice guidelines for recommendations regarding diagnostic and treatment options. Ideally, recommendations are based on the best available evidence, with the strength of the recommendation directly linked to the strength of the evidence.

This should especially apply to cardiovascular disease guidelines, considering the relatively high volume of cardiology trials published in well-respected journals and over US$2b of US National Institutes of Health research funding allocated annually.

Guidelines from the American Heart Association and American College of Cardiology (AHA/ACC) and European Society of Cardiology (ESC) published over the last decade were recently analysed to determine the level of evidence supporting the many recommendations guiding current clinical practice.

The study, “Levels of Evidence Supporting American College of Cardiology/American Heart Association and European Society of Cardiology Guidelines, 2008-2018”, published in JAMA, examined 26 AHA/ACC guidelines and 25 ESC guidelines published from 2008-2018, which together provide over 6,000 recommendations.

Of the nearly 3,000 recommendation statements from the AHA/ACC, 43.4% receive a Class I (strongest) recommendation, yet only 14.2% of these Class I recommendations were based on level A evidence (supported by data from multiple randomised trials or a single, large randomised trial) according to the ACC scale.

The AHA/ACC guideline with the highest percentage of Class I recommendations supported by level A evidence (the updated blood cholesterol guideline) had just 35% of these recommendations supported by level A evidence.

11% of 188 Class I general cardiology recommendations relevant to cardiologists are supported by level C evidence (which includes consensus expert opinion, mechanistic studies, and low-quality registry data).

The European guidelines performed slightly better, with 21.5% of the Class I recommendations supported by level A evidence.

For guidelines with current and prior versions available, both the AHA/ACC and ESC had a decline in the proportion of Class I recommendations supported by level A evidence over time, suggesting a shift towards evidence on intermediate or disease-oriented outcomes.

A report published in JAMA in 2014 on the durability of AHA/ACC recommendations found that recommendations based on lower quality evidence were significantly more likely to be changed over time compared to those based on higher quality evidence.

Basing strong recommendations on expert opinion also perpetuates a cycle in which clinical practice is driven by expert opinion and a pathophysiology-reasoned approach to care rather than by recommendations derived from high-quality clinical trial evidence. At a minimum, guideline recommendations should prioritise recommendations based on the highest quality evidence, and highlight the lack of data supporting recommendations based on opinion.