P5711Limits of the cardiovascular knowledge base: results from a systematic analysis of the evidence substantiating the European Society of Cardiology clinical practice guidelines

Abstract

Abstract Background The clinical practice guidelines of the European Society of Cardiology (ESC) have a tremendous impact on cardiological practice. To enable clinicians to deliver the best care, substantiated knowledge has to ground these guidelines. However, recently many have voiced questions about the evidence levels forming the basis of guidelines. The latest analysis of cardiovascular guidelines was performed in the United States in 2009 and revealed that they were mainly based on lower level evidence or expert opinion. Moreover, a systematic analysis of the evidence base supporting the ESC guidelines has not been performed before, such an analysis is needed to embody the trust placed in them. Purpose To systemically evaluate the scientific evidence base underlying ESC guidelines. Methods We extracted all recommendations from ESC guidelines published on the ESC website since 2009, with their class of recommendation (treatment is beneficial [class I]; uncertain [class IIa/b]; harmful [class III]) and their level of evidence (multiple randomized controlled trials (RCTs)/meta-analyses [level A]; single RCT/large non-randomized studies [level B]; expert opinions/small studies [level C]). We assessed the distributions of recommendation classes, the underlying evidence levels and their ratios. Furthermore, we compared subsequent guidelines to investigate if evidence levels substantiating recommendations improved over time. Results In the past decade, 43 documents were published as guidelines on the ESC website; 5 were excluded because they concerned definitions, focused updates or position papers, leaving 38 for analysis. A total of 4704 recommendations were extracted, including 2313 (49%) class I, 1341 (29%) class IIa, 677 (14%) class IIb and 373 (8%) class III recommendations. 800 (17%) recommendations were supported by level A evidence; 1439 (31%) by level B, and 2465 (52%) by level C evidence (figure 1a). The majority of class I recommendations (i.e. treatment is beneficial) were based on level C evidence. The same was true for class II and III recommendations. When guidelines were updated there was a median increase of 32 (IQR, 14–40) recommendations. Overall, the number of recommendations supported by level A evidence fell by 1%, while level B and C evidence rose (17% and 16%) (figure 1b). Despite introducing 50 class I recommendations, only a minority of the recommendations were supported by level A evidence (3), less by level B evidence (−7) and the majority by level C evidence (54). Figure 1 Conclusion(s) This first systematic analysis of the ESC guidelines shows that half of its recommendations are grounded in expert opinions or based on small studies. Over time the number of recommendations increased, this increase was primarily based in expert opinions and small studies. To maintain professional and public trust in the ESC guidelines further research is essential to enlighten whether, why and which gaps exist in the cardiovascular knowledge base. Acknowledgement/Funding ZonMw Top-grant