Guideline Recommendations for Cardiac Resynchronization Therapy Evolve But Does Clinical Practice Match the Pace?

This article refers to ‘Effect of expanding evidence and evolving clinical guidelines on the prevalence of indication for cardiac resynchronization therapy in patients with heart failure’ by L.H. Lund et al., published in this issue on pages xxx. Lund et al.1 in their article employ the Swedish Heart Failure Registry (SwedeHF) to analyse both the actual number of patients with a cardiac resynchronization therapy (CRT) device and the number of patients with a recommendation level I/IIa for CRT. Out of a total of 99 189 patients entered into the registry between 2000 and 2014, they identify 17 193 patients with an ejection fraction (EF) ≤39% and sufficient data recorded for these patients. In order to determine the prevalence of CRT indication, they used the European Society of Cardiology (ESC) guidelines issued by both the European Heart Rhythm Association (EHRA) and the Heart Failure Association (HFA) from 2005 to 2016 and found patients who qualified for CRT with either a I or IIa recommendation (‘is recommended or should be considered’). They concluded that with the expansion of 2013 CRT guideline recommendations to include patients in New York Heart Association (NYHA) class II the prevalence of the indication for CRT rose from 24.5% to 30.0% from 2005/2007 when the first ESC guidelines to include CRT were published. The prevalence then dropped to 26.8% when the lower limit for QRS duration was raised to 130 ms by the latest ESC guidelines.2-5 Notably, the actual number of patients with a CRT device in this cohort was only 6.8%. Such underutilization of CRT in Sweden has been reported previously using data from SwedeHF.6, 7 In order to partially explain this low CRT implantation rate, the authors correctly point out that guidelines continually evolve as new evidence is reported and also require time to be adopted. However, they performed a sensitivity analysis using only data from 2010 onwards and found a similar low (7.0%) implantation rate in this population compared with the 24.5–30.0% expected from guideline recommendations. The authors remind us that 2010 is 5 years after the publication of the first pivotal clinical trials demonstrating decreased mortality following CRT implantation in appropriate patients.8, 9 Other potential explanations proposed for this low rate are limited awareness of CRT indications and low referral rates, although the sources responsible for these referrals are not described. The paper does not present an analysis of the potential factors associated with the underutilization of CRT. The reader is referred to a previous paper from the SwedeHF recently published in this Journal.10 This report identified a lack of non-cardiology care, female sex and age > 75 years among the factors associated with underutilization of CRT. The prevalence of atrial fibrillation (AF) in this patient subset of the SwedeHF population was 43%. This high percentage of AF patients highlights why registry data are so useful in demonstrating that the population is different from that of patients enrolled in randomized controlled trials (RCT) because of the rigid inclusion criteria of RCTs. SwedeHF includes substantially more elderly patients and women with considerably more co-morbidities as compared to RCTs and therefore perhaps provides us with a more realistic representation of the actual Swedish heart failure population. It has been estimated that less than one-third of patients over 64 years hospitalized with heart failure would actually qualify for entry into a RCT.11 The analysis does have some limitations but most of these are highlighted by the authors and adequately addressed by the four sensitivity analyses that do not dramatically vary from the primary analysis. One of these, of course, is the fact that the ESC guidelines recommend CRT implantation in patients with an EF ≤35% and not <39% as used in this analysis. This is a well-recognized limitation when using SwedeHF data for assessing CRT indication. SwedeHF groups patients by EF into the following categories: <30%, 30–39%, 40–49% and ≥50%, thus complicating the analysis of the indication for CRT. However, their sensitivity analysis using EF ≤30% found a similar trend in the temporal changes when compared to the primary analysis. This grouping is fortunate in that the third category potentially provides information about the mid-range EF group (40–49%) in which there is currently considerable interest. A major limitation when using SwedeHF data to analyse CRT utilization is that in patients actually implanted with a CRT device QRS duration and morphology are not recorded in the registry. Thus, there is some uncertainty about whether these patients accurately satisfy the indication for a CRT device. It is assumed that patients who have been implanted with a CRT device have the correct indication and these data are used in the primary analysis to calculate the prevalence of CRT indication. However, as shown by both European CRT surveys, a substantial number of patients with a CRT device in Europe are implanted with weak guideline recommendations.12, 13 The authors report that the actual percentage of CRT implantations was 6.8% for the entire period studied (2000–2014). However, it would have been interesting to observe how this percentage changed following publication of the new guidelines. Furthermore, there is also limited information presented about the patients who appear to qualify for a CRT device but did not receive one. It would also have been interesting to understand why patients who qualify are not ultimately implanted. A further important limitation using the SwedeHF registry in this analysis is that it does not contain information about the presence of atrioventricular block. The most recent ESC guidelines provide a IA recommendation for a CRT in patients with heart failure who have an indication for ventricular pacing and high degree atrioventricular block. Although these patients with a recommendation level I should be included in the primary analysis, they cannot be separately identified in this registry. This limitation leads to an underestimation of the prevalence of CRT indication according to the most recent guidelines. Finally, we must not forget that this is a registry and although registry analyses are likely to be more representative of the actual heart failure population than data from patients entered into RCTs, the data collected are generally less detailed, not validated and endpoints are not adjudicated. We are grateful to Lund et al. for another useful contribution demonstrating the productivity of a well-designed and well-organized registry. Sweden remains a leader in reporting ‘real-life’ data and sets the bar for other countries in the attempt to optimize health care delivery to the most appropriate patients. Conflict of interest: none declared.