Prevention is better than cure: modifiable risk factors for heart failure better understood

This article refers to ‘A population-based study of 92 clinically recognized risk factors for heart failure: co-occurrence, prognosis and preventive potential’ by A. Banerjee et al., published in this issue on pages 466–480. Heart failure (HF) is a heterogeneous syndrome with several complex overlapping pathophysiological mechanisms and various classifications.1 It remains as one of the most significant causes of morbidity and mortality worldwide, with an estimated global prevalence of more than 64 million people (>8 per 1000 inhabitants).1 The lifetime risk for HF among those aged >45 years is as high as 46% in previous reports, although its incidence has declined over the last two decades.2, 3 Furthermore, mortality from HF in the community has significantly increased since the start of the COVID-19 pandemic.4 While advancements in HF therapies had a significant impact on the survival and quality of life among those living with the condition, prevention is better than cure as Desiderius Erasmus once told. However, the first step to prevention is a conceptual understanding of risk factors (RFs) that have preventative or modifiable potential. A myriad of RFs for HF has been previously described, with evidence sought from major randomized controlled trials (RCTs) to small scale observational studies. However, major RCTs often recruit highly selected cohorts and commonly exclude extremes of age, those with malignancies and major comorbidities, all of which are significant RFs for HF.5 Furthermore, RFs often do not exist in isolation, meaning that insights into their co-existence and preventative potential is of the greatest essence. In this issue of the Journal, Banerjee et al.6 seek to further our understanding on the prevalence and prognostic impact of a wide range RFs for HF based on their review of current evidence from RCTs and observational studies, and their national-level analysis of 170 000 individuals (aged >30 years) with incident HF in England between 1997 and 2017 from primary and secondary care electronic health record (EHR) datasets. The authors initially review the level current evidence from RCTs on strength of association between 92 RFs and incident HF and subsequently classify these RFs based on modifiable interventions (or lack thereof) from RCTs into three groups, namely interventions for primary prevention of HF (RCT-HF), cardiovascular disease (RCT-CVD), and no interventions (RCT-0). Several important findings can be extrapolated from this study. First, the authors demonstrate that only 10 RFs had a high (GRADE A) level of evidence on association with incident HF from RCTs or high-quality observational studies, while the rest had either GRADE B (n = 24) or C (n = 58) level of evidence. In terms of strength of association with incident HF, expressed as relative risk (RR), the authors found only four RFs with very strong (RR >3.5) association, including myocardial infarction, hypertrophic cardiomyopathy, pre-eclampsia, and atrial fibrillation, while another five RFs had a strong association (RR >2.5; hypertension, smoking, reduced cardiorespiratory fitness, connective tissue diseases and sinus node dysfunction). A concerning finding is the lack of association between most RFs (n = 64) and incident HF, although this alone does not negate such a relationship since absence of evidence is not evidence of absence. Quite the contrary, these findings highlight the need for prospective studies focussing on the association between important yet understudied RFs and incident HF, especially when plausible biological mechanisms support their association. For example, thyrotoxicosis often precipitates HF due to a high output state,7 while amyloidosis is a common cause of HF with preserved ejection fraction (HFpEF).8 However, both were found to have no evidence of association with incident HF by the authors based on their review of the literature. Another finding of concern is the lack of RCT-based evidence for many RFs as reported by Banerjee et al. Only seven RFs were found to reduce the risk of incident HF (unstable angina, stable angina, myocardial infarction, hypertension, cytostatic drugs, diabetes mellitus and renal failure), while only six RFs had observational evidence for reduced incident HF or RCT evidence for new-onset CVD: smoking, obesity, aortic valve disorders, amyloidosis, reduced physical activity and reduced cardiorespiratory fitness, although arguably the latter two may not necessarily be discrete categories. The authors show that 8% of patients with incident HF had no RFs of the 92 identified in their analysis, while 4 out of 10 patients had ≥4 RFs. Specifically, modifiable RFs such as hypertension, stable angina and obesity were the most prevalent among RF combinations, which highlights future preventative opportunities for incident HF among multimorbid patients with such RFs. These findings reaffirm those from a population-based case–control study of 962 incident cases of HF in Olmsted County, United States, between 1979 and 2002, where the population attributable risk for incident HF in the community was highest in patients with coronary artery disease and hypertension, together accounting for 40% of all new HF cases in their analysis.9 However, many of these modifiable RFs are currently undertreated. For example, Banerjee et al. find that only half of hypertensive patients were on first-line antihypertensive therapy (calcium channel blockers: 53.7%; angiotensin-converting enzyme inhibitors: 51.7%; angiotensin receptor blockers: 16.9%). This is likely due to undertreatment by physicians, whether at a primary or secondary care level, as well as poor adherence to medications, which has been previously shown to be around 50% at 1 year post commencement.10 Similarly, the utilization of evidence-based preventative therapies among patients with stable angina was relatively low in their analysis (statin use: 63.1%; antiplatelet therapy: 73.5%). The present findings have important clinical implications as they highlight the need for a more holistic approach to the management of patients with multiple comorbidities and the focus on established RFs for incident HF for which there is sufficient evidence from RCTs in terms of prognostic benefit. The authors should be applauded for the valuable insights they offer on current evidence around RCT-based interventions for RFs for HF and CVD as well as their systematic mapping of concomitant RFs, the findings of which emphasize the need for prospective sufficiently powered RCTs to examine understudied RFs in greater detail and establish causal inferences between single and multiple RFs and incident HF. Furthermore, the reusable International Classification of Diseases-Tenth Revision (ICD-10) codes reported in their study for almost all RF phenotypes (91/92) can aid future researchers, especially when conducting hierarchical clustering analyses from EHR datasets to look at clusters of RF combinations that confer the worst prognosis and/or are most amenable to preventative therapy.11 Notwithstanding, the findings reported in this study should be interpreted with several limitations in mind, mainly attributable to its observational nature. For example, one factor not accounted for in the present analysis was the length of exposure to RFs, which is known to correlate with the incidence of HF and CVD mortality, but this is rightly acknowledged by the authors and was due to variable missingness of certain RFs in their dataset. Furthermore, their dataset did not discriminate between acute and chronic HF or between the most recent classification of HF as per major international societies, which includes three categories based on left ventricular ejection fraction (LVEF), namely HF with preserved ejection fraction (HFpEF; LVEF ≥50%), mildly reduced ejection fraction (HFmrEF; LVEF 40%–49%) and reduced ejection fraction (HFrEF; LVEF<40%).12 Nevertheless, their comprehensive capture and mapping of a wide range of RFs in a large national dataset that is reflective of a real-world population, along with the insights they offer, has significant implications for clinical practice and future research directions. Conflict of interest: none declared.