Temporal trends in the outcomes of acute heart failure: between consolatory evidences and real progress

This article refers to ‘Temporal trends in mortality and readmission after acute heart failure: a systematic review and meta-regression in the past four decades’ by A. Kimmoun et al., published in this issue on pages 420–431. Acute heart failure (AHF) constitutes a broad spectrum of disease states, with heterogeneous clinical presentations, but commonly characterized by either a rapid onset or a progressive worsening of signs and symptoms, necessitating immediate treatment, and leading to urgent hospitalization.1 Despite many clinical trials conducted to date in these patients, rates of adverse outcomes in patients admitted for AHF remain very high, and currently there is not a single approved agent that is known to alter the natural history and/or clinical outcomes of AHF.2 Hospitalizations for AHF represent a change in the trajectory of the disease process and each successive readmission is associated with incrementally higher risk of mortality.3 Although short-term readmissions are due to haemodynamic congestion and socioeconomic factors limiting access to recommended treatments, medication adherence, self-care, and follow-up,4 long-term readmissions are the result of the continuous deterioration of the cardiac substrate and worsening of the associated non-cardiac comorbidities.4, 5 Consequently, a substantial number of patients with heart failure (HF) are rehospitalized for reasons not directly related to cardiovascular causes.6 This vulnerability to a diversity of illnesses may explain why cardiac interventions to prevent readmissions are limited to just a portion of the whole burden of hospitalizations.6 Most current post-discharge efforts focus on managing congestion and close haemodynamic monitoring,2 and although these are important goals, broader strategies to focus on HF-related comorbidities, patient-centred management, as well as innovative research strategies may be necessary. In this issue of the Journal, Kimmoun et al.7 report the results of a meta-analysis of 285 AHF studies between 1980–2017, representing more than 15 million patients from 70 countries and four continents. Primary outcomes were temporal trends, assessed by meta-regression, for 30-day and 1-year all-cause death and 30-day and 1-year all-cause readmissions. Overall, outcome rates were high, 7% for 30-day and 24% for 1-year all-cause death. The 30-day and 1-year all-cause readmissions were also high, 18% and 46%, respectively. Between 1980–2017, the authors reported a decline in 30-day all-cause death [odds ratio (OR) for a 10-year increment: 0.74, 95%confidence interval (CI) 0.61–0.91] that persisted at 1 year (OR 0.86, 95% CI 0.77–0.96). In contrast, 30-day readmission rate worsened (OR for a 10-year increment: 1.19, 95% CI 1.01–1.41), while 1-year all-cause readmission rate remained unchanged (OR 0.76, 95% CI 0.53–1.1). Importantly, temporal trends in AHF-related death and readmission in the past four decades did not differ among studied continents. A decline in AHF-related 1-year all-cause death rate was correlated with high proportions of beta-blockers (BBs) but not with any other classes of oral cardiovascular medications. Decline in 30-day death was correlated with high proportions of BB prescriptions, and also with high proportions of renin–angiotensin–aldosterone system inhibitors (RAASi), but not with diuretics or digoxin. When considering the effect of combined RAASi and BBs at admission on 30-day and 1-year all-cause death, only the combination of high proportion of RAASi and BBs was associated with the lowest odds of 30-day and 1-year all-cause death. By contrast, no associations were found between proportions of oral cardiovascular medications at admission and 30-day or 1-year all-cause readmissions. Very few studies analysed temporal changes in AHF outcome rates and the present research represents the largest AHF meta-analysis for a total exceeding 15 million AHF patients.7 Using data from the 285 studies, the authors demonstrate a decline in 30-day and 1-year all-cause mortality between 1980–2017 and stable readmission rates over the same time periods. The study also confirms that despite the favourable trends in AHF survival in the past decades, death after AHF remained high, with one quarter of AHF patients dying in the following year. The results are comparable to those reported by the successive analysis performed at 4-year interval by the European Society of Cardiology HF Long-Term Registry,8-10 showing 1-year mortality rates between 22% and 26%, and very high (45%) and unchanged rate of 1-year all-cause hospitalizations. Similar trends were also reported in a French study, showing a substantial decrease in AHF mortality, while rates of patients hospitalized for HF were almost unchanged during the 2002–2012 period.11 Importantly, temporal trends in AHF-related death and readmission in the past four decades did not differ among studied continents and by clinically relevant variables accounting for severity of AHF at admission – age, gender, systolic blood pressure and ejection fraction. Although, subgroup analysis did not include natriuretic peptides, which have been shown to assess the risk better than clinical variables,12 these results denote overall favourable trends for improving short and long-term survival after AHF and stable trends for readmissions, but with mention that it is not entirely clear if both components of readmissions, HF-related and not HF-related, follow the same trend. Even if the authors claimed that the trends in all primary outcomes were linear over the past four decades, suggesting that changes in coding, or management of AHF did not influence the results of the present study, their results cannot be dissociated from the continuous improvement of HF management over time. Although the inference of a 30-day outcome may be due to many confounders, the improvement in short-term mortality over time may be at least partly explained by a better in-hospital management (diagnosis, triage, comprehensive assessment of cardiac substrate and comorbidities, treatment and follow-up plan) of AHF over decades. Another limitation of the present study is related to the definition of AHF, because thresholds for admission could change over time leading to change in diagnostic thresholds, but this is an inherent limit of this type of long-term study. The heterogeneity of the pathophysiology and clinical presentations of this syndrome, as well as the variable relationship between the chronic condition and the episodes of acute decompensation remain major factors hindering a simple and thorough definition of this condition. Furthermore, due to the absence of objective criteria for hospital admission, different organizations of systems of care, variability in out-of-hospital services to manage mild decompensations, hospital bed availability, cultural perception of severity of both physician and patient, the hospitalization threshold has large geographical variations and cannot be excluded enrolment of low-risk patients. Not surprising, the present study shows the disconnection between trend of improvement of early survival and stable trend of early readmissions. Traditionally, early readmissions after hospital discharge indicate incomplete in-hospital resolution of congestion, poor communication of plans at discharge, or socioeconomic factors resulting in inadequate access to care in early follow-up.4, 5 However, beyond the competing risk of mortality, because patients who die early after hospitalization cannot be readmitted, stable rates of 30-day readmission may simply reflect effective efforts to decrease post-discharge mortality.4 Additionally, some readmissions are necessary to avoid deaths, while other readmissions may provide additional opportunities to implement further therapies, improve patient education, or establish clearer follow-up care strategies.4, 5 Another important finding of this research is the association between reduction in mortality and increased use of neurohormonal inhibition, but not of diuretics or digoxin, in the past decades. These data strongly suggest that patients at risk of AHF, i.e. those with chronic HF with reduced ejection fraction and/or other cardiovascular diseases, should be heavily treated in order to prevent death when exposed to future AHF hospitalization. The results are similar to another study, where HF therapies at discharge were associated with a consistent reduction in mortality in patients with recent AHF, regardless of left ventricular ejection fraction or the presence of comorbidities.13 In contrast, oral cardiovascular medical therapies did not influence the likelihood of readmissions, and the trend was roughly stable over time. This is not unexpected, because HF readmissions represent only 30–35% of all-cause hospitalizations and additionally is also possibly that HF medical therapies may destabilize some common comorbidities.6 Finally, the authors should be congratulated for performing this well-conducted meta-analysis aimed at evaluating the temporal trends in mortality and readmissions in AHF.7 These research data serve to update previous epidemiological knowledge and may be the new standard for comparison, for general AHF epidemiology and clinical trial design. The results of the present research mirror conceptual changes happened over time in AHF (Figure 1). Despite the absence of any new therapy, physicians have learned that treating beyond clinical congestion, preventing residual congestion, identifying cardiovascular and non-cardiovascular targets to therapies, pre-discharge initiation/up-titration of HF therapies, and finally early post-discharge visits are associated with improved survival.2 In terms of ‘medical systems’, the complex medical, social, and economic factors contributing to high readmission rates in AHF necessitate an integrated team approach delivered via multidisciplinary care.2, 6 From a research perspective, the overwhelming lack of evidence to support improvement in long-term outcomes when targeting symptoms with in-hospital short-term infusions suggests using another approach, to consider AHF as a marker of high risk and to use admission as a trigger to initiate HF therapies with proven benefit that would permit long-term exposure of AHF patients to therapeutical beneficial effects.14 Also, considering large geographical heterogenicity of AHF patients and the process of ‘globalization of randomized clinical trials’ that raises serious difficulties in study conduction, pre-trial registries15 have the advantage to provide site-specific and region-specific information for improving efficiency and quality of clinical trials and may have the potential to bridge current knowledge gaps in AHF study design and execution. This would also permit to design ‘smaller trials’, enrolling patients with characteristics and outcomes which are equally distributed across sites and countries, since ‘mega-trials’ may be vulnerable to both reductions in placebo event rates and dilutions of the treatment effect, most pronouncedly with disease-specific events such as recurrent HF.16 This type of trial will ensure that targeted pathophysiology is present in the target population and would more favourably match disease pathology with mechanisms of action of the drug.17 Conflict of interest: O.C. reports research grants from Servier, Novartis and Vifor; other from Boehringer Ingelheim. All other authors have nothing to disclose.