Introduction

The incidence, prevalence, and adverse outcomes of heart failure continue to gain ominous impact [1, 2]. Indeed, our successes at managing acute cardiovascular conditions has translated into a booming prevalence of chronic heart failure. Yet, acute heart failure poses even greater management challenges than chronic heart failure, as optimal decision-making must be accomplished in a limited time frame [3]. Specifically, in keeping with current definitions, acute heart failure is a recent onset clinical syndrome characterized by typical symptoms that may be accompanied by signs caused by a structural and/or functional cardiac abnormality, resulting in a reduced cardiac output and/or elevated intracardiac pressures at rest or during stress [1].

Several interventions have been recommended or considered for acute heart failure, yet uncertainty persists on many of them [1, 3]. This has to do with difficulty in enrolling patients in randomized trials, but also with the fragmentation of interventions and clinical scenarios. Network meta-analysis is a methodological tool suitable for synthesizing complex evidence stemming from different randomized trials comparing alternative interventions for a common condition. Accordingly, network meta-analyses can be particularly useful to summarize randomized trials on acute heart failure care.

We thus aimed to provide a concise yet comprehensive overview of recent network meta-analyses on acute heart failure care, to inform clinical practitioners and guide further research efforts [4].

Methods

PubMed was searched on December 15, 2017, for network meta-analyses and mixed treatment comparisons published in scholarly peer-reviewed journals and focusing on the management of acute heart failure, cardiac shock, or cardiogenic shock with the following string: ((acute AND heart AND failure) OR (cardiogenic OR cardiac) AND shock)) AND ((network AND (meta-analysis* OR meta-analysis)) OR ((mixed AND treatment AND comparison*)). We adopted a broad selection approach, thus including reviews on acute heart failure as well as conditions leading, even indirectly, to cardiovascular insufficiency.

We abstracted key features of included reviews, as well as summarizing key findings and bibliometric indexes. Quality of reporting was appraised in keeping with the PRISMA statement on network meta-analysis [5•]. Internal validity was appraised in keeping with Biondi-Zoccai and Zarin et al. [6••, 7•], focusing on the following domains: search, selection, abstraction, appraisal, effect estimate, inconsistency estimate, and confounding estimate. Each domain was scored as validly, invalidly, or unclearly addressed. Results were tabulated without additionally summarizing efforts.

All reviewing activities were conducted by one expert reviewer (GBZ), and then independently checked by another reviewer (LG), with discrepancies resolved after consensus.

Results

We identified an initial set of 43 citations, with 5 being finally excluded because not using a network approach, 4 being a qualitative review, 8 being an observational study, 3 being a randomized trial report, 16 being an animal experimental study, and 2 focusing on cancer. Eventually, five network meta-analyses were shortlisted (Table 1), including a total of 101 randomized trials and 19,085 patients [8,9,10,11,12].

Table 1 Recent network meta-analyses on acute heart failure
Full size table

Specifically, Lee et al. included 13 randomized trials with 2843 patients on different means for mechanical hemodynamic support during high-risk percutaneous coronary intervention with or without cardiogenic shock [11]. Khan and colleagues focused on 11 randomized trials and 5200 subjects receiving antiarrhythmics for shock-refractory ventricular arrhythmias [12] Belletti et al. [8], Gibbison and colleagues [9], and Zhou et al. [12] all focused on pharmacologic interventions for severe sepsis or septic shock, including, respectively, 33, 23, and 21 randomized trials, with 3470, 3753, and 3819 patients.

Quality of reporting was moderately high in the five reviews (Table 2), with major weaknesses in protocol registration, inconsistency appraisal, and confounding appraisal. Internal validity was moderately high as well (Table 3), despite some lapses on inconsistency assessment, and common deficiencies in confounding appraisal. In particular, only Belletti et al. appropriately checked for small study effects or publication bias as recommended with comparison-adjusted funnel plots, regression tests, or network meta-regression [6••,8].

Table 2 Quality of reporting of recent network meta-analyses on acute heart failure
Full size table
Table 3 Internal validity of recent network meta-analyses on acute heart failure
Full size table

The main findings of shortlisted reviews were in favor of lidocaine for shock-refractory ventricular arrhythmias and several vasopressors for severe sepsis and septic shock (Table 4). Conversely, no evidence of benefit was found for mechanical hemodynamic support in high-risk percutaneous coronary intervention or corticosteroids in septic shock. Bibliometric indexes and usage metrics, albeit confounded by journal and publication date, suggested that all network meta-analyses, except for Khan et al., were extensively quoted and disseminated (Table 5) [10].

Table 4 Main findings of recent network meta-analyses on acute heart failure
Full size table
Table 5 Bibliometric indexes and social usage of recent network meta-analyses on acute heart failure
Full size table

Discussion

This focused overview of recent network meta-analyses on acute heart failure care highlights the potential benefits as well as drawbacks of this novel tool for complex evidence synthesis [6••]. Specifically, we were able to shortlist five recent network meta-analyses focusing on pharmacological and mechanical interventions in patients with acute heart failure [8,9,10,11,12]. These works appeared of moderately high quality in terms of reporting and internal validity. Their main findings highlighted the clinical usefulness of lidocaine and several vasopressors in, respectively, shock-refractory ventricular arrhythmias and severe sepsis or septic shock [8, 10, 12]. Conversely, they suggested that corticosteroids are not beneficial in septic shock, nor are mechanical hemodynamic support devices in high-risk percutaneous coronary intervention [9, 11].

Despite these apparent fruitful efforts, the actual elephant in the room is the paucity of network meta-analysis on acute heart failure care published in recent years, as well as the limited number and scope of randomized trials among included reviews. Indeed, several of the most important clinical decisions were not the focus of reviewing efforts (e.g., diagnostic approaches, diuretics, antithrombotics). In addition, most evidence networks were largely star-shaped, testifying the lack of comprehensive comparative data for any given intervention-condition combination [6••]. Even the sobering results in favor of lidocaine reported by Khan et al. [12] should be put into the broader perspective stemming from the non-significant findings of the large randomized trial led by Kudenchuk et al., despite its inclusion in the above network meta-analysis [13].

Our operative choice of including in our overview of reviews also network meta-analyses on severe sepsis or septic shock merits further elaboration. Specifically, we chose to include these reviews [8, 9, 12] for a number of reasons. First, pathophysiologically, most cases of severe sepsis/septic shock are associated with at least mild diastolic and systolic dysfunction, thus making them pertinent for any review on acute heart failure [14]. Second, low blood pressure is common in acute heart failure and the optimal management of low blood pressure in severe sepsis/septic shock may indeed inform on the optimal management of low blood pressure in acute heart failure. Third, infection can often complicate acute heart failure, creating a clinical conundrum consisting of sepsis and heart failure, which may benefit from details on the optimal management of severe sepsis/septic shock. Yet, the reader should be aware that none of the included reviews provided details on cardiac systolic or diastolic dysfunction, thus clearly limited the external validity of these reviews.

From a more technical viewpoint, it remains disappointing that after several years of outstanding recommendations, no review was registered pre hoc (e.g., in the University of York Centre for Reviews and Dissemination PROSPERO system) [15]. In addition, inconsistency appraisal was not universally and explicitly performed. Similarly important drawbacks were apparent for small study effects, which were either not appraised at all or inappropriately checked in most reviews [6••, 16••]. Finally, network meta-regression approaches were never used, at odds with their potential for sensitivity analyses and hypothesis generation [17].

Implications for Research and Practice

Despite the above reporting and methodological limitations, it is interesting to notice that almost all reviews were already highly successful either in terms of peer usage or scholarly citation, confirming the appeal of network meta-analyses for practitioners as well as researchers [18]. Yet, such success should not distract our shared efforts to improve the methods and scope of network meta-analyses in acute heart failure care, as ample room for improvement persists. More poignantly, we can infer from the present overview that several vasopressors may be beneficial for severe sepsis and septic shock, in particular norepinephrine and levosimendan, that lidocaine is useful for shock-refractory ventricular arrhythmias, and that no mechanical hemodynamic support device is currently beneficial for high-risk percutaneous coronary intervention, irrespective of the presence of cardiogenic shock.

Conclusion and Future Directions

Notwithstanding that recent network meta-analyses on acute heart failure may begin to offer guidance for comparative effectiveness and improved clinical decision-making, further synthesizing efforts are needed to provide a more comprehensive and updated synthesis of the multitude of clinical alternatives for physicians caring for patients with acute heart failure.