Heart Failure Disease Management Interventions: Time for a Reappraisal

It is over two decades since the publication of the first heart failure (HF) disease management trial results. Consequently, it is timely to consider progress and lessons for these important health services. The early programme trials have undoubtedly been influential. For example, a trial of a nurse-directed multidisciplinary intervention for elderly patients with congestive heart failure (n = 282) found reductions in readmissions for HF and medical costs with improvements in quality of life at 90 days.1 A trial of a multidisciplinary, home-based intervention of a single home visit by a cardiac nurse one to 2 weeks after discharge for patients with chronic congestive HF (n = 200) found fewer unplanned readmissions plus out-of-hospital deaths and associated days in hospital and reduced hospital-based costs at 6 months and improvements in quality of life at 3 months.2 Another trial of a specialist nurse intervention for patients with HF due to left ventricular systolic dysfunction (n = 165) found fewer deaths or readmissions with HF, all-cause readmissions and admissions and days in hospital for HF.3 Cumulatively, these three trials alone have received around 4000 citations and historically formed the impetus for over 60 subsequent trials, over a dozen meta-analyses4 and a movement toward multidisciplinary disease management programmes for other chronic conditions. Yet, the sizeable positive results of these early trials of extremely brief interventions have struggled to be replicated consistently since, even with considerably larger and more complex interventions5 and meta-analyses.4 This influence derived from an overwhelming focus on ‘headline’ positive effects – to the neglect of key design weaknesses, including atypically healthy populations, vaguely described interventions, and non-described comparison group found in these and other trials and meta-analyses.4 Proponents for HF disease management interventions, even currently, retrospectively gloss over the possibilities of bias created by these design flaws,4, 5 continuing to label HF disease management interventions as ‘proven’.6 The positive results of these early trials are curious. Indeed, the early trials1-3 could be expected to have worse findings than subsequent trials, which could draw upon and harness insights from the previous ones. The difficult challenges of daily HF disease management are now much better understood: over 100 subsequent qualitative research studies in the years since have identified both the immense daily complexity of maintaining effective disease management and the consistently very poor knowledge, skills and confidence that the vast majority of HF patients and families possess to do so both before and after specialist consultations.7 Yet, the interventions of these early trials were briefer, focused only on patients, and more rudimentary than subsequent trials.4 Over the past decade, we8, 9 and others10-13 have suggested that the risk of publication bias remains high in this field because researchers have considerable vested career interests in publishing and championing interventions with positive findings. Current labelling of the interventions as being ‘proven’ also risks confirmatory bias: the tendency to search for, interpret, favour, and recall information to affirm one's prior beliefs or hypotheses. Indeed, larger studies of HF disease management interventions performed by independent investigators with no career or personal interest in results have had inconsistent and negative results.5, 14 Casting these past interventions as ‘proven’, proponents now call for a nuanced approach to HF management.6 Indeed, we identified these weaknesses over a decade ago, including a plea that appeared on the cover of the Lancet including the recognition that ‘Management programmes for HF have been championed on the basis of limited high-quality evidence’.15 In terms of approaches, now, as before, we call for more independent and nuanced, high quality evidence4, 15-17 into understanding the outcomes of HF disease management interventions. Learning from negative trial results is useful and important but has been uncommon.18 As complex interventions, concerted efforts are needed to both describe and understand the nature and effects of programmes and heterogeneity4 and how outcomes are influenced by both context19 and intervention mechanisms.7, 20, 21 Simplistic questions of ‘which interventions work?’22 – often associated with catchy study acronyms - should be replaced by research which unpacks the complexities of how factors associated with programmes, people and places interact to influence outcomes; in short to address ‘what works for whom, when and why?’.15-17, 19 This more nuanced approach to designing, implementing and evaluating interventions should utilize mixed quantitative and qualitative methods.4, 15-21 Instead of dismissing inconsistencies in programmes and simply attributing these variations to intervention fidelity, research into programmes should investigate variations in outcomes by exploring programme mechanisms and contexts.19 For example, although research into intervention mechanisms is large and growing in research volume,23 review approaches,24 frameworks25 and methods,26 there is an absence of detailed primary studies in heart failure management. Mechanisms as the ‘black box’ between an intervention's inputs and outputs27 are causal, often unobservable and may be associated with benefits or harms. Mechanisms influence outcomes through the interaction of various intervention parts.28 Reflecting the PICO (Population, Intervention, Comparison, and Outcome) approach, in future trials and meta-analyses there is an urgent need for better methods. Populations should be better described and targeted. Interventions need to be more systematically theorized, comprehensively detailed and driven by theories of change. Comparisons of typical HF care should be well described. Outcomes (including patient experiences) should be measured and explained. Adopting such approaches will identify hidden but potentially influential factors such as the crucial characteristics of a programme,29 and determinants of self-care30, 31. Of course, a one size fits all approach to HF management is untenable and it is incumbent on clinicians to not only apply these interventions to the right people at the right time but also to consider issues such as patient (and partner, family, caregiver) choice and preferences. Most HF interventions have been designed, driven and evaluated by clinicians and researchers with limited patient input. Given the patient is the dominant expert in their condition and its management, future interventions would benefit from systematic integration of patient and caregiver values, preferences and perceived needs into intervention content and design. Outcomes such as health-related quality of life, satisfaction and caregiver burden may well take precedence over survival and hospital resources should be assessed accordingly. The boundaries between science and advocacy have become blurred around HF disease management programmes. Too often those advocating disease management approaches most strongly have either ignored negative findings or dismissed these outright as resulting from biases that serve to ‘mask’ otherwise effective interventions.6 This advocacy, particularly when combined with overly simple research questions and simplistic methods, risks good science. While advocates of programmes remain eager to cast current evidence retrospectively as ‘proof’, this continues to risk confirmatory bias: a tendency which influences professionals in situations in which they seek to appear more knowledgeable and credible than justified in ambiguous situations.32 In addition to bias, as with other fields, blurring the lines between advocacy and science increases the risk of unethical attempts to falsely ‘create’ research success and impact. From the use of questionable research practices (such as merging outcomes or statistical manipulation to produce more impactful results, to the outright fabrication of data and study results), there remains an ongoing need for independent and more nuanced evaluations of HF disease management programmes free from actual or perceived conflicts of interest. Conflict of interest: none declared.