Meta‐Analysis of Medication Adherence Interventions Among Heart Failure Patients: The Lumpers Make Their Case

Those who make many species are the ‘splitters,’ and those who make few are the ‘lumpers.’

—Charles Darwin

Dating back to Hippocrates, patients have been nonadherent to medical therapy.1 In the 2000 years since, it remains unclear how much progress has been made. Research shows that medication nonadherence remains highly prevalent and is associated with adverse outcomes and higher costs of care.2, 3 For example, a year after sustaining an acute myocardial infarction, less than 50% of patients are filling prescriptions for statin, beta‐blocker, and angiotensin‐converting‐enzyme inhibitor medications, even when these medications are offered with no copay.4 In 2003, a World Health Organization statement espoused that improved medication adherence “may have a far greater impact on the health of the population than any improvement in specific medical treatments.”5

The World Health Organization cites the 5 main issues with medication adherence as problems with the patient, medical condition, therapy, socioeconomics, and health system.5 Nonadherence is particularly problematic among patients with heart failure, where several different classes of medications have been shown to improve health outcomes, offering patients the blessing of treatment but also the curse of having to adhere to complex regimens to realize the benefit. Research shows that almost half of patients hospitalized with heart failure need to start at least 1 new medication, 24% having indications for at least 2 medications, and 14% for ≥3 medications in order to comply with current guidelines.6 This need for a cadre of new medications, coupled with the overwhelming nature of the diagnosis and associated debilitating symptoms, is likely to obfuscate patients rather than to enlighten them on how best to care for themselves. Not surprisingly, patients often do not have long‐term adherence to heart failure medications, as evidenced by low‐fill rates in the ambulatory setting.7, 8 As former Surgeon General Dr C. Everett Koop famously said, “Drugs don't work in patients who don't take them.”2

Over the years, many trials have tested various medication adherence interventions among heart failure patients. In this issue of Journal of the American Heart Association, Ruppar and colleagues provide a 30 000‐foot view of all medication adherence interventions among patients with heart failure.9 They performed a comprehensive systematic review between 1996 and 2013, then conducted a meta‐analysis to examine the effect of medication adherence interventions on mortality and readmissions. Their search resulted in 57 eligible trials where interventions included medication education, disease education, improved integration of care, self‐management teaching, self‐monitoring, and other strategic combinations. Sufficient data were available in 48 studies to calculate mortality outcome effect sizes; sufficient data were available in 32 studies to calculate hospital readmission effect sizes. Using random effects model meta‐analysis methods, medication adherence interventions among patients with heart failure were found to significantly reduce mortality (relative risk 0.89, 95% CI: 0.81–0.99) and to decrease odds for hospital readmission (odds ratio 0.79, 95% CI: 0.71–0.89). This comprehensive and contemporary review of the medication adherence intervention literature among patients with heart failure is a novel integrative summary to an age‐old question, and the findings are encouraging.

However, it should be noted that other researchers have performed similar systematic reviews of medication adherence interventions and have shied away from meta‐analysis. For example, a recent review by Molloy et al examining all medication adherence interventions in heart failure stopped after systematic review due to a “high degree of heterogeneity” among studies.10 Trials in this area have varied widely in their strategic interventions, baseline patient population, and results. The difference between the meta‐analysis by Ruppar and prior systematic reviews comes down to a contrast between lumpers and splitters. Is the body of adherence research an Impressionist painting bringing visual clarity about our reality or a collection of small brush strokes cohabitating to form an abstract fiction? Studies that are evaluated in a meta‐analysis will inevitably differ, and the difficulty lies in deciding how similar they need to be for valid combination. Sometimes it is nice to have a parsimonious answer, but only if that answer is true. Moreover, publication bias leads to studies with a positive treatment effect being more likely to be pushed out into the medical literature.11 Given the wide range of interventions compiled by Ruppar, a healthy degree of skepticism is crucial.

The heterogeneity of the heart failure syndrome itself further complicates this combination approach. Some of the studies included in this meta‐analysis included all heart failure patients (ie, those with reduced as well as preserved ejection fraction). Given that there are no guideline‐recommended medical therapies for heart failure with preserved ejection other than diuretics,12 it makes the reader wonder how these studies played into the mix. Perhaps the greatest challenge with this study is how to interpret some if its more puzzling findings. The authors found that readmission outcomes were better when adherence interventions did not have patients self‐monitoring their medications and that interventions designed to train healthcare providers to have better skills for addressing medication adherence resulted in greater mortality risk for patients. These findings go against common beliefs that patients should be closely managing their own medications and that healthcare providers should be addressing medication adherence regularly. Currently, patients already have poor understanding of their medications and low rates of self‐monitoring. More importantly, the onus cannot just be on the patient; clinicians should also address these issues during patient encounters but often do a poor job. Recent survey data suggest that over 60% of patients in cardiology practices do not remember discussing medication adherence with their physicians.13 Fortunately, the authors of this meta‐analysis acknowledge that their findings should be interpreted with caution, but conclusions such as these serve as examples of inherent problems with meta‐analyses.

Whether one chooses to have the picture painted for them or to have a narrative review alone of medication adherence interventions in heart failure, the fact remains that there are now several medications that have proven efficacy at reducing morbidity and mortality among patients with heart failure and reduced ejection fraction and that these benefits cannot be realized without better adherence and higher persistence in clinical practice. One of the common features of successful interventions has been regular follow‐up with healthcare providers, often through focused heart failure disease management clinics.3 Given the burden of medication initiation with new and worsening heart failure and issues with polypharmacy, transitions from hospital‐to‐home after a hospital admission for heart failure decompensation are particularly vulnerable periods for patients. Early postdischarge follow‐up may help improve the chances of better medication adherence and long‐term outcomes.14 With the complexity of the problem, other adherence interventions will need to be personalized and targeted for the patient, provider, and the health system in which they practice.

At a glance, lumping interventions to improve heart failure medication adherence suggests a positive vision for the future. However, these summary findings do not help the reader strategize which components of medication adherence interventions are most effective in specific heart failure populations. The fine brush strokes of tailoring these interventions to the population at hand are currently left in the hands of clinicians and healthcare systems.

Disclosures

Dr Khazanie has no disclosures. Dr Allen discloses institutional research grant support from National Institutes of Health (K23 HL105896) and Patient‐Centered Outcomes Research Institute (PCORI‐1310‐06998); and consulting relationships with St Jude, Janssen, and Novartis.

J Am Heart Assoc. 2016;5:e003827 doi: 10.1161/JAHA.116.003827