Unpicking medication discontinuation and non-adherence in heart failure

This invited commentary refers to ‘Association between socioeconomic position and discontinuation of pharmacotherapy for heart failure after cardiac resynchronization therapy’, by M.H.J.P. Frausing et al., https://doi.org/10.1093/eurjcn/zvae083. Over the past decade, there have been multiple landmark trials that have advanced both pharmacological and device therapy treatments for those living with heart failure with reduced ejection fraction (HFrEF).1,2 Guidelines were quick to adapt to these developments, advocating the use of four main drug classes, commonly referred to as the ‘four foundational pillars’,3 or guideline-directed medical therapy (GDMT), namely beta-blockers (BBs), mineralocorticoid receptor antagonists (MRAs), angiotensin receptor neprilysin inhibitors, and sodium–glucose cotransporter-2 inhibitors. Collectively, there is robust randomized controlled trial (RCT) evidence for their effectiveness in reducing risk of death and hospitalization in HFrEF.4 Further to prolonging survival, they individually returned modest improvements in health-related quality of life.5 Having established efficacy, research focus has shifted to explore the obstacles to optimal GDMT sequencing, implementation, and up-titration in real-world clinical settings.6,7 Although current guidance may appear easy to achieve, most of the participants included in these RCTs are not reflective of the typical heart failure patients seen in clinics. International registry data characterizing real-life clinical practice,8 suggest several other factors influence GDMT implementation, for example clinical inertia, medical (tolerance), clinician (awareness), patient (preference), and system (costs) barriers.9 Initiating and up-titrating pharmacotherapy is just one part of the optimal medication therapy jigsaw puzzle. Once medication is prescribed, an estimated 30–50% of people do not adhere to their regimen.10 Therefore, it is equally important that healthcare providers ensure adequate and long-term adherence. Within heart failure, poor adherence has been linked to worse outcomes like high rates of heart failure-related hospitalizations and premature mortality.11 There are myriad reasons as to why people with heart failure might have sub-optimal medication concordance.12 Unlu et al.13 have highlighted the problem of polypharmacy in heart failure, particularly among the elderly, which is associated with adverse outcomes. Page et al.14 point to fragmented systems of care, which conspire to increase the risk of non-concordance and therefore heart failure exacerbations. Noureldin et al.10 have implicated health literacy as an important factor in a patients’ ability to follow complex medication regimens. Qualitative research describes the negative impacts of some medications, notably diuretic therapy, which can lead some patients to question their value.15 In this issue of the European Journal of Cardiovascular Nursing, Frausing et al.16 report the results of a retrospective, single-centre linked registry study conducted in a cohort of Danish patients with HFrEF (n = 1880) who received cardiac resynchronization therapy (CRT). The aim was to determine (i) rates of heart failure medication discontinuation; (ii) socioeconomic predictors of non-adherence; and (iii) correlates with multimorbidity. This research, which showcases Denmark’s enviable demographic and clinical data registries (Civil Registration System, Danish Pacemaker and ICD Register, Danish National Patient Registry, National Prescription Registry, Statistics Denmark), has been carefully conducted, exquisitely described, and beautifully presented. The authors used multivariable Cox proportional hazard regression models (minimally and fully adjusted) to calculate hazard ratios and 95% confidence intervals for the risk of drug discontinuation/non-adherence across a range of well-defined variables including living status, educational attainment, income strata, and multimorbidity. The fully adjusted model accounted for important covariates such as age, sex, and the presence of coronary artery disease, atrial fibrillation, and diabetes. The authors reported that the cumulative incidence of discontinuation of heart failure medications varied according to the type of medication prescribed. By 10 years, MRAs were the most commonly discontinued therapy (24%, 95% CI: 20–27%), followed by angiotensin-converting enzyme (ACE) inhibitors (10%, 95% CI: 9–12%), then BBs at 6% (95% CI: 5–8%). Discontinuation was not affected by educational attainment nor income tertiles; however, living alone was associated with discontinuation of BB and MRAs, and multimorbidity associated with discontinuation of ACE inhibitors and MRAs. When the authors examined adherence, they found that almost all patients (95%) obtained one or more of their prescriptions at 180 days after CRT implant and most (85–95%) remained adherent throughout the median follow-up period (4.7 years, interquartile range 2.6–7.6). No substantial differences in adherence by drug type were detected for the initial period; however over time, adherence to MRA therapy was lower than for the other drugs analysed. Through these findings, the authors concluded that in this population, (i) overall drug discontinuation rates were low; (ii) overall drug adherence rates were high; (iii) an established correlate (socioeconomic status) was not associated with drug discontinuation; (iv) living alone was associated with discontinuation of BB; and (v) multimorbidity was associated with discontinuation of ACE inhibitors and MRAs. The authors rightly speculate that the low discontinuation rates, high adherence rates, and lack of association with socioeconomic status might reflect the organization of care in Denmark (universally tax-financed free healthcare), and the implementation of clinical guidelines that mandate up-titration of optimal medical therapy to maximally tolerated doses, prior to referral for CRT implantation.17 However, they more contentiously conclude that ‘lack of social support at home and the presence of many coexisting conditions significantly influence drug persistence in patients with CRTs’. Statisticians are regularly at pains to point out that correlation is not causation,18,19 and it is perhaps here where Frausing et al.16 could have continued with the more cautious nomenclature they had previously adopted. Indeed, the authors go on to partially qualify this statement by presenting a myriad of reasons as to why patients may discontinue some therapies, and they suggest strategies that might augment the interrogation of statistically derived associations. For example, they propose embedding patient-reported outcome measures within clinical practice and linking these to registries.16 This methodology, they propose, may help researchers move beyond statistical inferences, accelerating our understanding. That discontinuation rates are low and adherence rates are high in patients receiving a CRT in Denmark is something that should be lauded. One might indirectly extrapolate from this that Danish cardiac centres are high performing and operate guideline-adherent systems of care. Further, we might postulate that the presence of robust registries and healthcare providers’ engagement with these have enabled continuous monitoring of performance and thus quality improvement. In terms of the association between living alone and multimorbidity with medication discontinuation, these findings need further investigation. While they are very plausible, and Frausing et al. summarize the evidence that justifies their potential as an important mediator of adherence, further research is needed to confirm their validity. This research is a good example of how linked registry-based data can be leveraged to identify possible targets for treatment optimization in heart failure. Arguably, the most salient finding is that the systems of cardiac care that operate in Denmark appear to be effective in ensuring that heart failure patients post CRT implantation are concordant with medication prescriptions in both the short and long term. Learning how Denmark has achieved this will help other centres with sub-optimal drug initiation and adherence develop their clinical services to ensure patients receive optimal medical therapy, regardless of where they reside. Faye Forsyth (Conceptualization [lead], Writing—original draft [lead], Writing—review & editing [equal]) and Sara Tavares (Conceptualization [supporting], Writing—original draft [supporting], Writing—review & editing: [equal]) No funding was obtained for this research. Not applicable. No new data were used in this editorial.