Discussion

Despite therapy uptake being increasingly recognized as an unmet need in HFrEF, comprehensive assessments spanning from physician prescriptions to patient adherence in the era of quadruple pharmacotherapy have been limited. In this contemporary nationwide cohort of >35 000 patients with HFrEF, we observed overall high GDMT uptake in terms of prescriptions and patient adherence. The proportion of patients receiving quadruple therapy increased rapidly since 2021 (time of ESC HF Guidelines), reaching ∼60% in 2023. The proportion of patients with good adherence was ∼90% or higher for all drug classes and ∼85% for quadruple therapy overall. As of 2023, use of SGLT2i overtook MRA, which remains the least prescribed (73%), least adhered to (90%), and least persisted (77% at 1 year) among the GDMTs. In an explorative analysis, patient adherence and 1-year persistence were associated with lower risk for HHF or cardiovascular death after adjustments for measured confounders (Structured Graphical Abstract). Our findings exemplify that high GDMT penetrance is achievable in an HFrEF cohort predominantly reflecting specialized cardiology care in a universal healthcare system, but that even in this setting there persist major areas of improvement, including suboptimal attainment of target doses and frequent discontinuation of prescribed therapies. Addressing patient adherence and persistence to GDMT might further improve outcomes.

Underuse of HFrEF GDMT has been shown to be a major challenge in the era of triple therapy across all studied healthcare settings,5 but with limited data available since the addition of SGLT2i as a fourth pillar of HFrEF pharmacotherapy in 2021.1,18 Previous data from SwedeHF showed that virtually no improvements were made in the implementation of HFrEF therapies between 2003 and 2012 and that this lack of progress was coupled with stable mortality rates.19 Since then, as our data show, all evidence-based pharmacotherapies for HFrEF have seen increasing implementation, although target doses of neurohormonal therapies were still achieved in a minority of patients. This progress mirrors recently reported prognostic improvements for patients with HFrEF in Sweden.16

What are the potential reasons for the observed positive shift in HFrEF therapy implementation in Sweden? First, it is plausible that increased attention to extensive underuse of therapies has contributed to improved awareness among healthcare providers. Second, the widespread integration of the HF-specific national registry (SwedeHF) into cardiology clinics has likely heightened attention to treatment gaps at both national and local levels, aligning with the principle that measurement drives improvement. Third, the increased adoption of HF nurse-led clinics in Sweden enables tighter follow-up during up-titration or vulnerable phases that carry risk for treatment de-escalation. Fourth, recent years’ novel therapies might act as ‘enablers’ to other HFrEF therapies. For example, in PARADIGM-HF and EMPEROR-Reduced, respectively, ARNi and SGLT2i were linked with greater use of MRA.9,10 Indeed, in our data, the implementation of MRA improved year-by-year throughout the 2016–23 period, but with a steepening slope following the adoption of SGLT2i for HFrEF therapy in 2021. Fifth, past emphasis on sequential addition and uptitration of one GDMT followed by the next may have caused delays, whereas the 2021 Guideline recommendation of simultaneous or near simultaneous introduction of all four may have had considerable effects on all aspects of GDMT uptake. Furthermore, following STRONG-HF,20 the 2023 focused update of the ESC HF guidelines strengthened the emphasis on rapid treatment intensification pre- and post-discharge among patients hospitalized for HF.21 Finally, our cohort consisted of patients with HFrEF enrolled in a quality registry predominantly reflecting secondary/tertiary care, which might also partly explain the observation of better GDMT implementation than in studies using electronic health records from broader and less selected populations, but with less data granularity for strictly identifying indications (e.g. EF) for drugs.6 While our study is performed in a setting where GDMT implementation is more likely achievable, it should be underscored that a large proportion of patients (∼40%) did not receive quadruple therapy at the end of the study period, and receiving four GDMT components at target doses remained rare (∼6%). Although some patients might have characteristics that truly prevent the achievement of full GDMT, our findings further highlight that the implementation of guidelines in the real world is a complex process, with barriers at several levels, i.e. physician-related, patient-related, therapy-related, and healthcare system-related, which are difficult to address even in well-structured universal healthcare system.

Patient adherence and persistence to therapy are well-known challenges in chronic cardiovascular conditions and core determinants of real-world treatment uptake.3,22 In a recent cardiologist survey from the Heart Failure Association of the ESC, 42% of participants referred to patient adherence as a major clinical barrier to implementation of GDMT in HFrEF.23 However, comprehensive assessments of patient adherence to HFrEF therapy in the era of quadruple therapy have been lacking. Our data suggest that patient adherence to pharmacotherapy in HFrEF (90%–95% of patients having good adherence across drug classes) is overall higher than in other chronic cardiovascular settings such as hypertension and hyperlipidaemia, possibly due to HF being more likely symptomatic and recognized by the patient as a serious condition. Our adherence estimates are comparable to reports from Norway (proportion with good adherence: beta-blockers 83%, RASi/ARNi 84%–91%, MRA 61%),24 a universal healthcare system with similarities to Sweden, but appear considerably higher than in the USA (beta-blocker 61%, RASi/ARNi 35%–59%, MRA 36%).25,26 We believe that two main attributes of the our study setting might contribute to higher adherence than might be observed in other contexts. First, high co-payments have been associated with lower adherence in HF,27 and in Sweden, medication costs are covered once an annual out-of-pocket spend threshold (in 2023 ∼230€) has been reached. Second, the SwedeHF setting has relatively high penetrance of follow-up in HF nurse-led clinics, enabling more frequent out-patient contacts which has also been linked with better patient adherence in HF.28 Despite the relatively high adherence observed in our study, there is still room for improvement particularly for the long-term persistence to quadruple therapy which was only ∼67% at 1 year, mainly limited by lower persistence to MRA (77%).

In an explorative outcome analysis, patient adherence to all GDMTs and 1-year persistence to all GDMTs except for SGLT2i were associated with lower risk for HHF/cardiovascular death before and after adjustments for measured confounders. For SGLT2i, although the point estimate was consistent with what was observed for the other drugs, the association between 1-year persistence and HHF/cardiovascular death was not statistically significant, likely due to the limited sample size. It is important to note that the observational nature of this analysis precludes causal conclusions; even after extensive adjustments, unmeasured confounders might partly explain our findings. In the CHARM trial, poor adherence even to placebo was associated with adverse prognosis, highlighting that prognostic associations of poor therapy adherence might not only be explained by diluted therapeutic benefits but also by several potential confounders, such as overall worse health behaviour, disease severity, and comorbid condition predisposing to tolerability issues.29

A first step in addressing patient adherence is to recognize when there is risk for non-adherence. We identified three key clusters of patient characteristics that appeared broadly and independently linked with greater risk for non-adherence to HFrEF GDMT: (i) characteristics linked with tolerability issues, e.g. hypotension (RASi/ARNi), bradycardia (beta-blockers), and hyperkalaemia (MRA); (ii) factors linked with overall poor health behaviour (e.g. smoking, alcohol abuse); and (iii) poor socioeconomic status. Improved digital health infrastructures and the integration of HF registries in the clinical workflow might facilitate identification of—and possibly screening for—gaps in treatment uptake, from implementation to patient adherence and persistence. Although meta-analyses suggest that medication adherence interventions in HF improve clinical outcomes,30 HF-specific evidence on how to best improve medication adherence is limited. As highlighted in a scientific statement from the Heart Failure Association of the ESC,3 potential strategies to improve patient adherence might include nudges such as text message reminders,31,32 patient education, integration of HF nurses and pharmacists in patient follow-up,33 and medication regimen rationalization. Single pill combinations have been shown to improve adherence as well as outcomes in other cardiovascular diseases and are being studied in HFrEF (NCT04633005, NCT06029712).34

Strengths and limitations

The primary strength of this study was the use of a large, contemporary, and well-characterized HF cohort, with virtually complete access to drug prescriptions, and virtually complete follow-up for cause-specific outcomes. The ability to ascertain prescribed doses from free text dose instructions represents a strength in relation to several prior studies of patient adherence, which have relied on the time interval between initial pharmacy refills to infer the intended dose. There are also key limitations that need to be acknowledged when interpreting our results. First, the Swedish healthcare system, and the SwedeHF setting in particular, has several attributes that might positively affect GDMT uptake, making generalizability to other healthcare systems uncertain. Moreover, only a subset of the Swedish HF population is captured in SwedeHF, where patients are predominantly seen in specialty care, are less likely female, receive better treatment, and have better prognosis.35 Therefore, it is likely that the treatment uptake observed in this study approaches a ‘best case scenario’, reflecting the secondary/tertiary care environment of a country with strategies to address barriers to GDMT implementation are in place. Our results may not be generalizable to other healthcare systems and less specialized or primary care settings, where lower uptake both in terms of implementation and patient adherence might be expected. Second, although standard methods were used to estimate adherence via pharmacy refills, these methods involve a multitude of investigator degrees of freedom and have inherent limitations, including that retrieved medications are not necessarily consumed and inability to fully account for doses consumed during hospital stays. Third, we lacked data on clinical parameters (potassium, blood pressure, heart rate) during follow-up and could not distinguish between non-adherence/non-persistence caused by patient preferences, non-compliance, and clinically motivated reasons such as side effects or palliation. Fourth, given the aim of providing a general picture of treatment uptake, this study included prevalent users, who are expected to have higher adherence/persistence than first-time users since early discontinuations due to uncovered side effects are less likely, although a sensitivity analysis in de novo users showed overall consistent results. Fifth, implementation was defined according to prescriptions dispensed in pharmacies, meaning that those who never dispensed their prescription were considered as non-users. This might contribute to a slight underestimation of therapy implementation and a slight overestimation of patient adherence. Finally, the observational design precludes causal conclusions, and our findings should therefore be interpreted as descriptive.