Abstract
Aims
Whether medication adherence to angiotensin receptor‐neprilysin inhibitor (ARNI) in real‐world practice is associated with the reduced risk of all‐cause mortality or hospitalization relative to that with traditional renin–angiotensin system (RAS) blockade remains unclear. This study investigated the influence of medication adherence of ARNI and traditional RAS blockade in heart failure with reduced ejection fraction (HFrEF).
Method
We conducted a nationwide longitudinal cohort study with patients with HFrEF using data from the Korean National Health Insurance Service data (2017–2021) covering the entire population. A total of 13 483 patients with HFrEF who received ARNI were matched 1:1 with 13 483 patients who received traditional RAS blockade using propensity score matching. The primary outcome was a composite of all‐cause mortality or any hospitalization within one year. Medication adherence was assessed by calculating the proportion of days covered (PDC) relative to total medication prescribed. ARNI and traditional RAS blockade adherence rates were directly compared to analyse their respective associations with the primary outcome.
Results
Patients in the ARNI group had a lower rate of the primary outcome than those in the traditional RAS blockade group [hazard ratio (HR) 0.78; 95% confidence interval (CI) 0.75–0.81; P < 0.001]. Mean PDC values spanning 1 year were 92.6 ± 14.5% and 90.9 ± 17.7% in the ARNI and RAS blockade groups, respectively (P < 0.001). Among patients with PDC ≥ 80%, the risk of primary outcome was significantly lower in the ARNI group than in the RAS blockade group (HR 0.75; 95% CI 0.72–0.78; P < 0.001) while a risk reduction with ARNI was not observed among patients with PDC < 80% (HR 0.95; 95% CI 0.85–1.05; P = 0.313). The beneficial effect was more pronounced among patients with PDC ≥ 80% than that among patients with PDC < 80% (P for interaction <0.001).
Conclusions
In a real‐world cohort with HFrEF, ARNI was superior to traditional RAS blockade in reducing the risk of all‐cause mortality and hospitalization. The benefit of ARNI was pronounced among patients with high medication adherence but not among those with low medication adherence, highlighting the importance of adherence to ARNI treatment for HFrEF.
Trial Registration
PARADE‐HF ClinicalTrials.gov number, NCT05329727.
Keywords: angiotensin receptor‐neprilysin inhibitor, medication adherence, renin–angiotensin system blockade, real‐world evidence
Real‐world comparison of medication adherence between ARNI and RAS blockade in patients with HFrEF.
Introduction
The prevalence of heart failure (HF) is increasing with the ageing population worldwide. 1 HF is a significant burden on the global healthcare system due to its high morbidity and mortality. 2 Four pillar drugs, including angiotensin receptor‐neprilysin inhibitor (ARNI), have improved the outcomes of HF with reduced ejection fraction (HFrEF). 3 , 4 , 5 In the PARADIGM‐HF trial, the landmark randomized controlled trial (RCT), ARNI significantly reduced mortality and HF hospitalization compared with the renin–angiotensin (RAS) blockade with enalapril in HFrEF. 6 Based on these findings, recent clinical guidelines recommend the use of ARNI as a first‐line drug for reducing morbidity and mortality of HFrEF. 3 , 7
Nonetheless, the PARADIGM‐HF trial had some limitations in generalizing the study findings to real‐world clinical practice. The PARADIGM‐HF trial excluded patients with advanced chronic kidney disease (CKD), symptomatic hypotension and systolic blood pressure <100 mmHg at screening. Furthermore, 19.8% of the enrolled patients dropped out due to intolerance to ARNI or enalapril. In the result, the PARADIGM‐HF trial comprised a younger population (mean age 63.8 years) who had lesser comorbidity than the real‐world population with HFrEF. 8 Medication adherence in RCTs may differ from real‐world clinical practice due to various factors, such as medication cost, safety concerns, polypharmacy, a high proportion of older patients and clinical inertia, resulting in low adherence in real‐world practice. 9 Thus, there is a pressing need to investigate the effect of ARNI and medication adherence in the contemporary clinical practice of HFrEF with a real‐world population.
Therefore, this PARADE‐HF trial (comparative effectiveness between ARNI and traditional RAS blockade in patients with HFrEF) aimed to compare 1 year mortality and any hospitalization with ARNI versus traditional RAS blockade in patients with HFrEF using a nationwide real‐world database. We hypothesized that high adherence to ARNI could have incremental value in improving HF outcomes compared with traditional RAS blockade in real‐world practice.
Methods
Data source
This nationwide population‐based cohort study used data from the Korean National Health Insurance Service (NHIS) Database. The Korean NHIS is a single medical insurer managed by the Korean Ministry of Health, Welfare, and Family Affairs. The NHIS offers obligatory healthcare for all Korean citizens, covering 97% of the Korean population. The NHIS database includes anthropometric information, drug prescriptions, and claims data for medical services using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD‐10) codes. Information about the participants' death dates was obtained from the electronic medical database of Statistics Korea. The detailed methodology of the NHIS database has been previously described. 10 The current database with HF has been approved by the memorandum of understanding between the Korean Society of Cardiology and the NHIS (database number: NHIS‐2022‐1‐376). The data analysed in the current study included the NHIS claims data from July 2017 to December 2021.
Ethical consideration
The study was approved by the Institutional Review Board for Human Research of Yonsei University Wonju Severance Christian Hospital (approval number: CR321358), and the study protocol followed the ethical guidelines of the Declaration of Helsinki. The informed consent was waived because the NHIS database was assembled after anonymization according to strict confidentiality guidelines.
Study patients
This study included patients with HFrEF who were new users of ARNI or users of traditional RAS blockade between July 2017 and December 2021. The diagnosis of HFrEF was defined as (1) the use of ARNI [because the Korea Health Insurance Review and Assessment Service only approved the prescription of ARNI in patients with HF with left ventricular (LV) EF ≤ 40%] and no previous use of ARNI for at least 3 years before the index date or (2) more than two claims for one of the following ICD‐10 diagnostic codes: HF with LV systolic dysfunction (I50.04), LV failure (I50.1), dilated cardiomyopathy (I42.0) or ischaemic cardiomyopathy (I25.5), and at least one examination of natriuretic peptide or echocardiography within 6 months of the prescription date of medication. The index date of the study participants was defined as the first prescription date that met the inclusion criteria. Patients who fulfilled the following criteria were excluded from the final analysis: age <18 years, cancer within 5 years, end‐stage renal disease with dialysis, heart transplantation or LV assistant device, human immunodeficiency virus infection, admission to nursing hospitals at index date, follow‐up period <6 months, no previous use of traditional RAS blockade before the index date and duration of prescription for ARNI or traditional RAS blockade <90 days. Of the selected patients with HFrEF, 18 342 had received ARNI (ARNI group), and 37 178 had received traditional RAS blockade (RAS blockade group) for at least 3 months.
Propensity score matching (1:1) was performed to assemble a matched cohort with similar baseline characteristics. After matching, 13 483 patients were selected for each group (ARNI and traditional RAS blockade groups). Figure S1 shows the schematic flow of patients throughout the study.
Drug exposure and covariates
From the index date, patients were followed up for a maximum of 1 year until the outcome occurred after a 90 day supply of medication. To assess medication adherence, we calculated the proportion of days covered (PDC) in the ARNI and traditional RAS blockade groups. PDC is the ratio of the total number of days of medication supplied during the follow‐up period divided by the number of days in the period. 11 The thresholds for adequate adherence were 80% or 90%. Baseline clinical characteristics, such as comorbidities, medications, and healthcare utilization, were measured based on claims 1 year before the index date. The burden of comorbidities was defined as more than two claims of ICD‐10 diagnostic codes, and medication use was defined as a prescription duration >90 days. The full list of covariates is presented in Tables 1 and S1. Frailty was assessed using the Charlson comorbidity index (CCI). 12
Table 1.
Baseline characteristics of study groups.
| Before matching | After matching | |||||
|---|---|---|---|---|---|---|
| ARNI | Traditional RAS blockade | ASD | ARNI | Traditional RAS blockade | ASD | |
| N | 18 342 | 37 178 | 13 483 | 13 483 | ||
| Age | 66.8 ± 13.5 | 69.6 ± 14.1 | 0.202 | 68.2 ± 13.3 | 69.0 ± 14.1 | 0.055 |
| Women | 6115 (33.3) | 16 623 (44.7) | 0.235 | 5082 (37.7) | 5097 (37.8) | 0.002 |
| SES | 0.063 | <0.001 | ||||
| Low | 5851 (31.9) | 11 500 (31.3) | 4231 (31.4) | 4144 (30.7) | ||
| Middle | 5887 (32.1) | 10 972 (29.9) | 4228 (31.4) | 4195 (31.1) | ||
| High | 6604 (36.0) | 14 229 (38.8) | 5024 (37.3) | 5144 (38.2) | ||
| Region | 0.144 | 0.027 | ||||
| Metropolitan | 3258 (17.8) | 6527 (17.6) | 2373 (17.6) | 2319 (17.2) | ||
| City | 5110 (27.9) | 8106 (21.8) | 3411 (25.3) | 3343 (24.8) | ||
| Rural | 9974 (54.4) | 22 545 (60.6) | 7699 (57.1) | 7821 (58.0) | ||
| Burden of comorbidities | ||||||
| CCI, median (IQR) | 4 (2–5) | 4 (2–5) | 0.044 | 4 (2–5) | 4 (2–6) | 0.027 |
| Diabetes mellitus | 5332 (29.1) | 8395 (22.6) | 0.149 | 3619 (26.8) | 3647 (27.0) | 0.005 |
| Hypertension | 14 302 (78.0) | 26 457 (71.2) | 0.157 | 10 335 (76.7) | 10 334 (76.6) | <0.001 |
| Dyslipidaemia | 10 563 (57.6) | 18 108 (48.7) | 0.179 | 7416 (55.0) | 7328 (54.3) | 0.013 |
| IHD | 11 320 (61.7) | 19 493 (52.4) | 0.188 | 7985 (59.2) | 7896 (58.6) | 0.013 |
| Cerebrovascular disease | 1861 (10.1) | 4231 (11.4) | 0.040 | 1427 (10.6) | 1509 (11.2) | 0.020 |
| PAD | 477 (2.6) | 1050 (2.8) | 0.014 | 376 (2.8) | 384 (2.8) | 0.004 |
| AF | 6079 (33.1) | 10 380 (27.9) | 0.114 | 4352 (32.3) | 4414 (32.7) | 0.010 |
| CKD | 1851 (10.1) | 5154 (13.9) | 0.116 | 1548 (11.5) | 1513 (11.2) | 0.008 |
| Medication | ||||||
| SGLT2 inhibitor | 1291 (7.0) | 1193 (3.2) | 0.174 | 726 (5.4) | 686 (5.1) | 0.013 |
| CCB | 3932 (21.4) | 12 951 (34.8) | 0.301 | 3476 (25.8) | 3421 (25.4) | 0.009 |
| Aldosterone antagonist | 9845 (53.7) | 10 606 (28.5) | 0.529 | 5928 (44.0) | 5952 (44.1) | 0.004 |
| Diuretics | 13 933 (76.0) | 21 780 (58.6) | 0.377 | 9733 (72.2) | 9874 (73.2) | 0.024 |
| β‐blocker | 13 954 (76.1) | 22 797 (61.3) | 0.322 | 9676 (71.8) | 9538 (70.7) | 0.023 |
| Digoxin | 2968 (16.2) | 4740 (12.7) | 0.098 | 2046 (15.2) | 2106 (15.6) | 0.012 |
| Statin | 10 268 (56.0) | 19 658 (52.9) | 0.062 | 7428 (55.1) | 7354 (54.5) | 0.001 |
| Healthcare utilization | ||||||
| Duration of RAS blockade before index date (months) | 37.0 (46.9) | 42.6 (47.8) | 0.117 | 40.1 (49.0) | 40.6 (47.1) | 0.010 |
| No. of outpatient clinic visits, median (IQR) | 22 (12–37) | 23 (12–39) | 0.035 | 22 (13–37) | 23 (12–38) | 0.008 |
| Hospitalization within 1 year | 11 780 (64.2) | 20 481 (55.1) | 0.187 | 8047 (59.7) | 8068 (59.8) | 0.031 |
| No. of hospitalization, median (IQR) | 2 (1–3) | 1 (1–2) | 2 (1–3) | 1 (1–2) | ||
| Duration of hospitalization (days) | 9 (4–19) | 6 (1–15) | 9 (3–18) | 7 (2–17) | ||
| ED visit | 472 (2.6) | 670 (1.8) | 0.053 | 319 (2.4) | 326 (2.4) | 0.003 |
Note: Data are expressed as frequencies (percentages), means ± standard deviations and medians (interquartile ranges).
Abbreviations: AF, atrial fibrillation; ARNI, angiotensin receptor‐neprilysin inhibitor; ASD, absolute standardized difference; CCB, calcium channel blocker; CCI, Charlson comorbidity index; ED, emergency department; IHD, ischaemic heart disease; PAD, peripheral artery disease; RAS, renin–angiotensin system; SES, socioeconomic status; SGLT2, sodium‐glucose co‐transporter‐2.
Study endpoint
The primary outcome of this study was a composite of all‐cause mortality or any hospitalization within 1 year. All‐cause mortality was defined as death using data from the National Statistics Korea. Hospitalization was defined as admission to a tertiary or general hospital, with admission to nursing hospitals excluded to reflect the burden of hospitalization accurately. However, due to the limited assessment of the NHIS database, event adjudication for the primary cause of events, such as cardiovascular mortality or hospitalized HF, was not fully verified; therefore, cardiovascular mortality or hospitalized HF was not evaluated as study endpoints.
Statistical analysis
Data are presented as the mean ± standard deviation and numbers (percentages) for continuous and categorical variables, respectively. Demographic and clinical characteristics were compared by calculating ASD between the groups. A multiple logistic regression model was used to calculate the propensity score, with the treatment group as the dependent variable and all baseline characteristics presented in Tables 1 and S1 as independent variables. We used 1:1 greedy nearest neighbour matching on the propensity score using a calliper width of 0.1 and duration of RAS blockade before the index date within 6 months. We considered an absolute standardized difference (ASD) <0.1 as an acceptable balance of covariates between groups. After matching, the distribution of propensity score was similar between groups (Figure S2b). All ASDs between the two groups were <0.1 for baseline covariates. The incidence rates of outcomes were calculated as the number of events per 100 person‐years. The cumulative incidence of outcomes was graphically presented using a Kaplan–Meier curve. A Cox proportional hazards regression model was used to evaluate the relationship between treatments and study outcomes, and the results were presented as hazard ratios (HRs) and 95% confidence intervals (CIs). For data analysis before matching, we used a multivariable Cox proportional hazards regression model, and all variables with an ASD ≥ 0.1 were included as independent covariates in the model. For data analysis after matching, we estimated the relative hazard by using a univariate Cox proportional hazards regression model with a robust sandwich variance estimator that accounts for correlations within the matched pairs. 13 The risk of outcome in the ARNI group relative to the RAS blockade group was analysed according to the PDC level (≥80, <80 or ≥90, and <90%). We also performed an on‐treatment sensitivity analysis. The on‐treatment period was defined as the consecutive treatment period from the index date to 90 days after the end date of the last ARNI prescription. If the next prescription was filled within 30 days of the expected end date of the previous prescription, the treatment was assumed to be uninterrupted. When the on‐treatment analysis was performed, the study outcomes of the participants in the ARNI group and their matched participants in the RAS blockade group were only followed up during the on‐treatment period. We conducted subgroup analyses stratified by pre‐specified subgroups.
All statistical analyses were performed using the SAS Enterprise Guide software, version 7.1 (SAS Institute Inc., Cary, NC, USA) and R software, version 4.1 (R Foundation for Statistical Computing, Vienna, Austria). A two‐sided P value <0.05 was considered statistically significant.
Results
Baseline characteristics
Tables 1 and S1 present the baseline characteristics of the study population before and after propensity score matching. The ARNI group had a lower mean age and proportion of women than the RAS blockade group. The prevalence of diabetes, hypertension, dyslipidaemia, ischaemic heart disease and atrial fibrillation was higher in the ARNI group whereas the prevalence of CKD was higher in the RAS blockade group. The ARNI group had a higher frequency of HF medications, including SGLT2 inhibitors, aldosterone antagonists and β‐blockers, but a lower frequency of calcium channel blocker use, than the RAS blockade group. The frailty score (CCI) was comparable between the groups. After propensity score matching, 18 343 patients were included in each group, which resulted in well‐balanced baseline characteristics (Table 1). The distribution of propensity score before and after matching is presented as in Figure S2.
Primary outcome, all‐cause mortality, and any hospitalization
Table 2 and Figure 1 demonstrate the incidence rates and risk of the primary outcome between the two groups over a 1 year follow‐up period. The primary outcome, a composite of all‐cause mortality or any hospitalization, occurred in 4726 patients (49.1%) in the ARNI group and 5525 patients (64.1%) in the RAS blockade group (HR 0.78; 95% CI 0.75–0.81; P < 0.001). The ARNI group also showed a lower risk of all‐cause mortality (HR 0.86; 95% CI 0.78–0.94; P < 0.001) and any hospitalization (HR 0.77; 95% CI 0.74–0.80; P < 0.001) than the RAS blockade group. The results remained significant in the on‐treatment analysis (HR 0.76; 95% CI 0.73–0.79; P < 0.001) (Table S2).
TABLE 2.
Incidence rates and hazard ratios for all‐cause mortality or any hospitalization within 1‐year (reference group = RAS blockade).
| Before matching | After matching | |||||
|---|---|---|---|---|---|---|
| ARNI | Traditional RAS blockade | P | ARNI | Traditional RAS blockade | P | |
| N | 18 342 | 37 178 | 13 483 | 13 483 | ||
| Primary outcome | ||||||
| No. of event (IR) | 6475 (49.46) | 15 569 (63.27) | 4726 (49.12) | 5525 (64.06) | ||
| HR (95% CI) | ||||||
| Crude | 0.75 (0.73–0.78) | 1 (ref.) | <0.001 | 0.78 (0.75–0.81) | 1 (ref.) | <0.001 |
| Adjusted model a | 0.79 (0.77–0.82) | 1 (ref.) | <0.001 | |||
| All‐cause mortality within 1 year | ||||||
| No. of event (IR) | 1074 (6.60) | 2722 (8.17) | 832 (6.98) | 945 (8.10) | ||
| HR (95% CI) | ||||||
| Crude | 0.77 (0.71–0.83) | 1 (ref.) | <0.001 | 0.86 (0.78–0.94) | 1 (ref.) | 0.001 |
| Adjusted model a | 0.87 (0.80–0.94) | 1 (ref.) | 0.001 | |||
| Any hospitalization within 1 year | ||||||
| No. of event (IR) | 6230 (47.59) | 15 088 (61.32) | 4535 (47.14) | 5364 (62.19) | ||
| HR (95% CI) | ||||||
| Crude | 0.75 (0.73–0.77) | 1 (ref.) | <0.001 | 0.77 (0.74–0.80) | 1 (ref.) | <0.001 |
| Adjusted model a | 0.79 (0.76–0.81) | 1 (ref.) | <0.001 | |||
Note: Incidence rate is expressed per 100 person‐years.
All variables with ASD ≥ 0.1 before matching were included in the model as adjusting variables.
Abbreviations: ARNI, angiotensin receptor‐neprilysin inhibitor; ASD, absolute standardized difference; CI, confidence interval; HR, hazard ratio; IR, incidence rate; RAS, renin–angiotensin system.
Figure 1.
Primary outcome, all‐cause mortality and any hospitalization. The primary outcome is a composite of all‐cause mortality and hospitalization (A). The cumulative incidence of the primary outcome, all‐cause mortality (B) and any hospitalization (C) are compared between the two groups using the Kaplan–Meier method. The graphs are truncated after 12 months. The cumulative incidence is modelled using cubic splines. ARNI, angiotensin receptor neprilysin inhibitor; CI, confidence interval; RAS, renin–angiotensin system.
Primary outcome in pre‐specified subgroup
The consistency of the treatment effect on the primary outcome was evaluated in predefined subgroups (Figure 2). The results indicated that the effect of ARNI on the primary outcome was consistent with that of RAS blockade across most subgroups. The positive impact of ARNI was particularly notable in the older population, those who did not use aldosterone antagonists and during the time period 2020–2021 (amid the COVID‐19 pandemic) compared with that in the time period 2017–2019 (pre‐COVID‐19 pandemic).
Figure 2.
Primary outcome in pre‐specified subgroups. The consistency of the treatment effect on the primary outcome is assessed in pre‐specified subgroups. RAS, renin–angiotensin system; SGLT, sodium‐glucose cotransporter.
Medication adherence and outcomes
The mean PDC during 1 year was 92.6 ± 14.5% and 90.9 ± 17.7% in the ARNI and RAS blockade groups, respectively (P < 0.001, ASD = 0.105). The distribution of PDC in the study population is shown in Figure S3. A total of 88.0% (11 870) and 85.7% (11 553) of the patients had a PDC ≥ 80% in the ARNI and RAS blockade groups, respectively (P < 0.001, ASD = 0.111). In the group with a PDC ≥ 80%, the risk for the primary outcome was significantly lower in the ARNI group than that in the RAS blockade group (HR 0.75; 95% CI 0.72–0.78; P < 0.001) while the risk reduction of ARNI was not observed in the group with a PDC < 80%. The beneficial effect was more pronounced in the group with a PDC ≥ 80% than that in the group with a PDC < 80% (P for interaction <0.001) (Table 3). ARNI reduced all‐cause mortality only in patients with a PDC ≥ 80% (HR 0.85; 95% CI 0.76–0.94; P = 0.002), but not in patients with a PDC < 80%. However, the interaction between the PDC groups was not significant. Furthermore, the risk reduction for any hospitalization with ARNI was more significant in the PDC ≥ 80% group (HR 0.85; 95% CI 0.76–0.94; P = 0.002) than that in the PDC < 80% group (P for interaction <0.001). Cox linear regression analysis showed statistical significance of the differences in primary outcome rates between the ARNI and RAS blockade groups at a PDC threshold of 91%. The statistical significance of this benefit was sustained for PDC values up to 100% (Figure S4).
Table 3.
Hazard ratios for association between ARNI, RAS blockade and 1 year outcomes according to the proportion of days covered (reference group = RAS blockade).
| ARNI | Traditional RAS blockade | ||||||
|---|---|---|---|---|---|---|---|
| N (%) | Event (IR) | N (%) | Event (IR) | HR (95% CI) | P | P for interaction | |
| Primary outcome | |||||||
| ≥80% | 11 553 (85.7) | 4000 (48.63) | 11 870 (88.0) | 4899 (65.47) | 0.75 (0.72–0.78) | <0.001 | <0.001 |
| <80% | 1930 (14.3) | 726 (52.04) | 1613 (12.0) | 626 (54.77) | 0.95 (0.85–1.05) | 0.313 | |
| ≥90% | 10 463 (77.6) | 3558 (47.54) | 10 676 (79.2) | 4430 (66.40) | 0.73 (0.69–0.76) | <0.001 | <0.001 |
| <90% | 3020 (22.4) | 1168 (54.67) | 2807 (20.8) | 1095 (56.05) | 0.97 (0.90–1.06) | 0.528 | |
| All‐cause mortality | |||||||
| ≥80% | 11 158 (82.8) | 615 (6.27) | 11 196 (83.0) | 708 (7.38) | 0.85 (0.76–0.94) | 0.002 | 0.584 |
| <80% | 2325 (17.2) | 217 (10.33) | 2287 (17.0) | 237 (11.49) | 0.90 (0.75–1.08) | 0.250 | |
| ≥90% | 9953 (73.8) | 495 (5.65) | 9797 (72.7) | 549 (6.55) | 0.86 (0.76–0.97) | 0.014 | 0.771 |
| <90% | 3530 (26.2) | 337 (10.68) | 3686 (27.3) | 396 (12.06) | 0.88 (0.76–1.02) | 0.098 | |
| Any hospitalization | |||||||
| ≥80% | 11 553 (85.7) | 3850 (46.80) | 11 870 (88.0) | 4,768 (63.72) | 0.74 (0.71–0.78) | <0.001 | <0.001 |
| <80% | 1930 (14.3) | 685 (49.10) | 1613 (12.0) | 596 (52.14) | 0.94 (0.84–1.05) | 0.254 | |
| ≥90% | 10 463 (77.6) | 3431 (45.84) | 10 676 (79.2) | 4,328 (64.87) | 0.72 (0.69–0.75) | <0.001 | <0.001 |
| <90% | 3020 (22.4) | 1104 (51.67) | 2807 (20.8) | 1,036 (53.03) | 0.97 (0.89–1.06) | 0.525 | |
Note: Incidence rate is expressed per 100 person‐years.
Abbreviations: ARNI, angiotensin receptor‐neprilysin inhibitor; ASD, absolute standardized difference; CI, confidence interval; HR, hazard ratio; IR, incidence rate; PDC, proportion of days covered; RAS, renin–angiotensin system.
Discussion
The PARADE‐HF trial, a nationwide real‐world comparative effectiveness study, compared the outcomes of ARNI with those of traditional RAS blockade in patients with HFrEF. ARNI was more effective in reducing the risk of all‐cause mortality or any hospitalization than traditional RAS blockade in a real‐world cohort of 26 966 patients with HFrEF. Furthermore, high adherence to ARNI therapy was associated with improved 1 year mortality and hospitalization outcomes compared with traditional RAS blockade in the contemporary clinical practice of HFrEF.
Our study offers novel insights into the comparison of real‐world outcomes of ARNI and traditional RAS blockade in patients with HFrEF. Our study found that ARNI therapy was associated with improved 1 year mortality and hospitalization outcomes compared with traditional RAS blockade in HFrEF, even after controlling for baseline differences through propensity score matching. The risk reduction of primary outcomes in our study by 22% over 1 year, compared with the RAS blockade, is comparable with that observed in the PARADIGM‐HF trial. 6 The results of our study expand upon the findings of the PARADIGM‐HF trial by including a broader population in contemporary clinical practice. The results of our study are also consistent with previous real‐world studies from the United States, the United Kingdom and China, which have shown the superiority of ARNI over RAS blockade in terms of clinical outcomes. 14 , 15 , 16 , 17 , 18 However, our study stands out because of its large nationwide sample size and the recent time frame of 2017–2021, which allow for a comprehensive and up‐to‐date analysis of the real‐world impact of ARNI on patients with HFrEF.
Our results suggest that high adherence to ARNI therapy was associated with improved outcomes; however, low adherence to ARNI therapy was associated with outcomes similar to those of traditional RAS blockade. These findings emphasize that adherence to guideline‐directed medical therapy for HF is crucial for improving outcomes. 19 , 20 , 21 A previous study involving 897 patients with HFrEF from the GWTG‐HF registry found that patients with a PDC ≥ 80% for ARNI had a lower risk of all‐cause mortality and any hospitalization than those with a PDC < 80% for ARNI. 22 Our study differs from the GWTG‐HF registry study in that we compared the effect of PDC between ARNI and RAS blockade users. Our study provides a unique and important comparison between the real‐world outcomes of ARNI and RAS blockade in patients with HFrEF, specifically focusing on the impact of medication adherence on treatment outcomes. In both the ARNI and RAS blockade groups in our study, the mean PDC spanning 1 year was greater than 90%. Few studies have evaluated adherence to the ARNI regimen. In the GWTG registry, among 897 patients prescribed ARNI at discharge, only 32.9% of patients had PDC ≥ 80%. 22 There are two possible reasons for the high adherence observed in the present study. First, in South Korea, universal health insurance facilitates access to various treatments, including ARNI and comprehensive HF management for all patients. Second, our study included only patients who were continuously prescribed ARNI or RAS blockade for >90 days, which likely contributed to the observed elevated PDC. Our study population thus consisted predominantly of patients with medication adherence, making it an appropriate real‐world cohort to compare the effects of ARNI and RAS blockade in patients with high‐adherence HFrEF. A recent HF expert consensus recommends that ARNI should be directly initiated without initially performing RAS blockade to improve health status and achieve reverse cardiac remodelling, with traditional RAS blockade only considered for patients who are intolerant to or have contraindications for ARNI. Our study results align with this recommendation, showing that even in patients with HFrEF with high adherence to and good control of RAS blockade, ARNI improves clinical outcomes. This finding has important implications for clinical practice, suggesting that switching medication to ARNI can provide additional benefits, even for patients with already well‐managed RAS blockade.
The results from the pre‐specified subgroup analysis indicated that the positive impact of ARNI was consistent with that of RAS blockade across most subgroups. The positive effect of ARNI was notable in the older population. Although this study did not provide the data of LV EF, previous studies have shown that ARNI is more effective in patients with lower EF. 23 The elderly population may have multiple comorbidities and lower EF, which could explain the greater effectiveness of ARNI in this subgroup. Additionally, the results showed that the effect of ARNI was pronounced in patients who did not use aldosterone antagonists. The PARAGON‐HF trial revealed that aldosterone antagonists were linked to frequent hyperkalaemia. 24 , 25 These results may emphasize the caution required when combining ARNI and aldosterone antagonists, as hyperkalaemia may worsen. Our study also revealed that ARNI was more effective in the 2020–2021 period (the COVID‐19 pandemic) than that in the 2017–2019 period (pre‐COVID‐19 pandemic). The use of HF medications was comparable between the two periods in most countries. 26 , 27 In Korea, the medical healthcare system remained resilient during the pandemic without significant collapse. 28 These results may underscore the importance of contemporary treatment strategies rapidly evolving in the field of HF. The reasons for the increased effectiveness of the ARNI during the pandemic period remain unclear and require further investigation.
Study limitations
The present study has several limitations. First, this is an observational study with the potential for residual confounding factors. Although we used propensity score matching to control for baseline differences, the role of residual confounding factors cannot be completely excluded. Second, the study is limited to the assessment of all‐cause mortality and hospitalization, as event adjudication for the primary cause of events, such as cardiovascular mortality or hospitalized HF, using the Korean NHIS database was inconclusive. Third, PDC reflects the prescription of fills but not the actual medication intake. Finally, our study is limited by the availability of data from the NHIS database. For example, we were unable to assess the presence of other important parameters, such as drug dose, echocardiographic parameters and biomarker levels. Further studies are needed to evaluate the effect of these parameters on the clinical outcomes of ARNI.
Conclusions
Our study suggests that high adherence to ARNI therapy was associated with improved 1 year mortality and hospitalization outcomes compared with traditional RAS blockade in the contemporary clinical practice of HFrEF. These results underscore the importance of medication adherence for optimizing the benefits of ARNI therapy in HFrEF.
Conflict of interest
The authors declare that there is no conflict of interest.
Funding
This work was supported by Novartis Pharmaceutical, Bio and Medical Technology Development Program of the National Research Foundation (NRF), funded by the Korean government (MSIT) (Nos. RS‐2024‐00440824 and RS‐2024‐00340592), the Korea Disease Control and Prevention Agency (Nos. 2019‐ER6303‐00, 2019‐ER6303‐01, 2019‐ER6303‐02, 2022‐ER0908‐00, 2022‐ER0908‐01, and 2022‐ER0908‐02) and the Korea National Institute of Health (KNIH) (Nos. 2023ER080600 and 2023ER080601). The funding organizations did not play a role in study design, data collection, data analysis, data interpretation or manuscript writing. The corresponding author and statistician had complete access to all study data, and the corresponding author was responsible for making the final decision to submit the manuscript for publication.
Supporting information
Table S1. Baseline Characteristics of Study Groups.
Table S2. Risk of all‐cause mortality and any hospitalization during on‐treatment period.
Figure S1. Schematic of study flow.
Figure S2. The distribution of propensity score before and after matching.
Figure S3. The distribution of PDC for the study population.
Figure S4. Hazard ratios for association between ARNI or RAS blockade and one‐year outcomes according to the continuous PDC variable.
Acknowledgements
The authors thank all the participants in this study.
Cho, D.‐H. , Choi, J. , Youn, J.‐C. , Kim, M.‐N. , Lee, C. J. , Son, J.‐W. , and Yoo, B.‐S. (2025) Angiotensin receptor‐neprilysin inhibitor adherence and outcomes in heart failure with reduced ejection fraction. ESC Heart Failure, 12: 603–612. 10.1002/ehf2.15117.
Dong‐Hyuk Cho and Jimi Choi contributed equally to this work and are joint first authors.
References
- 1. Park JJ, Lee CJ, Park SJ, Choi JO, Choi S, Park SM, et al. Heart failure statistics in Korea, 2020: a report from the Korean Society of Heart Failure. Int J Heart Fail 2021;3:224‐236. doi: 10.36628/ijhf.2021.0023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Cho DH, Yoo BS. Current prevalence, incidence, and outcomes of heart failure with preserved ejection fraction. Heart Fail Clin 2021;17:315‐326. doi: 10.1016/j.hfc.2021.03.002 [DOI] [PubMed] [Google Scholar]
- 3. McDonagh TA, Metra M, Adamo M, Gardner RS, Baumbach A, Böhm M, et al. 2021 ESC guidelines for the diagnosis and treatment of acute and chronic heart failure. Eur Heart J 2021;42:3599‐3726. doi: 10.1093/eurheartj/ehab368 [DOI] [PubMed] [Google Scholar]
- 4. Choi HM, Park MS, Youn JC. Update on heart failure management and future directions. Korean J Intern Med 2019;34:11‐43. doi: 10.3904/kjim.2018.428 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Lee KS, Noh J, Park SM, Choi KM, Kang SM, Won KC, et al. Evaluation and management of patients with diabetes and heart failure: a Korean Diabetes Association and Korean Society of Heart Failure Consensus statement. Int J Heart Fail 2023;5:1‐20. doi: 10.36628/ijhf.2022.0028 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. McMurray JJ, Packer M, Desai AS, Gong J, Lefkowitz MP, Rizkala AR, et al. Angiotensin‐neprilysin inhibition versus enalapril in heart failure. N Engl J Med 2014;371:993‐1004. doi: 10.1056/NEJMoa1409077 [DOI] [PubMed] [Google Scholar]
- 7. Heidenreich PA, Bozkurt B, Aguilar D, Allen LA, Byun JJ, Colvin MM, et al. 2022 AHA/ACC/HFSA guideline for the management of heart failure: a report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation 2022;145:e895‐e1032. doi: 10.1161/CIR.0000000000001063 [DOI] [PubMed] [Google Scholar]
- 8. Huo X, Armitage J. Use of run‐in periods in randomized trials. JAMA 2020;324:188‐189. doi: 10.1001/jama.2020.6463 [DOI] [PubMed] [Google Scholar]
- 9. Jarjour M, Henri C, de Denus S, Fortier A, Bouabdallaoui N, Nigam A, et al. Care gaps in adherence to heart failure guidelines: clinical inertia or physiological limitations? JACC Heart Fail 2020;8:725‐738. doi: 10.1016/j.jchf.2020.04.019 [DOI] [PubMed] [Google Scholar]
- 10. Cheol Seong S, Kim YY, Khang YH, Heon Park J, Kang HJ, Lee H, et al. Data resource profile: the National Health Information Database of the National Health Insurance Service in South Korea. Int J Epidemiol 2017;46:799‐800. doi: 10.1093/ije/dyw253 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Prieto‐Merino D, Mulick A, Armstrong C, Hoult H, Fawcett S, Eliasson L, et al. Estimating proportion of days covered (PDC) using real‐world online medicine suppliers' datasets. J Pharm Policy Pract 2021;14:113. doi: 10.1186/s40545-021-00385-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Formiga F, Moreno‐Gonzalez R, Chivite D, Franco J, Montero A, Corbella X. High comorbidity, measured by the Charlson comorbidity index, associates with higher 1‐year mortality risks in elderly patients experiencing a first acute heart failure hospitalization. Aging Clin Exp Res 2018;30:927‐933. doi: 10.1007/s40520-017-0853-1 [DOI] [PubMed] [Google Scholar]
- 13. Austin PC. The use of propensity score methods with survival or time‐to‐event outcomes: reporting measures of effect similar to those used in randomized experiments. Stat Med 2014;33:1242‐1258. doi: 10.1002/sim.5984 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Albert NM, Swindle JP, Buysman EK, Chang C. Lower hospitalization and healthcare costs with sacubitril/valsartan versus angiotensin‐converting enzyme inhibitor or angiotensin‐receptor blocker in a retrospective analysis of patients with heart failure. J Am Heart Assoc 2019;8:e011089. doi: 10.1161/jaha.118.011089 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Chen W, Liu Y, Li Y, Dang H. Sacubitril/valsartan improves cardiac function in Chinese patients with heart failure: a real‐world study. ESC Heart Fail 2021;8:3783‐3790. doi: 10.1002/ehf2.1349 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Desai RJ, Patorno E, Vaduganathan M, Mahesri M, Chin K, Levin R, et al. Effectiveness of angiotensin‐neprilysin inhibitor treatment versus renin–angiotensin system blockade in older adults with heart failure in clinical care. Heart 2021;107:1407‐1416. doi: 10.1136/heartjnl-2021-319405 [DOI] [PubMed] [Google Scholar]
- 17. McMurray JJV, Trueman D, Hancock E, Cowie MR, Briggs A, Taylor M, et al. Cost‐effectiveness of sacubitril/valsartan in the treatment of heart failure with reduced ejection fraction. Heart 2018;104:1006‐1013. doi: 10.1136/heartjnl-2016-310661 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Tan NY, Sangaralingham LR, Sangaralingham SJ, Yao X, Shah ND, Dunlay SM. Comparative effectiveness of sacubitril‐valsartan versus ACE/ARB therapy in heart failure with reduced ejection fraction. JACC Heart Fail 2020;8:43‐54. doi: 10.1016/j.jchf.2019.08.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Hood SR, Giazzon AJ, Seamon G, Lane KA, Wang J, Eckert GJ, et al. Association between medication adherence and the outcomes of heart failure. Pharmacotherapy 2018;38:539‐545. doi: 10.1002/phar.2107 [DOI] [PubMed] [Google Scholar]
- 20. Schulz M, Griese‐Mammen N, Anker SD, Koehler F, Ihle P, Ruckes C, et al. Pharmacy‐based interdisciplinary intervention for patients with chronic heart failure: results of the PHARM‐CHF randomized controlled trial. Eur J Heart Fail 2019;21:1012‐1021. doi: 10.1002/ejhf.1503 [DOI] [PubMed] [Google Scholar]
- 21. Kim IC, Youn JC, Jang SY, Lee SE, Cho HJ, Choi JO, et al. Physician adherence and patient‐reported outcomes in heart failure with reduced ejection fraction in the era of angiotensin receptor‐neprilysin inhibitor therapy. Sci Rep 2022;12:7730. doi: 10.1038/s41598-022-11740-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Carnicelli AP, Li Z, Greiner MA, Lippmann SJ, Greene SJ, Mentz RJ, et al. Sacubitril/valsartan adherence and postdischarge outcomes among patients hospitalized for heart failure with reduced ejection fraction. Heart Failure 2021;9:876‐886. doi: 10.1016/j.jchf.2021.06.018 [DOI] [PubMed] [Google Scholar]
- 23. Solomon SD, Vaduganathan M, Claggett B, Packer M, Zile M, Swedberg K, et al. Sacubitril/valsartan across the spectrum of ejection fraction in heart failure. Circulation 2020;141:352‐361. doi: 10.1161/circulationaha.119.044586 [DOI] [PubMed] [Google Scholar]
- 24. Jering KS, Zannad F, Claggett B, Mc Causland FR, Ferreira JP, Desai A, et al. Cardiovascular and renal outcomes of mineralocorticoid receptor antagonist use in PARAGON‐HF. JACC Heart Fail 2021;9:13‐24. doi: 10.1016/j.jchf.2020.08.014 [DOI] [PubMed] [Google Scholar]
- 25. Riello RJ 3rd, Pitt B. ARNI and MRA combination in PARAGON‐HF: odd couple or dynamic duo? JACC Heart Fail 2021;9:25‐27. doi: 10.1016/j.jchf.2020.10.005 [DOI] [PubMed] [Google Scholar]
- 26. Keshvani N, Mehta A, Alger HM, Rutan C, Williams J, Zhang S, et al. Heart failure quality of care and in‐hospital outcomes during the COVID‐19 pandemic: findings from the Get With The Guidelines‐Heart Failure registry. Eur J Heart Fail 2022;24:1117‐1128. doi: 10.1002/ejhf.2484 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Yousufuddin M, Yamani MH, Kashani KB, Zhu Y, Wang Z, Seshadri A, et al. Characteristics, treatment patterns, and clinical outcomes after heart failure hospitalizations during the COVID‐19 pandemic, march to October 2020. Mayo Clin Proc 2023;98:31‐47. doi: 10.1016/j.mayocp.2022.09.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Cho DH, Yoo BS, Son JW, Kim IC, Park SM, Choi DJ. COVID‐19 ‐ implications for patients with heart failure: the Korean Society of Heart Failure's clinical recommendations. Int J Heart Fail 2020;2:111‐114. doi: 10.36628/ijhf.2020.0015 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Baseline Characteristics of Study Groups.
Table S2. Risk of all‐cause mortality and any hospitalization during on‐treatment period.
Figure S1. Schematic of study flow.
Figure S2. The distribution of propensity score before and after matching.
Figure S3. The distribution of PDC for the study population.
Figure S4. Hazard ratios for association between ARNI or RAS blockade and one‐year outcomes according to the continuous PDC variable.