Comparative Effectiveness of Renin-Angiotensin System Inhibitors in Heart Failure With Nonreduced Ejection Fraction: A Multicenter Cohort Study

Abstract

Background

The comparative effectiveness of angiotensin-converting enzyme inhibitors (ACEis) versus angiotensin receptor blockers (ARBs) in heart failure with nonreduced ejection fraction (HFnon-rEF) remains uncertain. We evaluated long-term outcomes of these therapies in a real-world cohort following hospitalization for acute heart failure.

Methods

This retrospective multicenter study (2005-2019) included 5837 patients with HFnon-rEF (left ventricular ejection fraction ≥ 40%). Patients were categorized by discharge prescription into ACEi, ARB, or non-renin-angiotensin system inhibitor (RASi) groups. A 14-day landmark approach ensured pharmacological stabilization and reduced time-related bias. Inverse probability of treatment weighting was used to balance covariates (SMD < 0.1). The primary outcome was a composite of cardiovascular (CV) death or heart failure rehospitalization at 1 year.

Results

The primary composite outcome was similar between ARB and ACEi users (adjusted hazard ratio [HR], 0.94; 95% confidence interval [CI], 0.75-1.17; P = .556). However, ARB therapy was associated with lower all-cause mortality compared with ACEi (adjusted HR, 0.74; 95% CI, 0.55-0.99; P = .041) and non-RASi (adjusted HR, 0.80; 95% CI, 0.67-0.94; P = .008). Subgroup analyses showed generally consistent directional associations between ARB use and lower mortality, although these findings were exploratory.

Conclusions

In this real-world cohort of patients with HFnon-rEF, ARB use was associated with lower all-cause mortality compared with ACEi, despite similar CV outcomes. These findings do not establish superiority and should be interpreted as hypothesis-generating.

Keywords

Heart failure with preserved ejection fraction heart failure with mildly reduced ejection fraction angiotensin receptor blocker ACE inhibitor mortality real-world evidence

Introduction

Acute heart failure (AHF) is characterized by the rapid onset or worsening of heart failure (HF) symptoms and signs, commonly resulting from structural or functional cardiac abnormalities that lead to congestion and hemodynamic instability.^1,2 Among patients with HF and nonreduced ejection fraction (HFnon-rEF), including HF with preserved ejection fraction (HFpEF; left ventricular ejection fraction [LVEF] ≥ 50%) and HF with mildly reduced ejection fraction (HFmrEF; LVEF 40%-49%), long-term prognosis remains poor, with substantial risks of rehospitalization and death.^1–4

Therapeutic strategies for HFnon-rEF have evolved considerably in recent years. Guideline-directed therapies such as angiotensin-converting enzyme inhibitor (ACEi) and angiotensin receptor blocker (ARB) are frequently prescribed to reduce cardiovascular (CV) risk and manage comorbid conditions.^5–9 In addition, newer pharmacological agents, including sodium-glucose cotransporter-2 inhibitors and mineralocorticoid receptor antagonists, have demonstrated clinical benefit in randomized trials involving this population.^10–12 Nevertheless, optimal treatment selection remains challenging, particularly in HFmrEF, whose clinical phenotype overlaps with both HFpEF and HF with reduced ejection fraction (HFrEF).^13–15

Most prior clinical trials have evaluated HFpEF and HFmrEF together, although emerging evidence suggests that patients with HFmrEF may respond more favorably to therapies traditionally used for HFrEF.^4,13,16–18 However, no large-scale randomized trial has been specifically designed to compare renin-angiotensin system inhibitor (RASi) strategies within the HFmrEF subgroup, and direct comparative evidence between ACEi and ARB in HFnon-rEF remains limited.

Given the uncertainty regarding optimal RASi selection in HFnon-rEF, particularly following hospitalization for AHF, real-world comparative data may help inform clinical decision making. Therefore, we aimed to evaluate the comparative effectiveness of ACEi and ARB therapy on clinical outcomes in patients with HFpEF and HFmrEF using a large multicenter cohort.

Patients and Methods

Data Source

The Chang Gung Research Database (CGRD) is a large-scale, de-identified electronic health record database derived from the Chang Gung Memorial Hospital (CGMH) system in Taiwan, which includes four tertiary medical centers and three teaching hospitals with more than 10,000 beds. The database contains longitudinal information on demographics, diagnoses, laboratory results, prescriptions, imaging studies, echocardiographic parameters, and hospitalization records from January 1, 2001 through March 31, 2021. The validity of CGRD-based research has been described previously.¹⁹ This study was approved by the Institutional Review Board of CGMH (No. 202100393B0C601).

Study Design and Population

This retrospective multicenter cohort study included patients hospitalized for acute decompensated HF between January 1, 2005 and December 31, 2019. The index admission was defined as the first eligible hospitalization for HF during the study period, and the index date was defined as the date of discharge. To reduce inclusion of prevalent cases, patients with prior HF-related admissions within the integrated CGMH system before January 1, 2005 were excluded. Eligible patients were adults aged ≥18 years with nonreduced ejection fraction, defined as HFpEF (LVEF ≥50%) or HFmrEF (LVEF 40%-49%), with an in-hospital echocardiographic examination available during the index admission. Patients who died during hospitalization or within 14 days after discharge were excluded to allow assessment of initial treatment exposure and reduce the influence of peridischarge instability. Patients without follow-up contact beyond the 14-day landmark and without a recorded death event were also excluded. Additional exclusion criteria included type 1 diabetes mellitus, human immunodeficiency virus infection, malignancy, autoimmune disease, infective endocarditis, major organ transplantation, or prior cardiac device implantation (implantable cardioverter-defibrillator, permanent pacemaker, or cardiac resynchronization therapy). A total of 5837 patients met the study criteria and were categorized according to discharge prescriptions as ACEi users, ARB users, or patients receiving neither agent (non-RASi group) (Figure 1).

Figure 1.

Study enrollment and flow diagram.

Baseline Variables and Comorbidities

Patient demographics and clinical characteristics were extracted, including age, sex, socioeconomic status, and relevant comorbidities. These included hypertension, type 2 diabetes mellitus, dyslipidemia, atrial fibrillation, ischemic heart disease, prior stroke, peripheral arterial disease, venous thromboembolism, chronic obstructive pulmonary disease (COPD), gout, gastrointestinal bleeding, intracranial hemorrhage, and prior cardiac surgery (Coronary artery bypass graft (CABG) or valve replacement). Renal function was evaluated using estimated glomerular filtration rate (eGFR), and echocardiographic data were reviewed where available. Renal function was assessed using the last serum creatinine measurement obtained before discharge; eGFR was calculated with the IDMS-traceable MDRD (Isotope dilution mass spectrometry-traceable modification of diet in renal disease) equation (eGFR = 175 × SCr⁻¹·¹⁵⁴ × age⁻⁰·²⁰³ × 0.742 if female). Chronic kidney disease (CKD) was defined as eGFR < 60 mL/min/1.73 m². Acute kidney injury (AKI) events during hospitalization were not consistently coded in the database; therefore, AKI could not be directly adjusted in the propensity model. Because AKI often determines temporary withdrawal or nonreinitiation of RASis, this factor was considered a potential unmeasured confounder and discussed as a limitation.

Medication Assessment

Medication exposure was defined according to discharge prescriptions after clinical stabilization, representing the initial treatment strategy following hospitalization for AHF. All inpatient and discharge medications were reviewed using World Health Organization Anatomical Therapeutic Chemical codes and Taiwan National Health Insurance reimbursement records. To reduce time-related bias, a 14-day landmark design was applied. Patients were required to survive to day 14 after discharge, during which treatment exposure (ACEi, ARB, or non-RASi) was determined based on receipt of ≥7 days of therapy. However, although this approach was intended to mitigate time-related bias, it does not fully eliminate immortal-time bias or survivor-selection bias.

Outcomes and Follow Up

The primary outcome was a composite of CV death and HF rehospitalization. Secondary outcomes included all-cause mortality, HF rehospitalization, acute myocardial infarction (AMI), and ischemic stroke. Outcomes were identified using inpatient ICD-9-CM and ICD-10-CM codes (Supplemental Appendix 1). CV death was defined according to the Taiwan National Death Registry as death due to cardiac, hypertensive, or cerebrovascular causes.

Patients were followed by the index discharge. To minimize reverse causation and early postdischarge instability, outcome follow up for the primary analysis was initiated after a 30-day washout period. Follow up continued until the occurrence of a study event, death, or March 31, 2021, whichever came first. For the primary analysis, follow up was administratively censored at 1 year after the index discharge, ensuring a uniform observation window across treatment groups and minimizing bias related to differential follow-up duration.

Statistical Analysis

Baseline characteristics among the three treatment groups (ACEi, ARB, and non-RASi) were compared using Chi-squared tests for categorical variables and one-way analysis of variance (ANOVA) for continuous variables. To reduce potential confounding and simulate randomization, inverse probability of treatment weighting (IPTW) was applied based on propensity scores, generating a weighted cohort in which the distribution of observed baseline covariates was balanced across treatment arms.²⁰ This approach enhances the validity of treatment comparisons in observational studies.^20,21 Findings were generally consistent across IPTW-only and doubly adjusted models. Propensity scores were independently estimated for each pairwise comparison (ACEi vs non-RASi; ARB vs non-RASi; and ARB vs ACEi) using logistic-regression models including all baseline covariates. Stabilized weights were used to improve precision and reduce variance. Covariate balance was assessed using the standardized mean difference (SMD), with an absolute SMD < 0.10 indicating adequate balance. The study employed an intention-to-treat framework, where patients were analyzed according to their initial medication strategy at discharge. This approach was selected to reflect real-world clinical practice and to maintain the prognostic value of the baseline treatment decision, regardless of subsequent medication titration or switching during follow up.

Time-to-event outcomes—including all-cause mortality, CV death, HF rehospitalization, and AMI—were assessed over a one-year follow-up period. Cox proportional hazards models were used to estimate hazard ratios (HRs) for mortality endpoints. For nonfatal outcomes with competing risks (eg, HF rehospitalization, AMI, and ischemic stroke), subdistribution hazard models based on the Fine and Gray method were employed. All-cause and cause-specific mortality were analyzed using Cox proportional-hazards models with robust standard errors to account for weighting. Subgroup analyses were prespecified for age, sex, HFpEF/HFmrEF category, renal function, and key comorbidities. SMDs within subgroups were reassessed, and subgroup results with residual imbalance (SMD > 0.10) were considered exploratory. All analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC), with statistical significance defined as a two-sided P value < .05.

Results

Baseline Characteristic of Adult Patients With HF Among Three Groups

In Table 1, a total of 5837 adults with HFnon-rEF were classified into ACEi (n = 340), ARB (n = 951), and non-RASi (n = 4546) groups. ACEi and non-RASi users were older on average (mean age 69.2 and 71.4 years, respectively) and more frequently had HFpEF (74.7%) than ARB users (mean age 70.8 years). The ARB group contained a larger proportion of women (53.6%) and exhibited higher prevalences of hypertension (63.6%), diabetes (43.4%), dyslipidemia (27.7%), ischemic stroke (17.8%), and gout (9.8%). ARB recipients were also more often treated with β-blockers (50.6%) and statins (27.0%). After applying stabilized inverse probability weighting, all covariates were well-balanced, with SMDs either below 0.1 or approaching 0.1, indicating no differences in baseline characteristics between the pairwise comparison groups. The distribution of propensity scores demonstrated sufficient overlap between treatment groups, supporting the validity of the weighting model (Supplemental Figure S4).

Table 1.

Baseline Characteristics of the Study Patients.

	Unweighted						After Stabilized IPTW*
							Absolute SMD
	ACEi		ARB		Non-RASi		ACEi vs Non-RASi	ARB vs Non-RASi	ACEi vs ARB
Variable	N	(%)	N	(%)	N	(%)	ACEi vs Non-RASi	ARB vs Non-RASi	ACEi vs ARB
Total	340		951		4546
Age group, years
18-64	121	(35.6)	284	(29.9)	1354	(29.8)	0.104	0.023	0.006
65-74	81	(23.8)	252	(26.5)	1049	(23.1)	0.070	0.001	0.002
≥ 75	138	(40.6)	415	(43.6)	2143	(47.1)	0.040	0.022	0.004
Mean (SD)	69.2	(14.0)	70.8	(13.6)	71.4	(14.3)	0.110	0.004	0.011
Gender							0.042	0.009	0.030
Male	186	(54.7)	441	(46.4)	2234	(49.1)
Female	154	(45.3)	510	(53.6)	2312	(50.9)
LVEF							0.013	0.009	0.012
HFmrEF	122	(35.9)	287	(30.2)	1152	(25.3)
HFpEF	218	(64.1)	664	(69.8)	3394	(74.7)
Comorbidities
Hypertension	167	(49.1)	605	(63.6)	2362	(52.0)	0.019	0.016	<0.001
Diabetes mellitus, type II	115	(33.8)	413	(43.4)	1493	(32.8)	0.010	0.008	0.013
Dyslipidemia	65	(19.1)	263	(27.7)	981	(21.6)	0.017	<0.001	0.003
Atrial fibrillation	61	(17.9)	144	(15.1)	876	(19.3)	0.011	0.013	0.003
Peripheral arterial disease	8	(2.4)	34	(3.6)	167	(3.7)	0.077	0.031	0.034
VTE (PE and DVT)	10	(2.9)	15	(1.6)	108	(2.4)	0.018	0.008	0.016
COPD	48	(14.1)	132	(13.9)	660	(14.5)	0.034	0.010	0.002
Gouty arthritis	22	(6.5)	93	(9.8)	405	(8.9)	0.003	0.012	0.001
Gastrointestinal bleeding	45	(13.2)	177	(18.6)	808	(17.8)	0.038	0.013	0.038
Intracranial hemorrhage	2	(0.6)	7	(0.7)	64	(1.4)	0.056	0.003	0.014
Old ischemic stroke	49	(14.4)	169	(17.8)	588	(12.9)	0.014	0.015	0.034
Ischemic heart disease	160	(47.1)	402	(42.3)	1791	(39.4)	0.038	0.002	0.033
Cardiac surgery	8	(2.4)	26	(2.7)	215	(4.7)	0.022	0.009	0.006
eGFR, mL/min/1.73m²
≥60	248	(72.9)	542	(57.0)	2516	(55.4)	0.004	0.001	0.010
30-59	58	(17.1)	202	(21.2)	965	(21.2)	0.002	0.002	0.043
16-30	11	(3.2)	63	(6.6)	386	(8.5)	0.077	0.020	0.018
Dialysis	23	(6.8)	144	(15.1)	679	(14.9)	0.063	0.012	0.055
Medications at admission
Antiplatelet	230	(67.7)	591	(62.2)	2369	(52.1)	0.021	0.012	0.013
Beta-blockers	159	(46.8)	481	(50.6)	2172	(47.8)	0.082	0.082	0.082
MRA	40	(11.8)	146	(15.4)	833	(18.3)	0.052	0.052	0.052
Statin	84	(24.7)	257	(27.0)	1082	(23.8)	0.014	0.009	0.010
Outcomes (within one year after discharge)
Primary outcomes	115	(33.8)	317	(33.3)	1481	(32.6)	0.057	0.003	0.039
CV death	27	(7.9)	62	(6.5)	347	(7.6)	0.074	0.029	0.039
HF	101	(29.7)	284	(29.9)	1295	(28.5)	0.028	0.005	0.028
AMI	13	(3.8)	32	(3.4)	174	(3.8)	0.057	0.064	0.087
Stroke	21	(6.2)	49	(5.2)	188	(4.1)	0.014	0.037	0.020
All-cause mortality	70	(20.6)	163	(17.1)	923	(20.3)	0.099	0.083	0.137

*Stabilized IPTW adjustment on age, sex, LVEF group, hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, peripheral arterial disease, VTE, COPD, gouty arthritis, gastrointestinal bleeding, intracranial hemorrhage, old ischemic stroke, ischemic heart disease, cardiac surgery, eGFR, antiplatelet, beta-blockers, MRA, and statin.

Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; AMI, acute myocardial infarction; ARB, angiotensin receptor blocker; COPD, chronic obstruction pulmonary disease; CV, cardiovascular; DVT, deep vein thrombosis; eGFR, estimated glomerular filtration rate; HF, heart failure; HFmrEF, HF with mildly reduced ejection fraction; HFpEF, HF with preserved ejection fraction; IPTW, inverse probability treatment weighting; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; PE, pulmonary embolism; RASi, renin-angiotensin-system inhibitor; SD, standard deviation; SMD, standardized mean difference; VTE, venous thromboembolism.

Conversely, ACEi users showed the greatest burden of ischemic heart disease (47.1%) and the highest rates of antiplatelet (66.7%) and mineralocorticoid-receptor antagonist (18.3%) therapy. Renal function differed across groups: patients with ARB had lower eGFR values and a higher prevalence of CKD than their ACEi counterparts. By contrast, inpatient biochemical measurements and echocardiographic parameters were broadly comparable among the three cohorts (Table 2).

Table 2.

Biochemical Tests and Echocardiography During Hospitalization.

	ACEi				ARB				Non-RASi				Absolute SMD
Variable	Available N	(%)	Mean	(SD)	Available N	(%)	Mean	(SD)	Available N	(%)	Mean	(SD)	ACEi vs Non-RASi	ARB vs Non-RASi	ACEi vs ARB
Laboratory data
Hemoglobin, g/dL	113	(33.2)	11.4	(2.2)	428	(45.0)	10.8	(2.2)	2482	(54.6)	11.0	(2.4)	0.156	0.087	0.264
Platelet, 1000/µL	183	(53.8)	198.9	(84.9)	567	(59.6)	213.7	(88.8)	2875	(63.2)	205.6	(92.1)	0.075	0.090	0.198
Creatinine, mg/dL	146	(42.9)	1.7	(2.2)	571	(60.0)	2.3	(2.8)	3189	(70.1)	2.1	(2.4)	0.158	0.077	0.237
Na, mg/dL	209	(61.5)	138.3	(4.9)	696	(73.2)	138.4	(5.2)	3464	(76.2)	138.4	(5.2)	0.025	0.000	0.036
K, mg/dL	218	(64.1)	3.8	(0.7)	727	(76.4)	3.8	(0.7)	3570	(78.5)	3.8	(0.7)	0.084	0.027	0.045
Uric acid, mg/dL	100	(29.4)	7.2	(2.3)	310	(32.6)	7.6	(2.5)	1182	(26.0)	7.5	(2.7)	0.098	0.045	0.149
ALT, U/L	141	(41.5)	109.9	(492.2)	426	(44.8)	43.0	(170.7)	1977	(43.5)	61.8	(215.9)	0.127	0.097	0.203
BNP, pg/mL	40	(11.8)	970.3	(1100.8)	181	(19.0)	1070.4	(1347.3)	982	(21.6)	1043.8	(1193.8)	0.064	0.021	0.033
DBP, mm Hg	338	(99.4)	75.0	(11.5)	943	(99.2)	73.9	(12.6)	4491	(98.8)	72.9	(12.3)	0.178	0.083	0.083
SBP, mm Hg	339	(99.7)	129.1	(19.1)	946	(99.5)	133.6	(21.0)	4497	(98.9)	129.5	(20.5)	0.020	0.195	0.240
Echo parameters
EF, %	340	(100.0)	57.1	(11.7)	951	(100.0)	59.4	(12.6)	4546	(100.0)	60.1	(12.1)	0.251	0.054	0.197
LVEDD, mm	338	(99.4)	51.0	(8.5)	946	(99.5)	51.1	(7.9)	4504	(99.1)	49.7	(8.2)	0.161	0.177	0.033
LA size, mm	325	(95.6)	41.4	(8.9)	917	(96.4)	41.6	(8.6)	4211	(92.6)	42.0	(9.1)	0.073	0.045	0.038
TVPG, mm Hg	229	(67.4)	32.8	(14.5)	657	(69.1)	31.3	(13.8)	3191	(70.2)	31.5	(14.7)	0.084	0.018	0.125

Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ALT, alanine aminotransferase; ARB, angiotensin receptor blocker; BNP, B-type natriuretic peptide; DBP, diastolic blood pressure; EF, ejection fraction; LA, left atrial; LVEDD, left ventricular end-diastolic diameter; RASi, renin-angiotensin-system inhibitor; SBP, systolic blood pressure; SD, standard deviation; SMD, standardized mean difference; TVPG, transvalvular pressure gradient.

Outcomes Analysis by Discharge RASi or None

Supplemental eTable 2 summarizes unweighted and IPTW-unadjusted HRs for all study endpoints. Compared with the non-RASi cohort, ACEi therapy was not associated with any significant difference in the primary composite outcome (CV death or HF rehospitalization) or all-cause mortality in either crude or adjusted models. Similarly, ARB therapy showed no significant effect on the primary composite outcome versus non-RASi (Figure 2). However, ARB therapy was associated with a 20% lower risk of all-cause mortality at 1-year follow up (adjusted HR 0.80; 95% CI 0.67-0.94; P = .008). The cumulative incidence curves after IPTW adjustment (Supplemental Figures S1-S3) were generally consistent with the regression analyses. While no clear separation was observed for the primary composite outcome or most CV endpoints, a divergence in all-cause mortality curves was noted, favoring ARB therapy.

Figure 2.

Hazard ratios for one-year outcomes according to discharge therapy (ACEi, ARB, and non-RASi groups). The panel displays unadjusted and IPTW-adjusted estimates for CV death, heart failure rehospitalization, AMI, stroke, and all-cause mortality. *Non-RASi (neither ACEi nor ARB) is the reference category. ^†ACEi is the reference for the ARB-versus-ACEi comparison. ^‡Adjusted for age, sex, LVEF category, hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, peripheral arterial disease, venous thromboembolism, chronic obstructive pulmonary disease, gout, gastrointestinal bleeding, intracranial hemorrhage, prior ischemic stroke, ischemic heart disease, prior CABG, valve surgery, estimated glomerular filtration rate, and baseline use of antiplatelets, β-blockers, MRAs and statins. Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; AMI, acute myocardial infarction; CV, cardiovascular; IPTW, inverse probability treatment weighting; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; RASi, renin-angiotensin system inhibitor.

Direct comparison of ARBs with ACEi revealed similar rates of CV death, HF rehospitalization, AMI, and stroke, but ARB conferred a 26% reduction in all-cause mortality (adjusted HR 0.74; 95% CI 0.55-0.99; P = .041). Findings were consistent in unadjusted analyses.

Subgroup Analysis by Discharge RASi

Associations between treatment groups and all-cause mortality were evaluated using two models (Table 3, Figure 3, and Supplemental eTable 3). ARB use was associated with lower all-cause mortality compared with non-RASi, particularly among older patients (>65 years), women, and those with HFpEF (Table 3). In direct comparisons between ARB and ACEi, ARB use was directionally associated with lower all-cause mortality, especially among male patients, those aged <65 years, and those with HFpEF. These associations were consistent across adjusted models; however, they should be interpreted with caution given the modest effect size and the absence of adjustment for multiple comparisons (Figure 3 and Supplemental eTable 3).

Figure 3.

Subgroup analysis of all-cause mortality: ARB versus ACEi at one year. *ACEi is the reference group. ^‡Model adjusted for the same covariates listed in Figure 2. Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.

Table 3.

Subgroup Analysis of Associations Between ARB Versus Non-RASi and One Year After Discharge for the All-Cause Mortality.

	Unweighted				Stabilized Inverse Probability Weighting
	Unadjusted Analysis		Adjusted Analysis^‡		Unadjusted Analysis		Adjusted Analysis^‡
Subgroups	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
Age group, years
<65	0.95 (0.61-1.47)	.806	0.81 (0.51-1.27)	.350	0.90 (0.57-1.41)	.648	0.80 (0.51-1.26)	.339
≥65	0.78 (0.65-0.94)	.007	0.81 (0.68-0.97)	.023	0.78 (0.65-0.93)	.006	0.78 (0.65-0.94)	.008
Sex
Male	0.78 (0.61-1.01)	.061	0.84 (0.65-1.08)	.179	0.79 (0.61-1.01)	.058	0.80 (0.62-1.03)	.085
Female	0.82 (0.66-1.02)	.076	0.81 (0.65-1.01)	.065	0.82 (0.66-1.03)	.090	0.79 (0.63-0.99)	.043
LVEF
HFmrEF	0.91 (0.67-1.22)	.524	0.93 (0.69-1.26)	.631	0.93 (0.68-1.27)	.652	0.94 (0.69-1.29)	.714
HFpEF	0.77 (0.63-0.94)	.010	0.78 (0.64-0.96)	.017	0.76 (0.62-0.93)	.007	0.75 (0.62-0.92)	.005
Comorbidities
Hypertension
Yes	0.74 (0.60-0.92)	.007	0.76 (0.61-0.95)	.014	0.75 (0.59-0.94)	.012	0.75 (0.60-0.95)	.015
No	0.91 (0.70-1.19)	.495	0.92 (0.71-1.20)	.535	0.88 (0.69-1.13)	.310	0.85 (0.67-1.09)	.206
Diabetes mellitus, type II
Yes	0.78 (0.61-1.00)	.048	0.82 (0.64-1.06)	.129	0.74 (0.56-0.98)	.037	0.77 (0.58-1.02)	.064
No	0.81 (0.64-1.01)	.059	0.80 (0.64-1.01)	.056	0.84 (0.68-1.04)	.110	0.80 (0.65-0.99)	.036
Ischemic heart disease
Yes	0.91 (0.70-1.18)	.470	0.89 (0.69-1.16)	.398	0.87 (0.67-1.14)	.319	0.87 (0.66-1.15)	.329
No	0.75 (0.60-0.93)	.009	0.78 (0.63-0.98)	.030	0.77 (0.62-0.95)	.014	0.76 (0.61-0.94)	.012
eGFR, mL/min/1.73m²
≥60	0.95 (0.76-1.19)	.655	0.90 (0.72-1.13)	.365	0.96 (0.76-1.20)	.708	0.90 (0.71-1.13)	.355
30-59	0.72 (0.50-1.04)	.083	0.76 (0.52-1.11)	.152	0.65 (0.44-0.96)	.030	0.67 (0.45-1.00)	.048
16-30	0.73 (0.42-1.28)	.271	0.75 (0.43-1.32)	.316	0.70 (0.42-1.17)	.171	0.69 (0.41-1.17)	.168
Dialysis	0.64 (0.42-0.97)	.034	0.71 (0.47-1.08)	.110	0.67 (0.45-1.02)	.059	0.74 (0.49-1.13)	.163
Medications at admission
Antiplatelet
Yes	0.90 (0.73-1.11)	.315	0.93 (0.75-1.14)	.470	0.90 (0.73-1.12)	.355	0.92 (0.74-1.14)	.454
No	0.65 (0.49-0.87)	.004	0.65 (0.48-0.87)	.004	0.69 (0.53-0.89)	.005	0.65 (0.49-0.84)	.001
Beta-blockers
Yes	0.99 (0.79-1.25)	.941	1.04 (0.83-1.32)	.730	0.95 (0.75-1.21)	.686	0.99 (0.78-1.26)	.912
No	0.66 (0.52-0.85)	.001	0.66 (0.52-0.85)	.001	0.69 (0.55-0.88)	.002	0.68 (0.53-0.86)	.001
MRA
Yes	0.66 (0.43-1.01)	.057	0.60 (0.38-0.93)	.023	0.66 (0.43-0.99)	.044	0.59 (0.39-0.89)	.013
No	0.84 (0.70-1.01)	.062	0.87 (0.73-1.05)	.149	0.84 (0.70-1.01)	.064	0.86 (0.72-1.03)	.106
Statin
Yes	1.09 (0.80-1.48)	.588	1.11 (0.81-1.52)	.536	1.13 (0.82-1.55)	.456	1.15 (0.83-1.58)	.403
No	0.73 (0.59-0.89)	.002	0.74 (0.60-0.90)	.003	0.72 (0.59-0.88)	.001	0.71 (0.58-0.87)	.001

*Non-RASi as the reference group.

‡

Adjustment on age, sex, LVEF group, hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, peripheral arterial disease, VTE, COPD, gouty arthritis, gastrointestinal bleeding, intracranial hemorrhage, old ischemic stroke, ischemic heart disease, cardiac surgery, eGFR, antiplatelet, beta-blockers, MRA, and statin.

Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; CI, confidence interval; COPD, chronic obstruction pulmonary disease; eGFR, estimated glomerular filtration rate; HFmrEF, heart failure with mildly reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction; HR, hazard ratio; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; RASi, renin-angiotensin system inhibitor; VTE, venous thromboembolism.

Discussion

Selection between ARB and ACEi remains an important therapeutic consideration in patients with HFpEF and HFmrEF. In this multicenter real-world cohort, ARB and ACEi use were associated with comparable risks of CV death, HF rehospitalization, myocardial infarction, and stroke within the first year after discharge. However, ARB use was associated with lower all-cause mortality compared with both ACEi and no RAS blockade.

This apparent divergence between all-cause mortality and CV-specific outcomes likely reflects contributions from non-CV mortality as well as residual confounding not fully captured in the available data. Accordingly, this finding should not be interpreted as evidence of therapeutic superiority, but rather as an association observed within a real-world context. Although inverse probability of treatment weighting achieved good covariate balance and improved comparability across groups, residual confounding, treatment selection bias, and time-related biases—including immortal-time bias—cannot be fully excluded. Consistent findings across weighted and doubly adjusted models provide supportive but not definitive evidence. Taken together, our findings add to the emerging real-world literature and should be considered hypothesis-generating. These observations were also consistent with the Kaplan-Meier curves presented in the Supplemental Materials.

Several clinical variables appeared to influence the prescribing of RASi in this cohort. ACE inhibitors were more frequently prescribed to patients with a higher burden of CV comorbidities—including diabetes, ischemic heart disease, atrial fibrillation, and COPD—as well as those with advanced CKD or prior intracranial hemorrhage.^6,9 They were also more commonly used in older patients and those with HFpEF or fewer hypertensive comorbidities. This may also partially explain the relatively low proportion of in-hospital echocardiography observed in the cohort. These patterns likely reflect clinician-driven treatment selection based on comorbidity burden, drug tolerability, and perceived risk profiles, and may partly explain the observed baseline differences between the ACEi and ARB groups.

Patients with HFnon-rEF are generally younger than those with HFrEF, and higher left ventricular ejection fraction is associated with lower long-term mortality.^21,22 However, HFpEF and HFmrEF are characterized by substantial non-CV morbidity and clinical heterogeneity, complicating treatment optimization.^21–25 Given their overlapping phenotypes, biomarker profiles, and treatment strategies, these groups were analyzed together in the present study.^22,24,26

Previous randomized and meta-analytic evidence has not demonstrated a clear mortality or CV benefit of renin-angiotensin system blockade in HFpEF or HFnon-rEF,^26–28 prompting interest in alternative neurohormonal strategies such as angiotensin-neprilysin inhibition.^14,28 In these phenotypes, sudden cardiac death is less common than in HFrEF, and overall mortality is often driven by non-CV comorbidities.^22–25 In our cohort, CV causes accounted for 37.7% of all deaths, consistent with prior reports.^22,24 Prospective evidence in HFmrEF remains limited, as few interventional trials have specifically targeted this subgroup, and observational data are still evolving.²² While registry data such as J-MINUET have suggested potential benefits of β-blockers in selected HFnon-rEF populations,²⁹ a recent Cochrane review found insufficient evidence to support a clear advantage of ARBs, ACE inhibitors, or β-blockers in HFpEF.²⁷

Beyond mortality, CV events and HF hospitalization also warrant consideration. Trials of candesartan and valsartan have demonstrated reductions in composite CV outcomes in HFnon-rEF, although effects on all-cause mortality were neutral.²⁸ Although both ACE inhibitors and ARBs target the renin-angiotensin-aldosterone system, they differ in pharmacological mechanisms and adverse-effect profiles. Prior studies have frequently evaluated these agents as a combined class when assessing efficacy and safety,^8,27,28 suggesting the importance of considering potential differences between them in clinical practice.

ARB are generally better tolerated than ACE inhibitors, as they do not increase bradykinin levels and are less commonly associated with cough or angioedema.^30,31 Renal function may also influence treatment selection: in our cohort, ARB users had lower eGFR, suggesting that clinicians may preferentially avoid ACE inhibitors in patients with advanced kidney disease, consistent with prior recommendations.³² Conversely, evidence from meta-analyses in patients with type 2 diabetes mellitus suggests that ACE inhibitors, but not ARB, are associated with reductions in all-cause and CV mortality.^33,34 These findings highlight the complexity of treatment selection and suggest that differences between ACE inhibitors and ARB may be context-dependent and influenced by underlying comorbidities.

This study has several limitations. First, the retrospective observational design based on multicenter electronic health record data is subject to coding inaccuracies and residual confounding. Misclassification arising from healthcare encounters outside the study system would likely be nondifferential and may have biased estimates toward the null. Second, echocardiographic assessments were performed according to local clinical practice without central adjudication, which may have introduced variability in LVEF classification. Restricting inclusion to patients with in-hospital echocardiography improved temporal alignment with the index event but reduced sample size. Third, detailed clinical variables, including hemodynamic status, functional class, frailty, and medication tolerability, were unavailable. In particular, the absence of coded AKI is relevant because AKI may influence both treatment decisions and subsequent prognosis.^35,36 Fourth, although inverse probability of treatment weighting achieved acceptable balance across measured covariates, residual confounding and confounding by indication cannot be fully excluded. Fifth, although a landmark design was applied to mitigate time-related bias, immortal-time bias, and survivor-selection bias cannot be fully excluded, particularly given the requirement for early survival and treatment exposure.³⁷ Sixth, postdischarge medication adherence, dose titration, discontinuation, and treatment switching were not fully captured. Accordingly, the intention-to-treat framework reflects the initial discharge treatment strategy rather than sustained long-term exposure. Seventh, a strict new-user design was not feasible, and inclusion of prevalent users may have introduced a healthy-user effect. Eighth, multiple endpoints and subgroup analyses were evaluated without formal adjustment for multiplicity, increasing the risk of type I error.³⁸ These subgroup findings should therefore be considered exploratory. Finally, secular changes in HF management over the prolonged study period may have influenced treatment patterns and outcomes. Despite these limitations, this study provides large-scale real-world evidence in patients with HFnon-rEF following AHF hospitalization. The observed association between ARB therapy and lower all-cause mortality should be interpreted cautiously and warrants confirmation in prospective studies.

Conclusion

In this multicenter, real-world cohort of patients with HFpEF and HFmrEF, discharge prescription of an ARB was associated with a lower risk of all-cause mortality compared with both ACE inhibitors and no RAS blockade. These findings indicate an association between ARB use and lower all-cause mortality in HFnon-rEF, but do not establish superiority or causality. Prospective, adequately powered randomized trials are warranted to confirm these observations and define the optimal RAS inhibition strategy in this population.

Supplemental Material

sj-docx-1-cpt-10.1177_10742484261452742 - Supplemental material for Comparative Effectiveness of Renin-Angiotensin System Inhibitors in Heart Failure With Nonreduced Ejection Fraction: A Multicenter Cohort Study

Supplemental material, sj-docx-1-cpt-10.1177_10742484261452742 for Comparative Effectiveness of Renin-Angiotensin System Inhibitors in Heart Failure With Nonreduced Ejection Fraction: A Multicenter Cohort Study by Ming-Shyan Lin, MD, Po-Chang Wang, MD, Jung-Jung Chang, MD, Pao-Hsien Chu, MD, PhD, Meng-Hung Lin, DrPH, Yu-Sheng Lin, PhD, Tien-Hsing Chen, MD, and Ming-Horng Tsai, MD, PhD in Journal of Cardiovascular Pharmacology and Therapeutics

Footnotes

Acknowledgments

The authors thank the Health Information and Epidemiology Laboratory, Chang Gung Memorial Hospital, Chiayi Branch for the comments and assistance in data analysis.

ORCID iDs

Ming-Shyan Lin

Pao-Hsien Chu

Meng-Hung Lin

Ethics Consideration and Consent to Participate

Despite the fact that the patient's identity (ie, the chart number or national identification number) was encrypted, every patient was assigned a personal identification number (PIN) with de-identified information. The study protocol was approved by the Chang Gung Memorial Hospital Institutional Review Board (IRB number 202100393B0C601), who declare that informed consent is not required. Further, all methods in our study adhered to the principles of the Declaration of Helsinki as well as relevant guidelines and regulations. The statement was mentioned in the Methods section.

Author Contributions

MSL, MHT: Conceptualization.

MSL, PCW, THC, MHL, PHC, YSL: Data curation

THC, MHL, PHC: Formal analysis.

MSL, MHT: Funding acquisition.

JJC, MSL, MHT: Investigation.

PHC, YSL, MSL, MHT: Methodology.

MHT: Supervision.

MSL, MHT: Writing—original draft.

All authors were involved in writing—review & editing, and have read and approved the final version of the manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by research funding from Chang Gung Memorial Hospital (grant number: CGRPG6L0051).

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

This study used and analyzed datasets that can be obtained upon reasonable request from the corresponding author.

Role of the Funder/Sponsor

Chang Gung Memorial Hospital had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Supplemental Material

Supplemental material for this article is available online.

References

Arrigo

Jessup

Mullens

, et al. Acute heart failure. Nat Rev Dis Primers. 2020;6(1):16.

Mentz

O'Connor

. Pathophysiology and clinical evaluation of acute heart failure. Nat Rev Cardiol. 2016;13(1):28-35.

Omote

Verbrugge

Borlaug

. Heart failure with preserved ejection fraction: Mechanisms and treatment strategies. Annu Rev Med. 2022;73:321-337.

Peikert

Martinez

Vaduganathan

, et al. Efficacy and safety of dapagliflozin in heart failure with mildly reduced or preserved ejection fraction according to age: The DELIVER trial. Circ Heart Fail. 2022;15(10):e010080.

Abdin

Schulz

Riemer

, et al. Sacubitril/valsartan in heart failure: Efficacy and safety in and outside clinical trials. ESC Heart Fail. 2022;9(6):3737-3750.

Sotomi

Hikoso

Nakatani

, et al. OCVC-heart failure investigator. Medications for specific phenotypes of heart failure with preserved ejection fraction classified by a machine learning-based clustering model. Heart. 2023;109(16):1231-1240.

Lin

Liao

, et al. Comparison of pharmacological treatment effects on long-time outcomes in heart failure with preserved ejection fraction: A network meta-analysis of randomized controlled trials. Front Pharmacol. 2021;12:707777.

Leite

Sampaio

Saraiva

Diaz

Barros

Fontes-Carvalho

. The impact of heart failure therapy in patients with mildly reduced ejection fraction: A network meta-analysis. ESC Heart Fail. 2023;10(3):1822-1834.

Zhang

Sun

Zhang

, et al. Efficacy of guideline-directed medical treatment in heart failure with mildly reduced ejection fraction. ESC Heart Fail. 2023;10(2):1035-1042.

10.

Solomon

McMurray

JJV

Claggett

, et al. Dapagliflozin in heart failure with mildly reduced or preserved ejection fraction. N Engl J Med. 2022;387(12):1089-1098.

11.

Vaduganathan

Mentz

Claggett

, et al. Sacubitril/valsartan in heart failure with mildly reduced or preserved ejection fraction: A pre-specified participant-level pooled analysis of PARAGLIDE-HF and PARAGON-HF. Eur Heart J. 2023;44(31):2982-2993.

12.

Khan

Fonarow

Khan

, et al. Renin-angiotensin blockade in heart failure with preserved ejection fraction: A systematic review and meta-analysis. ESC Heart Fail. 2017;4(4):402-408.

13.

Savarese

Vedin

D'Amario

, et al. Prevalence and prognostic implications of longitudinal ejection fraction change in heart failure. JACC Heart Fail. 2019;7(4):306-317.

14.

Mentz

Ward

Hernandez

, et al. PARAGLIDE-HF Investigators. Angiotensin-neprilysin inhibition in patients with mildly reduced or preserved ejection fraction and worsening heart failure. J Am Coll Cardiol. 2023;82(1):1-12.

15.

Zuin

Temporelli

Bilato

, et al. Robustness of randomized controlled trial evidence for SGLT2 inhibitors in heart failure with preserved and mildly reduced ejection fraction. Eur J Intern Med. 2026;145:106615.

16.

Bhatt

Szarek

Steg

, et al. Sotagliflozin in patients with diabetes and recent worsening heart failure. N Engl J Med. 2021;384(2):117-128.

17.

Heidenreich

Bozkurt

Aguilar

, et al. 2022 AHA/ACC/HFSA guideline for the management of heart failure: Executive summary: A report of the American College of Cardiology/American Heart Association Joint Committee on clinical practice guidelines. Circulation. 2022;145(18):e876-e894.

18.

Lund

Benson

Dahlstrom

Edner

Friberg

. Association between use of beta-blockers and outcomes in patients with heart failure and preserved ejection fraction. JAMA. 2014;312:2008-2018.

19.

Tsai

Lin

Lee

, et al. Chang gung research database: A multi-institutional database consisting of original medical records. Biomed J. 2017;40(5):263-269.

20.

Austin

Stuart

. Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat Med. 2015;34(28):3661-3679.

21.

Chen

Huang

Liang

, et al. Five-year mortality of heart failure with preserved, mildly reduced, and reduced ejection fraction in a 4880 Chinese cohort. ESC Heart Fail. 2022;9(4):2336-2347.

22.

Savarese

Stolfo

Sinagra

Lund

. Heart failure with mid-range or mildly reduced ejection fraction. Nat Rev Cardiol. 2022;19(2):100-116.

23.

Lyu

Tan

, et al. Clinical characteristics and prognosis of heart failure with mid-range ejection fraction: Insights from a multi-centre registry study in China. BMC Cardiovasc Disord. 2019;19(1):209.

24.

Lauritsen

Gustafsson

Abdulla

. Characteristics and long-term prognosis of patients with heart failure and mid-range ejection fraction compared with reduced and preserved ejection fraction: A systematic review and meta-analysis. ESC Heart Fail. 2018;5(4):685-694.

25.

Lin

Wang

Lin

, et al. Acute heart failure with mildly reduced ejection fraction and myocardial infarction: A multi-institutional cohort study. BMC Cardiovasc Disord. 2023;23(1):272.

26.

Kuno

Ueyama

Fujisaki

Briasouli

Takagi

Briasoulis

. Meta-analysis evaluating the effects of renin-angiotensin-aldosterone system blockade on outcomes of heart failure with preserved ejection fraction. Am J Cardiol. 2020;125(8):1187-1193.

27.

Martin

Manoharan

Davies

Lumbers

. Beta-blockers and inhibitors of the renin-angiotensin aldosterone system for chronic heart failure with preserved ejection fraction. Cochrane Database Syst Rev. 2021;5(5):CD012721. Published 2021 May 22.

28.

Gronda

Vanoli

Iacoviello

. The PARAGON-HF trial: The sacubitril/valsartan in heart failure with preserved ejection fraction. Eur Heart J Suppl. 2020;22(Suppl L):L77-L81.

29.

Sakagami

Soeda

Saito

, et al. Clinical impact of beta-blockers at discharge on long-term clinical outcomes in patients with non-reduced ejection fraction after acute myocardial infarction. J Cardiol. 2023;81(1):83-90.

30.

Wilkerson

Winters

. Angiotensin-converting enzyme inhibitor-induced angioedema. Emerg Med Clin North Am. 2022;40(1):79-98.

31.

Borghi

Cicero

Agnoletti

Fiorini

Pathophysiology of cough with angiotensin-converting enzyme inhibitors: How to explain within-class differences?

Eur J Intern Med. 2023;110:10-15.

32.

Chung

Ruospo

Natale

, et al. Aldosterone antagonists in addition to renin angiotensin system antagonists for preventing the progression of chronic kidney disease. Cochrane Database Syst Rev. 2020;10(10):CD007004.

33.

Bhatt

Fonarow

Greene

, et al. Medical therapy before, during and after hospitalization in medicare beneficiaries with heart failure and diabetes: Get with the guidelines - heart failure registry. J Card Fail. 2024;30(2):319-328.

34.

Khan

Rashid

Shafi

Ferreira

Butler

. Management of heart failure with reduced ejection fraction in patients with diabetes mellitus and chronic kidney disease. Semin Nephrol. 2023;43(3):151429.

35.

Brar

James

Hemmelgarn

Klarenbach

Pannu

. Interdisciplinary chronic disease collaboration. Association of angiotensin-converting enzyme inhibitor or angiotensin receptor blocker use with outcomes after acute kidney injury. JAMA Intern Med. 2018;178(12):1681-1690.

36.

Meraz-Muñoz

Jeyakumar

Luo

, et al. Cardiovascular drug use after acute kidney injury among hospitalized patients with a history of myocardial infarction. Kidney Int Rep. 2022;8(2):294-304.

37.

Suissa

. Immortal time bias in pharmaco-epidemiology. Am J Epidemiol. 2008;167(4):492-499.

38.

Ranganathan

Pramesh

Buyse

Common pitfalls in statistical analysis: The perils of multiple testing. Perspect Clin Res. 2016;7(2):106-107.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.80 MB