Contemporary Guideline‐Directed Medical Therapy for Heart Failure in the United States: The EMPACE Study

Abstract

Background

Randomized trials and clinical guidelines support early initiation of guideline‐directed medical therapy (GDMT) for heart failure (HF). The EMPACE (Treatment Patterns of Guideline‐Directed Medical Therapies in Heart Failure Patients in the Real‐World) study examined GDMT use in US clinical practice among patients hospitalized with heart failure (HHF).

Methods

This observational cohort study examined US patient data from Optum's deidentified Market Clarity database (June 2020–September 2023). GDMT use was assessed in the 12 months before and after HHF. Discontinuation was assessed over 12 months after initiation.

Results

Among 17 210 patients (73% HF with reduced ejection fraction [EF], 4%, HF with mildly reduced EF, 23% HF with preserved EF), mean age was 69.2 years, and 60% were male. Before HHF, among patients with HF with reduced EF (HFrEF), only 1% received quadruple therapy; use of individual therapies was beta blockers 68%, angiotensin‐converting enzyme inhibitors/angiotensin receptor blockers 64%, mineralocorticoid receptor antagonists 23%, angiotensin receptor–neprilysin inhibitors (ARNI) 14%, and sodium–glucose cotransporter‐2 inhibitor (SGLT2i) 5%. After HHF, GDMT use improved modestly: quadruple therapy 2%, beta blockers 84%, angiotensin‐converting enzyme inhibitors/angiotensin receptor blockers 72%, mineralocorticoid receptor antagonists 38%, ARNI 26%, and SGLT2i 13%. Among patients receiving therapy post discharge, mean time‐to‐initiation was longest for SGLT2i (88 days) and shortest for beta blockers (15 days). Mean time‐to‐quadruple therapy was 109 days. ARNI had the highest 12‐month discontinuation rate (62%), followed by mineralocorticoid receptor antagonists (57%), SGLT2i (55%), and beta blockers (51%). Among patients with HF with mildly reduced EF (HFmrEF) and HF with preserved EF (HFpEF), only 7% each received SGLT2i before HHF compared with 12% and 9% post‐HHF (each with mean time‐to‐initiation 28 days), respectively.

Conclusions

Among patients hospitalized for HFrEF in contemporary US clinical practice, there were significant gaps in prehospitalization quadruple therapy and only modest GDMT improvement post‐discharge, with delayed initiation and high discontinuation rates. Similar patterns were observed with SGLT2i among patients with HFmrEF and HFpEF.

Nonstandard Abbreviations and Acronyms

ARNI

angiotensin receptor–neprilysin inhibitors

BB

beta blockers

GDMT

guideline‐directed medical therapy

HFmrEF

heart failure with mildly reduced ejection fraction

HFpEF

heart failure with preserved ejection fraction

HFrEF

heart failure with reduced ejection fraction

HHF

hospitalized with heart failure

MRA

mineralocorticoid receptor antagonists

SGLT2i

sodium–glucose cotransporter‐2 inhibitors

CLINICAL PERSPECTIVE.

What Is New?

• Among patients hospitalized for heart failure (HF) with reduced ejection fraction in contemporary US clinical practice, there were significant gaps in prehospitalization quadruple therapy and only modest improvement in guideline‐directed medical therapy post‐discharge, with delayed initiation and high discontinuation rates for individual therapies. Similar patterns were observed with sodium–glucose cotransporter‐2 inhibitors (SGLT2i) among patients hospitalized for HF with mildly reduced ejection fraction and HF with preserved ejection fraction.

What Are the Clinical Implications?

• These findings highlight the critical need to improve early and sustained guideline‐directed medical therapy use before and after. HF hospitalization represents a key opportunity to optimized guideline‐directed medical therapy, and these data suggest these opportunities are frequently being missed.

• Improving guideline‐directed medical therapy initiation and persistence could significantly improve outcomes for patients with HF across the ejection fraction spectrum.

Patients hospitalized with heart failure (HHF) are at a markedly elevated risk for rehospitalization, mortality, and reduced quality of life. ¹ , ² , ³ Early implementation of guideline‐directed medical therapy (GDMT) is crucial for reducing HF rehospitalization and mortality rates in patients with HF. ⁴ The American College of Cardiology, American Heart Association, and Heart Failure Society of America 2022 guidelines recommended using angiotensin‐converting enzyme inhibitors (ACEi)/angiotensin receptor blockers (ARB) or angiotensin receptor–neprilysin inhibitors (ARNI), beta blockers (BB), steroidal mineralocorticoid receptor antagonists (MRA), and SGLT2 (sodium–glucose cotransporter‐2) inhibitors (SGLT2i) as Class I therapies for managing HF with reduced ejection fraction (HFrEF). These same guidelines recommend SGLT2i as a Class 2a therapy for HF with mildly reduced or preserved EF (HFmrEF and HFpEF). ⁵ In addition, the 2023 American College of Cardiology Expert Consensus introduced the use of SGLT2i for patients with HFpEF as a foundational therapy to improve clinical outcomes and health status. ⁶

Despite guideline recommendations, prior studies have consistently shown suboptimal use of HF GDMT in both inpatient and outpatient settings. ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² To inform and improve clinical practice, a comprehensive understanding of contemporary GDMT implementation patterns is essential, particularly as recent guidelines have expanded to include newer GDMT classes such as SGLT2i and ARNIs. Likewise, most prior studies have focused primarily on describing medication use, with limited data examining timing and quantifying delays in GDMT initiation, or the timing of drug discontinuation, in US clinical practice. ¹³ , ¹⁴ In this context, the EMPACE (Treatment Patterns of Guideline‐Directed Medical Therapies in Heart Failure Patients in the Real‐World) study was designed to address this gap by providing comprehensive data on GDMT use patterns in contemporary US clinical practice. The objective of this study was to characterize contemporary real‐world patterns of GDMT use before and after HHF, including timing of initiation and treatment persistence, to identify specific opportunities for targeted interventions to improve GDMT implementation.

METHODS

The data, analytical methods, and study materials are not publicly available due to data use agreement restrictions. Access to the Optum Market Clarity database requires a separate data use agreement with the data provider.

Data Source

EMPACE is a retrospective, observational cohort study of US adults hospitalized for a primary diagnosis of HF. The study used Optum's de‐identified Market Clarity (Market Clarity) database, a large US‐based data source integrating administrative claims and electronic health records from >75 million enrollees since 2007.

The data are sourced from an integrated delivery network across diverse types, including commercial insurance, Medicaid, and Medicare. This data source provides longitudinal patient‐level data on demographics, laboratory results, provider notes, procedures, diagnoses, medications, outpatient visits, vital signs, hospitalizations, observations, and costs (from the insurance administrative claims portion). Data were standardized to the Observational Medical Outcomes Partnership Common Data Model. The data used for this study are Boehringer Ingelheim's proprietary data and were used under license for this study and, therefore, are not publicly available.

Study Population

Patients were included in the study if they met the following criteria: (1) aged ≥18 years old; (2) discharged alive with HF as the primary diagnosis (index date); (3) had at least 365 days of history and insurance coverage prior to the index date (baseline period); (4) had available left ventricular EF (LVEF) data to stratify by phenotype (LVEF: HFrEF ≤40%, HFmrEF 41%–49%, HFpEF ≥50%); (5) had at least 90 days of observation after discharge from HHF; (6) had no pregnancy‐related records during the study period; and (7) had no history of LV assist device, heart transplant, or dialysis before discharge from HHF.

Study Design

The study design is described in Figure S1. The study period was from June 2020 to September 2023 for the cohort with HFrEF and from March 2022 to September 2023 for the cohorts with HFmrEF and HFpEF, based on the earliest regulatory approval of empagliflozin or dapagliflozin for the respective HF LVEF phenotypes. SGLT2i received its first Food and Drug Administration approval for HFrEF in 2020 (dapagliflozin in May 2020, empagliflozin in August 2021). Two distinct reference time points (index date 1 and index date 2) were used, depending on the analysis objective. For evaluating changes in GDMT use before and after HHF and for time‐to‐initiation analysis, the HHF discharge date was used as index date #1. For evaluating GDMT discontinuation rates, the date of initial GDMT prescription/dispensation was used as index date #2. Follow‐up continued until death, end of data availability, or the study end, whichever occurred first.

LVEF phenotypes (HFrEF: LVEF ≤40%, HFmrEF: LVEF 41%–49%, HFpEF: LVEF ≥50%) were determined using LVEF measurements from the hospitalization period or the measurement closest to index date 1 during either the 90 days following or the 365 days before index date 1. Baseline laboratory values were measured within 180 days before and closest to hospital discharge. Other baseline variables were defined using data collected within 365 days before HHF and included sociodemographic characteristics (eg, age, sex, race), clinical characteristics and comorbidities, including history of HF diagnosis/hospitalization, type 2 diabetes (T2D), cancer, chronic kidney disease (CKD), cardiovascular diseases, hypertension, and obesity (Material S1), and concomitant drug use (Table S1). Among patients with HFrEF, GDMT use within 3 months of HHF was classified into high GDMT use (≥6 points) and low GDMT use (<6 points) based on an 8‐point scale (2 points were given for each of ARNI, BB, MRA, or SGLT2i, and 1 point was given for ACEi/ARB). Among patients with HFmrEF or HFpEF, high GDMT use was defined as any use of SGLT2i (Table S1).

Study Outcomes

Study outcomes included (1) GDMT use before and after index hospitalization, and (2) timing of GDMT initiation and discontinuation post‐discharge. GDMT use was defined as a either a clinician prescription or a dispensed medication. GDMT initiation of respective classes was defined as the earliest documentation of a prescription from the discharge date onward (regardless of GDMT use before HHF). Dual, triple, and quadruple therapies were defined by overlapping use of 2, 3, or 4 GDMT classes, respectively. Time‐to‐initiation was defined as the days from HHF discharge to GDMT initiation. Patients on GDMT before HHF were included in time‐to‐initiation analyses, with GDMT use during HHF considered as initiated at the time of discharge. GDMT discontinuation was defined as the absence of a new prescription for ≥30 days following the (estimated) end date of a previous GDMT prescription. We defined discontinuation using a 30‐day gap for our primary analysis, consistent with common practice in claims‐based persistence analyses, and we assessed robustness using a 90‐day gap in sensitivity analyses. This approach aligns with methodological guidance and prior applications in the literature. ¹⁵ , ¹⁶

Time‐to‐discontinuation was defined as the time from the first GDMT class initiation post‐HHF discharge, including discharge date, until the end date of the last prescription. Treatment discontinuation analyses were conducted on the cohort that completed a full year of follow‐up from treatment initiation.

Statistical Analysis

For baseline demographic and clinical characteristics, statistical analysis included frequency counts and proportions for categorical variables and mean ± SD for continuous variables. Time to GDMT initiation and time to GDMT discontinuation were reported as mean ± SD in days. The proportion of patients who discontinued GDMT was calculated by dividing the number of patients who discontinued by the total number of patients who initiated that therapy. Patients with HFmrEF and HFpEF were not included in the discontinuation analysis, as the vast majority of patients did not have a full year of follow‐up after SGLT2i treatment initiation.

Approximately 70% of prescription records contained incomplete information, such as missing or incorrect end dates. All prescriptions in our analysis had drug identification, fill dates, and patient identifiers; the imputation was limited to the days supply field only. To address this, prescriptions with 1‐day or missing outpatient supplies were imputed as 90‐day supplies to reflect common prescribing practices in US outpatient settings and the predominant pattern observed across HF GDMT prescriptions in the database. This imputation approach is consistent with methods used in prior claims‐based medication adherence studies. ¹⁷ , ¹⁸

Sensitivity analyses were conducted where all 1‐day supply outpatient GDMT prescriptions were imputed with 30‐day supplies versus 90‐day in the primary analysis. All data analyses were performed using a combination of ATLAS (an open‐source application developed as part of Observational Health Data Sciences and Informatics), SQL, and R.

Ethics Statement

All analyses performed in this study were conducted in accordance with data use agreement terms, as specified by the data owners. This study used de‐identified patient data from a commercial database and was therefore exempt from institutional review board approval and the requirement for informed consent under 45 CFR 46.104(d) (4). The study was carried out in compliance with the protocol, the principles of the Declaration of Helsinki, the Guidelines for Good Pharmacoepidemiology Practice, and relevant Boehringer Ingelheim Standard Operating Procedures.

RESULTS

Baseline Characteristics of the Total Population

The study included 17 210 patients with HHF who were discharged alive. Of these, 73% (n=12 588) had HFrEF, 4% (n=668) had HFmrEF, and 23% (n=3954) had HFpEF. The mean age was 69.2±13.7 years, with 60% of patients being male, 74% White, and 18% Black. Notably, 43% of patients had a body mass index ≥30 kg/m². Most patients were hospitalized for worsening chronic HF (68%), with 47% having T2D, 37% CKD, 38% atrial fibrillation, and 90% hypertension (Table 1). Patients with HFrEF were 65% male with a mean age of 68.0±13.7 years, whereas patients with HFpEF were 56% female with a mean age of 72.8±12.8 years. Obesity prevalence (body mass index ≥30 kg/m²) was 51% in HFpEF versus 41% in HFrEF, with similar patterns for T2D (50% versus 45%), CKD (43% versus 36%), atrial fibrillation (42% versus 37%), and hypertension (95% versus 88%) (Table 1).

Table 1.

Baseline Demographic and Clinical Characteristics

	Patients with HHF
	Total	HFrEF	HFmrEF	HFpEF
	17 210 (100%)	12 588 (73%)	668 (4%)	3954 (23%)
Age, y, mean±SD	69.2±13.7	68.0±13.7	70.4±13.7	72.8±12.8
Male sex, N (%)	10 361 (60%)	8197 (65%)	416 (62%)	1748 (44%)
Black race, N (%)	3101 (18%)	2333 (19%)	105 (16%)	663 (17%)
White, N (%)	12 800 (74%)	9326 (74%)	508 (76%)	2966 (75%)
Other, N (%)*	1309 (7%)	929 (7%)	55 (9%)	325 (9%)
Length of index HHF stay, mean±SD	8.8±12.7	8.9±13.6	8.5±10.2	8.5±9.8
Baseline measurements/diagnostics (within 180 d before and closest to discharge)
Urine albumin/creatinine ratio, mean±SD	420.4±1080.1	381.4±1062	389.7±1091.5	522.4±1120
Estimated glomerular filtration, mL/min/1.73 m2	61.3±24.1	61.5±24.0	61.5±23.0	60.7±24.5
BMI, kg/m2, mean±SD	30.6±8.9	30±8.5	30.5±8.9	32.2±9.9
Systolic blood pressure, mm Hg, mean±SD	124.5±18.8	122.2±18.4	129.1±18.7	130.9±18.5
Baseline comorbidities within 365 d prior
History of HF diagnosis, N (%)	11 676 (68%)	8572 (68%)	441 (66%)	2663 (67%)
Type 2 diabetes, N (%)	8023 (47%)	5722 (45%)	315 (47%)	1986 (50%)
Chronic kidney disease, N (%)	6423 (37%)	4488 (36%)	247 (37%)	1688 (43%)
Atrial fibrillation, N (%)	6545 (38%)	4599 (37%)	276 (41%)	1670 (42%)
Hypertension, N (%)	15 519 (90%)	11 135 (88%)	612 (92%)	3772 (95%)
BMI value ≥30.0 kg/m2 (within 180 d prior)	7289 (42%)	5078 (40%)	269 (40%)	1942 (49%)
Charlson Comorbidity index, mean±SD	10.1±5.3	9.7±5.4	10.4±5.1	11.2±5.2

BMI indicates body mass index; HF, heart failure; HHF, hospitalized with heart failure; HFmrEF, heart failure with mildly reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction; and HFrEF, heart failure with reduced ejection fraction.

^*Includes individuals identified as Asian as well as those with missing, unknown, or unreported race information in the underlying data source.

Baseline GDMT Use Before HHF

Among patients with HFrEF, GDMT use during the baseline period before HHF included 31% receiving no GDMT, 29% on monotherapy, 27% on dual therapy, 11% on triple therapy, and 1% on quadruple therapy. Specifically, 68% of patients received BB, 64% ACEi/ARB, 23% MRA, 14% ARNI, and 5% SGLT2i. Among patients with HFmrEF and HFpEF, only 7% of patients in each group received SGLT2i (Figure 1).

Figure 1. GDMT use before and after hospitalization for HF stratified by ejection fraction phenotype.

*Includes patients who were prescribed GDMT on the discharge date. ACEi indicates angiotensin‐converting enzyme inhibitor; ARB, angiotensin (II) receptor blocker; ARNI, angiotensin receptor–neprilysin inhibitor; BB, beta blocker; GDMT, guideline‐directed medical therapy; HF, heart failure; HFmrEF, heart failure with mildly reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; MRA, mineralocorticoid receptor antagonist; and SGLT2i, sodium–glucose cotransporter‐2 inhibitor.

GDMT Use Within 1 Year From HHF Discharge

GDMT use demonstrated a modest increase within 1 year following HHF discharge (Figure 1). Among patients with HFrEF (n=12 588), the proportion of patients receiving no GDMT decreased to 10%, whereas monotherapy use increased to 50%. The use of dual and triple therapy remained relatively stable at 28% and 11%, respectively. The proportion of patients with HFrEF receiving quadruple GDMT therapy after discharge was 2%. The use of individual GDMT also modestly increased following HHF discharge, with BB use increasing to 84%, ACEi/ARB to 72%, MRA to 38%, ARNI to 26%, and SGLT2i to 13%.

Among patients with HFmrEF and HFpEF, SGLT2i use demonstrated a modest increase following HHF discharge, with 12% and 9% receiving the therapy, respectively.

Additionally, only a limited number of patients met criteria for achieving “high GDMT” within 3 months of HHF discharge, with rates of 18% for HFrEF, 13% for HFmrEF, and 10% for patients with HFpEF.

Baseline Characteristics of Patients With any GDMT Versus No GDMT Use During Follow‐Up

Across all HF phenotypes, patients who had any GDMT use during post‐discharge follow‐up were younger, had a higher mean body mass index, shorter length of stay during index HHF, more prior HHF, greater pre‐HHF GDMT use, and higher mean urine albumin‐to‐creatinine ratio compared with those with no GDMT use (Table S2). In addition, among patients with HFmrEF and HFpEF, those with any GDMT use (ie, SGLT2i) had a higher prevalence of T2D and CKD compared with those with no GDMT use (Table S2).

Baseline Characteristics of Patients Initiating Newer GDMT Classes During Follow‐Up

Among patients with HFrEF, those initiating newer GDMT (SGLT2i or ARNI) were younger (mean age: 62.5±12.6 and 63.5±13.6 years, respectively) and more likely to be male (69% and 72%, respectively) compared with those initiating other GDMT classes (mean age 64.4–67.3 years, males: 66%–67%) (Table S3). Of note, patients with HFrEF who initiated SGLT2i had a higher prevalence of T2D (65%) compared with patients with HFrEF who initiated other GDMT classes (44%–46%). The prevalence of hypertension and atrial fibrillation among patients with HFrEF was consistently high and similar across all GDMT groups. A similar trend was observed among patients with HFpEF who initiated SGLT2i compared with those without SGLT2i initiation (Table S3).

Time to GDMT Initiation

Among patients with HFrEF, the mean time from HHF discharge to initiation of individual GDMT was as follows: 14.7 days for BB, 20.6 days for ACEi/ARB, 37.1 days for MRA, 51.6 days for ARNI, 87.8 days for SGLT2i, and 108.7 days for quadruple therapy (Figure 2, Table S4). The time to initiate SGLT2i in patients with HFmrEF and HFpEF was substantially shorter (27.6 and 28.0 days, respectively) compared with patients with HFrEF (Figure 2, Table S4). Among patients with HFrEF, stratification by history of HF and GDMT use before HHF (baseline period) revealed that those with a history of HF showed longer times to initiation across all GDMT classes compared with those with no history of HF. Additionally, any GDMT use before HHF was associated with shorter initiation time for SGLT2i compared with no GDMT use before HHF (81.3–85.5 days versus 95.7–102.5 days) (Table S5).

Figure 2. Time to GDMT initiation (days) in patients with HHF stratified by LVEF phenotypes.

ACEi, angiotensin‐converting enzyme inhibitor; ARB, angiotensin (II) receptor blocker; ARNI, angiotensin receptor–neprilysin inhibitor; BB, beta blocker; GDMT, guideline‐directed medical therapy; HHF, hospitalization for heart failure; HFmrEF, heart failure with mildly reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; and SGLT2i, sodium–glucose cotransporter‐2 inhibitor.

Patients with baseline T2D initiated SGLT2i treatment more rapidly, with a mean of 67.1 days, compared with 94.2 days for those without T2D. A similar, though less pronounced, difference was observed in patients with baseline CKD, with treatment initiation averaging 72.3 days for those with baseline CKD versus ACEi/ARB at 57%, 77.2 days for those without CKD (Table S6).

Discontinuation of GDMT Among Patients With HFrEF

The discontinuation rate was highest for ARNI at 62%, followed by MRA at 57%, ACEi/ARB at 57%, SGLT2i at 55%, and BB at 51% (Table 2).

Table 2.

GDMT Discontinuation^* Rates and Time‐to‐Discontinuation Among Patients With HFrEF

	Total	New‐onset HF	Worsening chronic HF
Number of patients in group, N (%)**	7382 (100%)	2193 (30%)	5189 (70%)
ACEi/ARB	3546 (57%)	1047 (53%)	2499 (59%)
ARNI	1333 (62%)	368 (58%)	965 (63%)
BB	3522 (51%)	976 (47%)	2546 (52%)
MRA	1844 (57%)	510 (54%)	1334 (59%)
SGLT2i	528 (55%)	144 (51%)	384 (57%)
Time‐to‐discontinuation
ACEi/ARB, mean±SD	175.8±93.1	179.5±95.0	185.2±92.2
ARNI, mean±SD	188.7±94.4	203.2±98.6	187.2±89.4
BB, mean±SD	178.7±91.9	184.9±95.7	183.9±87.0
MRA, mean±SD	176.1±93.0	181.2±97.9	180.9±89.2
SGLT2i, mean±SD	209.6±97.6	218.6±99.9	204.0±92.8

ACEi indicates angiotensin‐converting enzyme inhibitor; ARB, angiotensin (II) receptor blocker; ARNI, angiotensin receptor–neprilysin inhibitor; BB, beta blocker; GDMT, guideline‐directed medical therapy; HF, heart failure; HFmrEF, heart failure with mildly reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; MRA, mineralocorticoid receptor antagonist; and SGLT2i, sodium–glucose cotransporter‐2 inhibitor.

^*GDMT discontinuation was defined as the absence of a new prescription/dispensation for ≥30 days following the (estimated) end date of a previous GDMT prescription.

^**Treatment discontinuation analysis was conducted on the cohort that completed a full 1‐year follow‐up from the first GDMT initiation. The proportion of patients who discontinued GDMT was calculated by dividing the number of patients who discontinued by the total number of patients who initiated that therapy within 1 year.

Patients hospitalized for worsening chronic HF had higher discontinuation rates compared with those with new‐onset HF across most GDMT classes (SGLT2i: 57% versus 51%; MRA: 59% versus 54%; ACEi/ARB: 59% versus 53%; BB: 52% versus 47%) (Table 2).

Time to Treatment Discontinuation

Among patients with HFrEF who initiated therapy, treatment persistence (ie, mean number of days staying on treatment without discontinuation) was longest for SGLT2i (209.6 days), followed by ARNI (188.7 days), with comparable persistence for BB (178.7 days), ACEi/ARB (175.8 days), and MRA (176.1 days) (Table 2). When stratified by history of HF, SGLT2i and ARNI demonstrated higher persistence among those with new‐onset HF (218.6 and 203.2 days, respectively) compared with those with a history of chronic HF (204.0 and 187.2 days, respectively) (Table 2).

Clinical Outcomes

In an exploratory unadjusted analysis, patients with high versus low GDMT use at 3 months post‐discharge had numerically lower rates of 1‐year all‐cause mortality (incidence rate 86.1 versus 168.0 per 1000 person‐years) and HF events (incidence rate 124.0 versus 134.2 per 1000 person‐years) (Table S7).

Sensitivity Analysis

Sensitivity analyses examining the impact of different imputation strategies demonstrated that our main findings remained robust. When using a 30‐day imputation for missing days supply (versus 90‐day in the primary analysis), discontinuation rates ranged from 47% to 56% across GDMT classes (Tables S8). When using a 90‐day gap to define discontinuation (versus 30‐day in the primary analysis), discontinuation rates ranged from 42% to 52% (Tables S9). The relative patterns across drug classes and the overall conclusion of suboptimal GDMT persistence remained consistent regardless of the methodological approach. Additionally, GDMT use patterns stratified by baseline systolic blood pressure and renal function are presented in Table S10. Large gaps in GDMT use were found regardless of systolic blood pressure and estimated glomerular filtration rate (eGFR) group. For some GDMTs, use rates were higher among patients with lower systolic blood pressure (SBP): quadruple medical therapy for HFrEF was 2.7% among patients with SBP <120 mm Hg and 1.6% among those with SBP ≥120 mm Hg, and ARNI use was 29.4% for those with SBP <120 mm Hg versus 22.1% for those with SBP ≥120 mm Hg. Regarding eGFR groups, although use rates were modestly lower among patients with HFrEF and eGFR <60 mL/min/1.73 m², low rates of use were observed even among patients with eGFR ≥60 (40.9% on MRA, 28.6% on ARNI).

DISCUSSION

In this contemporary observational study using a large integrated electronic health record and claims database in the United States, we observed significant gaps in GDMT implementation in patients with HF. The use of GDMT was low and delayed across the LVEF spectrum, particularly for ARNI and SGLT2i classes, with a limited proportion of patients achieving high GDMT use, even months after HHF discharge. High discontinuation and low persistence rates were also observed. These findings highlight the need for targeted quality improvement strategies to both prompt GDMT initiation and mitigate discontinuation after initiation.

Despite updated guidelines and well‐established benefits, the uptake of HF GDMT remains low, particularly for newer therapies, reflecting a substantial missed opportunity to improve patient outcomes in clinical practice. In our study, approximately one third of patients with HFrEF had no GDMT use before HHF, which improved to only 10% within 1 year after HHF discharge, suggesting multiple barriers to treatment implementation in clinical practice. Only modest improvements in GDMT use within 1 year of HHF discharge were seen in dual (28%), triple (11%), and quadruple (2%) therapies, suggesting a continued disconnect between clinical evidence, guideline recommendations, and real‐world practice. HHF represents a critical leverage point when patients are engaged with the health care system and multidisciplinary teams can coordinate GDMT optimization. The STRONG‐HF (Safety, Tolerability and Efficacy of Rapid Optimization, Helped by NT‐proBNP Testing, of Heart Failure Therapies) trial demonstrated that rapid GDMT up‐titration during and immediately after hospitalization reduced the relative risk of death and HF readmission by 34% at 180 days. ¹⁹ Our findings that mean time to GDMT initiation ranged from 15 to 88 days post‐discharge represent missed opportunities during this high‐risk period when early intervention could provide maximal benefit.

Our findings align with prior studies showing low GDMT uptake across multiple settings. ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² For instance, a US retrospective study found that only 12.5% of patients with HFrEF received optimal GDMT after HHF. ⁷ Similarly, the CONNECT‐HF (Care Optimization Through Patient and Hospital Engagement Clinical Trial for Heart Failure) study reported just a 1.8% improvement in GDMT use among patients with HFrEF within 12 months after discharge. ¹¹ Furthermore, a nationwide analysis (2016–2022) of patients with HHF showed low adoption of triple/quadruple therapy (14%). ²⁰ Prior data also demonstrated that patients receiving ≥2 classes of GDMT had a reduced risk of death, ⁸ , ²⁰ HHF, ⁸ and HF readmissions. ²⁰ These findings underscore the need for early and comprehensive GDMT initiation to improve patient outcomes.

In our study, most patients with HFrEF were prescribed traditional therapies such as BB (84%) and ACEi/ARB (72%) following HHF. However, newer therapies were underused, with only 13% prescribed SGLT2i and 26% prescribed ARNI. In addition, only 12% and 9% of patients with HFmrEF and HFpEF, respectively, received SGLT2i following HHF. These findings are consistent with previous US studies showing higher use of BB (84%–86%), ACEi/ARB (73%), and MRA (37%–39%), as compared with slow and varied adoption of SGLT2i (1.5%–20%) and ARNI (17%) at HHF discharge in patients with HFrEF ²⁰ , ²¹ , ²² that did not improve even within 12 months of follow‐up. ²¹ In addition, SGLT2i were more often prescribed to patients with HFrEF with T2D (65% of SGLT2i initiators had T2D versus 42% in the no‐SGLT2i group) and less so in those with CKD (31% of SGLT2i initiators had CKD versus 36% in the no‐SGLT2i group). This is consistent with other studies showing that SGLT2i use is more associated with T2D than CKD. ²² , ²³ Hence, despite being approved for T2D, CKD, and HF, SGLT2i remain underused, representing unrealized therapeutic potential.

These baseline data provide benchmarks for quality improvement initiatives and enable tracking of implementation progress over time. Prior studies of cardiovascular guideline implementation have shown the unfortunate reality that publication of guidelines does not generally lead to sudden changes in the use of GDMTs. For example, a prior analysis examined the impact of sacubitril/valsartan use for HFrEF before and after publication of the 2016 HF guidelines giving it a Class 1 recommendation. ²⁴ The publication of the national guidelines recommending ARNI had no significant influence on the trajectory of early adoption of sacubitril/valsartan. Use of ARNI slowly increased over time, with no significant change in slope before versus after guideline publication.

Clinical trial evidence supports early GDMT benefits in patients with HF. ²⁴ , ²⁵ , ²⁶ Among patients with HFrEF, the EMPEROR‐REDUCED (Empagliflozin Outcome Trial in Patients with Chronic Heart Failure and a Reduced Ejection Fraction) and DAPA‐HF (Dapagliflozin and Prevention of Adverse Outcomes in Heart Failure) trials showed SGLT2i benefit on cardiovascular mortality, HF hospitalizations, and urgent HF visits within weeks of initiation. ²⁷ In our study, conventional GDMT such as BB and ACEi/ARB were initiated more rapidly (15–21 days) compared with newer agents, such as SGLT2i and ARNI (52–88 days), in patients with HFrEF. This trend is consistent with findings from a previous multicenter study that documented delays in the adoption of newer therapies (33 days for SGLT2i) relative to conventional treatments (18–24 days). ¹³ Delays in initiating GDMT among medically eligible patients exposes them to excess mortality and rehospitalization risks, emphasizing the need for streamlined protocols to improve rapid GDMT implementation.

Discontinuation rates were high for all GDMT in patients with HFrEF: ARNI (62%), MRA (57%), ACEi/ARB (57%), SGLT2i (55%), and BB (51%). These data are overall consistent with a real‐world multinational study conducted in the United Kingdom, the United States, and Sweden, in which discontinuation rates in the United States were 30% to 40% for ARNI, ¹³ , ¹⁴ ~40% for MRA, ¹⁴ ~55% for ACEi, ¹⁴ 33% for ARB, ¹⁴ 54% for SGLT2i (dapagliflozin), ¹³ and ~30% for BB. ¹³ , ¹⁴ Patients with chronic HF had slightly higher discontinuation (52%–63%) than patients newly diagnosed with HF (47%–58%). Discontinuation of GDMT has been associated with worse clinical outcomes. ²⁸ Moreover, blinded discontinuation of empagliflozin has been shown to increase risk of cardiovascular mortality and other adverse outcomes. ²⁹ These findings highlight the critical need to improve treatment persistence and address barriers to long‐term use of GDMT. ³⁰ , ³¹ , ³² , ³³

Patient characteristics were associated with GDMT use in this study. Our cohort was older (mean age: 69 years) with a high comorbidity burden, correlating with lower GDMT initiation, possibly due to therapeutic inertia or tolerability concerns. The differential use patterns observed between newer branded agents (ARNI, SGLT2i) and generic GDMT (ACEi, BBs, spironolactone) likely reflect, in part, cost‐related barriers. Sukumar et al. demonstrated that medication cost is a major determinant of GDMT nonadherence, particularly for newer therapies. ³⁴ Understanding the relative contribution of clinical factors versus financial barriers to GDMT underuse requires dedicated pharmacoeconomic analyses beyond the scope of this descriptive study. Prior studies also showed that younger patients tend to receive more intensive GDMT. ³⁵ SGLT2i prescribing patterns varied by age, sex, and comorbidity, consistent with data from the Veterans Affairs health system. ³⁶ GDMT underuse was observed across BP and renal function strata (Table S9), indicating that these clinical factors alone do not explain the low use rates. These data highlighting significant gaps in GDMT even among patients with robust SBP and eGFR are consistent with prior real‐world studies ³⁷ , ³⁸ , ³⁹ . These findings highlight the need for targeted education on GDMT benefits, especially for older adults.

LIMITATIONS

Limitations of this study should be acknowledged. First, due to the later approval of SGLT2i for HFpEF (early 2022), the vast majority of patients with HFmrEF and HFpEF did not have a full 1‐year follow‐up period from the initiation of GDMT, limiting the ability to analyze treatment discontinuation in these groups. Second, the need to impute days supply for ~70% of prescriptions represents a limitation inherent to claims databases. However, our sensitivity analyses demonstrated that the finding—high discontinuation rates—remained robust across different imputation assumptions, with discontinuation rates consistently exceeding 40% regardless of imputation strategy (Tables S7 and S8). As a result, time‐to‐discontinuation may be affected by the imputation approach. However, the use of a 30‐day gap to define discontinuation may have partially mitigated this limitation, as it represents a more conservative threshold compared with the 60‐ or 90‐day gaps used in other studies. Third, EF was assessed pragmatically, based on data from the HHF period, within 3 months post‐HHF, or up to 12 months before HHF. Therefore, the recorded EF may not accurately reflect the patient's status at the time of HHF discharge, leading to a possible misclassification in HF phenotypes. Fourth, although the Optum Market Clarity database integrates claims and electronic health records data, providing more clinical context than claims alone, specific reasons for medication discontinuation (eg, patient‐reported side effects, clinical decision‐making rationale, laboratory abnormalities, patient preferences) are not systematically captured in structured data fields. Free‐text clinical notes, which might contain such information, are not available in the deidentified data set. Likewise, data regarding patient‐level out‐of‐pocket medication costs or insurance coverage were not available. Although there were significant gaps in even generic inexpensive GDMTs, the degree to which patient‐level out‐of‐pocket expenses, copayment tiers, or prior authorization requirements influenced the observed prescribing decisions and medication persistence is unclear. Future studies specifically designed to examine clinician‐ and patient‐reported reasons for GDMT nonprescription and discontinuation are warranted.

CONCLUSIONS

In this study of patients hospitalized for HF in contemporary US clinical practice, there were persistent and large gaps in GDMT use before and after hospitalization. Only 2% of patients with HFrEF received quadruple medical therapy at any point during the 1‐year post‐discharge period, with SGLT2i use being the lowest compared with other GDMT. Even when GDMT was initiated, it was often substantially delayed (with shorter delays for BB and ACEi and longer delays for ARNI and SGLT2i), or was followed by subsequent discontinuation. These findings highlight the urgent need for targeted strategies to address initiation and treatment persistence barriers. Further efforts are needed to ensure that HHF is leveraged as an opportunity to initiate and optimize GDMT more effectively and rapidly.

Sources of Funding

This study was funded by the Boehringer Ingelheim & Eli Lilly and Company Alliance.

Conflict of Interest/Disclosures

Stephen J. Greene has received research support from the American Heart Association, Amgen, AstraZeneca, Boehringer Ingelheim, Bristol Myers Squibb, Cytokinetics, Merck, Novartis, Otsuka, Pfizer, and Sanofi; has served on advisory boards or as consultant for Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol Myers Squibb, Corcept Therapeutics, Corteria Pharmaceuticals, CSL Vifor, Cytokinetics, Idorsia, Lilly, Lexicon, Merck, Mineralys, Novo Nordisk, Otsuka, Recordati, Roche Diagnostics, Sanofi, scPharmaceuticals, Sumitomo, Tricog Health, and Viatris. Niklas Schmedt, Juergen H. Prochaska, Phuong Tang, and Ayman Alhamdow are employees of Boehringer Ingelheim. Christian Carlsen is an employee of Eli Lilly and Company. Milou Brand, Eleanor Davies, Atif Adam, and Laurence Sophie Jouaville Abrouk are employees of IQVIA.

The authors meet the criteria for authorship, as recommended by the International Committee of Medical Journal Editors. The authors did not receive payment related to the development of the article. Boehringer Ingelheim was given the opportunity to review the article for medical and scientific accuracy, as well as intellectual property considerations.

Supporting information

Tables S1–10

Figure S1