Dapagliflozin in Patients Hospitalized for Heart Failure: Primary Results of the DAPA ACT HF-TIMI 68 Randomized Clinical Trial and Meta-Analysis of Sodium-Glucose Cotransporter-2 Inhibitors in Patients Hospitalized for Heart Failure.
Aug 29, 2025DOI: 10.1161/circulationaha.125.076575Journal ArticleAHA
Written by: David D BergSiddharth M PatelPaul M HallerAbby L CangeMichael G PalazzoloAndrea BellaviaJulia F KuderAkshay S DesaiSilvio E InzucchiJohn J V McMurrayEileen O'MearaSubodh VermaJan BělohlávekJarosław DrożdżBéla MerkelyModele O OgunniyiTomáš DrasnarJoseph L IzzoBalazs SarmanJohn E McGintyKrishnan RamanathanAngel J MulkayAndrzej PrzybylskiChristian T RuffMichelle L O'DonoghueSabina A MurphyMarc S SabatineStephen D WiviottDAPA ACT HF-TIMI 68 Trial Committees and Investigators
Find these authors on Roon
MO
Dr. Modele Ogunniyi
General Cardiology · Emory University
CR
Dr. Christian Ruff
General Cardiology · PHS
Introduction
Hospitalization for heart failure (HF) is the leading cardiovascular reason for hospital admission globally,1 and rates have continued to climb over the last decade.2 Patients hospitalized for HF have a high risk of death and other adverse outcomes during hospital admission and in the early post-discharge period,2 highlighting the need for effective treatment strategies focused on this phase of care.3, 4 Evidence from randomized trials suggests that initiating and optimizing disease-modifying chronic heart failure therapies, including angiotensin receptor-neprilysin inhibitors, beta-blockers, and mineralocorticoid receptor antagonists, during heart failure hospitalization and in the early post-discharge period may improve both short- and long-term outcomes.3, 4Sodium-glucose cotransporter-2 inhibitors (SGLT2i), including dapagliflozin, are indicated for the treatment of HF regardless of left ventricular ejection fraction (LVEF).5, 6 However, there are limited data regarding the efficacy and safety of initiating SGLT2i during HF hospitalization.7, 8 In the SOLOIST-WHF trial (n=1222), which enrolled patients with type 2 diabetes and a recent episode of worsening chronic HF, sotagliflozin significantly reduced total cardiovascular deaths and worsening HF events over a median follow-up of 9 months;9 however, only 596 patients were randomized on or before their date of discharge (76% on the date of discharge).8 In the EMPULSE trial (n=530), in-hospital initiation of empagliflozin in patients hospitalized for HF significantly improved a hierarchical composite outcome that included clinical events and HF-related health status as measured by the Kansas City Cardiomyopathy Questionnaire-23 item (KCCQ-23) through 90 days, though the result was driven by improvements in the latter.7Data from trials of outpatients with HF suggest that the cardioprotective effects of SGLT2i emerge early, supporting prompt initiation of these therapies.10–12 Nevertheless, hospitalized HF patients may be more susceptible to therapy-related complications due to hemodynamic and rhythm instability, fluctuating kidney function, and concomitant addition or intensification of other HF therapies.13 We, therefore, designed the DAPA ACT HF-TIMI 68 (Dapagliflozin Effect on Cardiovascular Events in Acute Heart Failure – Thrombolysis in Myocardial Infarction 68) trial to test the hypothesis that in-hospital initiation of dapagliflozin, as compared with placebo, can safely and effectively decrease the early risk of cardiovascular death or worsening HF among patients hospitalized for HF. In addition, we aimed to combine data from DAPA ACT HF-TIMI 68 with other placebo-controlled trials of in-hospital initiation of SGLT2i in hospitalized HF patients to generate pooled estimates of the treatment effects of SGLT2i on early efficacy outcomes in this clinical setting.
Methods
The DAPA ACT HF-TIMI 68 trial design has been reported previously.14 Briefly, DAPA ACT HF-TIMI 68 was an investigator-initiated, multinational, randomized, double-blind, placebo-controlled trial of in-hospital initiation of dapagliflozin in patients hospitalized for HF. The trial was designed by the TIMI Study Group at Brigham and Women’s Hospital and Harvard Medical School (Boston, MA), in collaboration with the trial steering committee (Supplemental Material). Worldwide Clinical Trials (contract research organization) provided data management services. The safety of patients in the trial was overseen by an independent data monitoring committee. The protocol and amendments were approved by the relevant institutional review board or ethics committee at each participating site, and all participants provided written informed consent. All statistical analyses were conducted by biostatisticians at the TIMI Study Group in accordance with the statistical analysis plan (Supplemental Material). The first draft of the manuscript was written by the first author, and all authors participated in subsequent revisions. The steering committee made the decision to submit the manuscript for publication and assumes responsibility for the accuracy and completeness of the data and analyses. The manuscript adhered to the 2025 CONSORT reporting guidelines. Data will not be made publicly available; however, we encourage parties interested in collaboration and data sharing to contact the corresponding author directly for further discussion.
Eligible patients were at least 18 years of age and were currently hospitalized with a primary diagnosis of HF, including signs and symptoms of fluid overload. The trial was originally designed to enroll only patients with LVEF ≤40%; however, following publication of the EMPEROR-Preserved trial,15 which demonstrated the efficacy of the SGLT2i empagliflozin in ambulatory patients with HF and LVEF >40%, the protocol was amended in November 2021 to include patients with LVEF >40%. Patients were required to have elevated natriuretic peptide levels (either N-terminal pro-B-type-natriuretic peptide [NT-proBNP] or B-type-natriuretic peptide [BNP]) during the index hospitalization with varying thresholds for inclusion according to LVEF and cardiac rhythm. Patients were randomized while still hospitalized, at least 24 hours and no later than 14 days after hospital admission, and as early as possible following initial stabilization. Stabilization criteria included no intensification in the dose of intravenous diuretics during the 12 hours prior to randomization and no use of intravenous vasodilators or inotropes during the 24 hours prior to randomization. A complete list of eligibility criteria is provided in the Supplemental Material.
Eligible patients were randomly assigned in double-blind fashion to receive dapagliflozin 10 mg once daily or matching placebo, in accordance with a sequestered, fixed randomization schedule, with the use of balanced blocks by site to ensure an approximate 1:1 allocation ratio. Randomization was stratified based on diagnosis of type 2 diabetes and history of HF prior to the index hospitalization. Open-label use of SGLT2i was prohibited during the randomized treatment period; use of all other pharmacological HF therapies at baseline and throughout the trial was at the discretion of the treating physician. Randomized treatment was administered for 60 days, with follow-up visits scheduled at 7, 30, and 60 days after randomization. HF-related health status using the KCCQ-12 item (KCCQ-12) was assessed at randomization and the final study visit.
The primary efficacy outcome was a composite of time to cardiovascular death or first worsening HF event. A worsening HF event was defined as any event meeting one of the following criteria: (1) worsening signs or symptoms of HF during the index hospitalization resulting in treatment with intravenous inotropes, mechanical circulatory support, invasive ventilatory support, or heart transplantation; (2) unplanned rehospitalization for HF; or (3) urgent ambulatory HF visit resulting in treatment with intravenous diuretic therapy. A clinical events committee (CEC) led by the TIMI Study Group, whose members were blinded to randomized treatment, adjudicated all components of the primary outcome (Supplemental Material).
Secondary efficacy outcomes included a composite of time to cardiovascular death, rehospitalization for HF, or urgent HF visit; a composite of time to cardiovascular death or rehospitalization for HF; a composite of time to rehospitalization for HF or urgent HF visit; each individual component of the primary efficacy outcome; and time to death from any cause. An additional secondary outcome included a hierarchical composite of time to cardiovascular death, number of worsening HF events, time to first worsening HF event, and change from baseline in KCCQ-12 total symptom score (TSS).
Key safety outcomes included symptomatic hypotension leading to hospitalization or study drug discontinuation and worsening kidney function resulting in at least a doubling of serum creatinine, hospitalization, study drug discontinuation, dialysis, or renal death. Other prespecified safety outcomes included major hypoglycemia, diabetic ketoacidosis, adverse events leading to study drug discontinuation, and serious adverse events considered related to study drug by the investigator.
The primary efficacy analysis was based on time from randomization to the first occurrence of any element of the primary composite outcome. We estimated that a sample size of 2400 patients would provide the trial with 80% power to detect a relative reduction of 27% in the risk of the primary outcome with dapagliflozin as compared with placebo at two months. The estimated treatment effect was based on the DAPA-HF trial of dapagliflozin in ambulatory patients with HF with reduced ejection fraction (HFrEF) as well as a subgroup analysis of DAPA-HF patients who had been hospitalized for HF in the preceding year; the relative risk reductions were 26% and 36% in the overall trial and subgroup analysis, respectively.10, 16 The sample size calculation assumed an event rate of 16% in the placebo group, a dropout rate of 0.5%, and a two-sided alpha level of 0.05. All efficacy analyses were conducted on an intention-to-treat basis.
The primary outcome was assessed using the log-rank test, with stratification based on diagnosis of type 2 diabetes and history of HF prior to the index hospitalization. The cumulative incidence of the primary outcome was estimated using the Kaplan-Meier method. A stratified Cox proportional-hazards model was used to estimate the treatment effect using hazard ratios (HR) and 95% confidence intervals (CI). The proportional-hazards assumption was assessed and confirmed graphically and tested with the use of Schoenfeld residuals. We assessed the consistency of the treatment effect on the primary outcome in prespecified subgroups using the Cox model. The primary outcome was also analyzed using a Bayesian survival model (Supplemental Material).
Secondary time-to-event outcomes were evaluated using stratified Cox proportional-hazards models. The secondary hierarchical composite outcome was evaluated using an unmatched win ratio approach.17 Change in KCCQ-TSS from baseline to two months in surviving patients was compared using an analysis of covariance model, adjusting for baseline value. Safety analyses were performed in patients who had undergone randomization and received at least one dose of study drug. Chi-square or Fisher’s exact test was used to compare the incidence of each adverse event. Analyses were performed using SAS v9.4 and R v4.2.3.
Since DAPA ACT HF-TIMI 68 was the final planned randomized trial of in-hospital initiation of an SGLT2i, a prespecified meta-analysis was performed of randomized trials (or subgroup analyses from randomized trials) evaluating initiation of SGLT2i in patients hospitalized for HF to summarize the totality of the evidence base. Detailed methods are provided in the Supplemental Material. The main outcomes were: (1) cardiovascular death or first worsening HF event, and (2) all-cause death. Random effects models with inverse-variance weighting were used to generate estimates for the treatment effect of SGLT2i on each outcome. Between-trial heterogeneity of treatment effect was examined using Cochran’s Q test.
Results
From September 2020 through March 2025, a total of 2401 patients underwent randomization at 210 sites in 5 countries (Figure 1), of whom 1218 were assigned to receive dapagliflozin and 1183 to receive placebo. A total of 2389 patients received at least one dose of study drug and were included in the safety analyses. A total of 262 (11.0%) patients discontinued study drug prematurely, including 139 (11.5%) in the dapagliflozin group and 123 (10.4%) in the placebo group. In total, 25 (2.1%) patients in the dapagliflozin group and 25 (2.1%) in the placebo group received at least one dose of an open-label SGLT2i during trial follow-up; however, only 9 (0.7%) and 6 (0.5%) in the dapagliflozin and placebo groups, respectively, permanently stopped study drug to start an open-label SGLT2i. Rates of withdrawal of consent (13 patients, 0.5%) and loss to follow-up with unknown vital status (4 patients, 0.2%) were low (Figure 1).
The median time from hospital admission to randomization was 3.6 days (interquartile range [IQR], 2.1–5.4 days). At the time of randomization, reported signs and symptoms of congestion were prevalent, including 39.8% with elevated jugular venous pressure, 36.4% with rales on auscultation of the lungs, and 61.8% with peripheral edema. Baseline characteristics were well balanced between the treatment groups (Table 1). The median age was 69 years (IQR, 58–77 years); 815 (33.9%) were women, 448 (18.7%) were Black, 852 (35.5%) had type 2 diabetes, and 1717 (71.5%) had a LVEF ≤40%. The index hospitalization was the first presentation of HF in 1074 (44.7%) patients. Of the remaining 1327 (55.3%) patients with worsening chronic HF, 742 (55.9%) had at least one prior HF hospitalization. At screening, the median NT-proBNP concentration (n=1582) was 4803 pg/ml (IQR, 2812–8768 pg/ml) and the median BNP concentration (n=830) was 1106 pg/ml (IQR, 717–1859 pg/ml). At randomization, the median systolic blood pressure was 119 mmHg (IQR, 108–133 mmHg), and the median eGFR was 63 ml/min/1.73m2 (IQR, 48–82 ml/min/1.73m2). At randomization, 82.5% of patients were on a beta-blocker, 70.1% were on a renin-angiotensin system inhibitor, including 27.0% on an angiotensin receptor–neprilysin inhibitor (ARNI), and 48.6% were on a mineralocorticoid receptor antagonist (MRA). At discharge, 88.2% were on a beta-blocker, 75.5% were on a renin-angiotensin system inhibitor, including 30.6% on an ARNI, and 54.4% were on an MRA. The median time from randomization to hospital discharge was 2 days (IQR, 1–4 days).
The primary outcome occurred in 133 patients (10.9%) in the dapagliflozin group and in 150 patients (12.7%) in the placebo group (HR, 0.86; 95% CI, 0.68–1.08; p=0.20) (Table 2 and Figure 2A). There was no heterogeneity in the treatment effect of dapagliflozin on the primary outcome across prespecified subgroups (Figure 3). The results of the Bayesian analysis are presented in the Supplemental Material.
Most worsening HF events were rehospitalizations for HF, which occurred in 90 patients (7.4%) in the dapagliflozin group, as compared with 102 patients (8.6%) in the placebo group (HR, 0.85; 95% CI, 0.64–1.13) (Table 2 and Figure 2B). Cardiovascular death occurred in 30 patients (2.5%) and 37 patients (3.1%) in the dapagliflozin and placebo groups, respectively (HR, 0.78; 95% CI, 0.48–1.27) (Table 2 and Figure 2C). Death from any cause occurred in 36 patients (3.0%) and 53 patients (4.5%) in the dapagliflozin and placebo groups, respectively (HR, 0.66; 95% CI, 0.43–1.00) (Table 2 and Figure 2D).
The difference in the least-squares mean change in KCCQ-12 TSS from baseline to two months between groups was +1.3 (95% CI, −0.7 to +3.3). For the secondary hierarchical composite outcome, dapagliflozin was superior in 42.3% of paired comparisons, placebo was superior in 38.7%, and there were 19.0% ties, yielding a win ratio of 1.08 (95% CI, 0.98–1.20) (Table 2).
Symptomatic hypotension leading to hospitalization or study drug discontinuation occurred in 43 patients (3.6%) in the dapagliflozin group and 26 patients (2.2%) in the placebo group. A worsening kidney function event occurred in 71 patients (5.9%) in the dapagliflozin group and 55 patients (4.7%) in the placebo group (Table 3). Rates of major hypoglycemia were very low and did not differ between groups. There were no cases of diabetic ketoacidosis. The rate of permanent discontinuation of study drug owing to an adverse event also did not differ between treatment groups.
Data from three trials were selected for inclusion in the meta-analysis. These studies tested dapagliflozin (DAPA ACT HF-TIMI 68),18 empagliflozin (EMPULSE),7 and sotagliflozin (subgroup analysis of patients initiated on study drug on or before the date of discharge in SOLOIST-WHF).8 Collectively, these 3 studies included 3,527 patients hospitalized for HF (Supplemental Material).7, 8 A total of 447 patients experienced cardiovascular death or a worsening HF event, and 153 patients died. In the meta-analysis, SGLT2i appeared to reduce the early risk of cardiovascular death or worsening HF by 29% (HR 0.71, 95% CI 0.54–0.93; p=0.012) and the risk of all-cause death by 43% (HR 0.57, 95% CI 0.41–0.80; p=0.001) (Figure 4). There was no between-trial heterogeneity of treatment effect for either outcome (Q=3.26 [p=0.20] and Q=1.40 [p=0.50], respectively).
Discussion
In this cardiovascular outcomes trial of patients hospitalized with a primary diagnosis of HF, in-hospital initiation of dapagliflozin did not significantly reduce the risk of cardiovascular death or worsening HF over two months. However, when the results of the DAPA ACT HF-TIMI 68 trial are considered in the context of other randomized trials evaluating in-hospital initiation of SGLT2i, the totality of evidence suggests a potential clinical benefit of this therapeutic class for reducing the early risks of cardiovascular death or worsening HF and of all-cause mortality when initiated in the inpatient setting.
In the SOLOIST-WHF trial, 596 patients were initiated on study drug on or before the date of discharge from their index HF hospitalization. In a post hoc analysis of this subgroup, sotagliflozin significantly reduced the composite of cardiovascular death or worsening HF (HR 0.54, 95% CI 0.35–0.82), and of all-cause mortality (HR 0.39, 95% CI 0.17–0.88) through 90 days.9 In the EMPULSE trial (n=530), all patients were initiated on study therapy (empagliflozin vs. placebo) while hospitalized for either newly diagnosed or worsening chronic HF, with a treatment duration of 90 days.7 Although EMPULSE was not powered to assess the impact of in-hospital initiation of SGLT2i therapy on clinical outcomes during and immediately following HF hospitalization, there were numerically lower rates of cardiovascular death or worsening HF (HR 0.69, 95% CI 0.45–1.08) and of all-cause mortality (HR 0.50, 95% CI 0.24–1.03) in patients randomly allocated to empagliflozin.
These studies had several important limitations, including modest sample sizes and lack of CEC adjudication of investigator-reported clinical events. Moreover, the short-term outcomes data for hospitalized HF patients from SOLOIST-WHF were based on a post hoc subgroup analysis, which carries a higher risk of bias compared to a full randomized trial population. In addition, there were important differences in the study populations, study designs, and follow-up durations between these two studies and the DAPA ACT HF-TIMI 68 trial. Nevertheless, these two trials contributed to a pre-trial likelihood of clinical benefit from the strategy of in-hospital initiation of SGLT2i. In a meta-analysis using all relevant trial data, there was a significant reduction in the early risk of cardiovascular death or worsening HF with SGLT2i treatment compared to placebo, with no evidence of between-trial heterogeneity in the treatment effect and with highly consistent event rates across studies. Still, given the limitations of the prior trials, which may have contributed to overly optimistic treatment effect estimates, as well as the neutral primary result of the largest and most rigorous trial contributing to the meta-analysis, we must be cautious about the magnitude of any clinical benefit from this strategy. Indeed, it may be that the true treatment effect of SGLT2i in this clinical setting is closer to the treatment effect estimate observed in DAPA ACT HF-TIMI 68.
Several factors may have contributed to the non-significant treatment difference on the primary outcome in our trial. In contrast to EMPULSE, we designed DAPA ACT HF-TIMI 68 to have a very short two-month follow-up duration given concerns about the equipoise of a placebo control beyond that time frame in the setting of data from chronic HF trials. It is possible that the risk of worsening HF events is less modifiable and related more to patient-specific factors or models of healthcare delivery in the first month following HF hospitalization. The same pattern was observed in EMPULSE with the largest treatment effect on worsening HF events emerging between 60- and 90-days post-randomization.7 In addition, to consider the impact of SGLT2i on in-hospital outcomes, we included worsening HF during the index hospitalization as a component of the primary composite outcome. Although there were very few such events in our trial, there was no suggestion that dapagliflozin reduced the risk of this outcome. Finally, there were fewer than expected primary outcome events, likely due to the high proportion of newly diagnosed HF patients, which reduced the overall power to detect a significant difference in the primary outcome.
Notably, the strongest potential efficacy signal in DAPA ACT HF-TIMI 68 as well as the other two trials of in-hospital initiation of SGLT2i was the early reduction in the risk of all-cause death. The effect on mortality appears to emerge very early, suggesting that in-hospital initiation has the potential to maximize this benefit.
We observed a modestly higher risk of symptomatic hypotension leading to hospitalization or study drug discontinuation with dapagliflozin. These data suggest that in more vulnerable populations, care should be taken to minimize other potential drivers of symptomatic hypotension, such as volume depletion or rapid escalation in the dosing of neurohormonal antagonists, when initiating SGLT2i. The rate of worsening kidney function events was low but numerically higher with dapagliflozin, consistent with the known initial decline in eGFR with SGLT2i before the long-term kidney protective effects become apparent.19, 20 Reassuringly, the rate of major hypoglycemia was low, and there were no cases of diabetic ketoacidosis. Overall, these data are highly consistent with the known safety profile of the drug class.
Conclusions
In-hospital initiation of dapagliflozin did not significantly reduce the risk of cardiovascular death or worsening HF over two months in patients admitted for HF in the DAPA ACT HF-TIMI 68 trial. However, the totality of randomized clinical trial data suggests that in-hospital initiation of SGLT2i may reduce the high risk of clinical events, including all-cause mortality, in the early post-discharge period.
Supplementary Material
Online Supplemental Data
CONSORT checklist
MO
Dr. Modele Ogunniyi
Cardiology · General Non-Invasive CardiologyEmory Universitymodele