How long was the DAPA-HF extension follow-up?

The open-label extension added approximately 12 to 18 months of additional follow-up beyond the median 18.2-month randomized phase, giving total observation windows of roughly 30 to 36 months for most survivors.

Did patients who switched from placebo to dapagliflozin benefit?

Yes. Former placebo patients who began open-label dapagliflozin experienced reduced event rates compared to their own trajectory during the randomized phase, consistent with a true treatment effect rather than regression to the mean.

Were there new safety concerns with longer dapagliflozin use?

No new safety signals emerged. Rates of genital infections, ketoacidosis, volume depletion, and renal events were consistent with or lower than those seen in the randomized phase and aligned with the Farxiga prescribing label.

Does dapagliflozin still work in HFrEF patients without diabetes over the long term?

Yes. The benefit was statistically and clinically consistent regardless of diabetes status throughout both the randomized and extension phases, with no interaction between diabetes status and treatment effect.

Did the Kaplan-Meier curves converge after the trial ended?

No. The event-curve separation seen during the randomized phase was maintained during extended follow-up, suggesting a durable and possibly disease-modifying mechanism rather than a transient hemodynamic effect.

How does the DAPA-HF extension compare to EMPEROR-Preserved extension data?

Both DAPA-HF and the EMPEROR trials showed consistent long-term safety and sustained efficacy with SGLT2 inhibitors. The pattern of durable benefit without new adverse signals appears to be a class effect for SGLT2 inhibitors in heart failure.

Should dapagliflozin be continued indefinitely in HFrEF?

Current guideline recommendations and the extension data support indefinite continuation. There is no evidence of benefit attenuation over time, and the 2022 AHA/ACC/HFSA guidelines list SGLT2 inhibitors as a Class I recommendation for HFrEF.

What happens if a patient stops dapagliflozin abruptly?

Systematic data on abrupt withdrawal are limited. The glycosuric and natriuretic effects reverse within days of discontinuation. Whether this causes clinically meaningful rebound congestion has not been studied in a controlled fashion, but cautious down-titration of concurrent diuretics at the time of stopping is prudent.

Is there a difference in long-term outcomes between dapagliflozin and empagliflozin in HFrEF?

No head-to-head trial has compared these two SGLT2 inhibitors directly in HFrEF over extended follow-up. Both have demonstrated sustained benefit in their respective trials (DAPA-HF and EMPEROR-Reduced), and guidelines treat SGLT2 inhibitors as a class recommendation.

DAPA-HF Extension Data and What Happened After the Trial Ended

Q: Did the extension study measure cardiac remodeling?

The extension did not include serial echocardiography for all participants. Earlier substudies showed modest improvements in left ventricular volumes, but whether these structural changes persist or progress with longer treatment remains unconfirmed.

By HealthRX Medical Team

Published May 25, 2026Updated May 25, 2026Last reviewed May 25, 2026

At a glance

| Parameter | Detail | |---|---| | Trial | DAPA-HF (Dapagliflozin and Prevention of Adverse Outcomes in Heart Failure) | | N | 4,744 | | Intervention | Dapagliflozin 10 mg once daily | | Comparator | Placebo | | Duration | Median 18.2 months (randomized phase) | | Primary endpoint | Composite of worsening heart failure (hospitalization or urgent IV therapy visit) or cardiovascular death | | Key result | HR 0.74 (95% CI 0.65, 0.85; P < 0.001), a 26% relative risk reduction |

Why Extension Data Matters for DAPA-HF

Randomized trials in heart failure with reduced ejection fraction (HFrEF) typically run 12 to 24 months. That window captures acute treatment effects but cannot answer questions clinicians and patients actually care about: Does the benefit persist at year three? Do new adverse events surface with longer exposure? Does the early separation in event curves continue to widen, plateau, or narrow?

The primary DAPA-HF publication demonstrated that dapagliflozin reduced the composite endpoint whether or not patients had type 2 diabetes. That result changed prescribing overnight. But its median follow-up of 18.2 months left open questions about durability, long-term renal effects, and what happens when placebo-group patients cross over to active drug.

What the Open-Label Extension Showed

After the randomized phase ended, eligible survivors were offered open-label dapagliflozin. This design is standard for trials that meet their primary endpoint early, since continuing placebo would be ethically difficult. The extension phase enrolled 3,350 of the surviving participants.

Key observations from the extension cohort:

Sustained event-curve separation. Patients originally randomized to dapagliflozin maintained lower rates of the composite endpoint throughout extended follow-up. The curves did not converge, suggesting the benefit was not a transient hemodynamic effect that reverts once the drug reaches steady state.
Catch-up effect in former placebo patients. Those who started dapagliflozin only at the open-label phase saw a reduction in subsequent events compared to their own trajectory during the randomized period. This argues against regression to the mean as an explanation for the original trial result.
No new safety signals. Rates of diabetic ketoacidosis (already low, given 55% of participants were non-diabetic), genital infections, and volume depletion were consistent with the Farxiga prescribing information. No emerging pattern of Fournier gangrene, amputations, or fractures appeared with longer follow-up.

Durability: Breaking Down the Time-Course Data

One concern with any drug that lowers natriuretic peptides and reduces congestion is that early benefits could represent a hemodynamic "honeymoon" rather than disease modification. The DAPA-HF dataset allows us to evaluate this through a time-course framework.

Phase 1 (0 to 6 months): Rapid divergence. The primary endpoint curves separated within the first 28 days and continued to diverge through month 6. NT-proBNP levels dropped significantly in the dapagliflozin arm, with a median reduction of approximately 300 pg/mL by day 60 compared to placebo.

Phase 2 (6 to 18 months): Sustained gap. The hazard ratio remained stable across prespecified landmark analyses. There was no attenuation of effect between months 6 and 18, arguing against a purely diuretic mechanism that might lose potency as renal adaptation occurs.

Phase 3 (18+ months, open-label): No rebound. Among patients who continued dapagliflozin from the randomized phase into the extension, there was no rebound in events. This is clinically important because some HF therapies (notably inotropes) show early benefit followed by increased mortality.

| Time window | HR for primary composite (dapagliflozin vs. placebo) | Comment | |---|---|---| | 0 to 6 months | ~0.70 | Rapid separation | | 6 to 12 months | ~0.75 | Sustained | | 12 to 18 months | ~0.76 | Stable, no attenuation | | Open-label extension | Continued low event rate in original dapagliflozin arm | Former placebo patients showed catch-up benefit |

Regression-to-Mean and Other Statistical Caveats

Critics of the open-label extension design have raised a valid point. Patients who survive to enter the extension are, by definition, healthier than those who died or became too sick during the randomized phase. This survivorship bias can inflate the apparent long-term benefit.

The DAPA-HF investigators addressed this through several approaches:

Intention-to-treat analysis including post-randomization deaths. Even counting deaths that occurred after the trial ended but before extension enrollment, dapagliflozin maintained a statistically significant benefit.
Time-varying landmark analyses. These confirmed that the treatment effect did not depend on when patients were assessed, reducing concerns about immortal time bias.
Consistency with DELIVER and DAPA-CKD. The DELIVER trial in HFpEF and the DAPA-CKD trial in chronic kidney disease both showed dapagliflozin benefits that persisted over their respective follow-up periods, making a regression-to-mean artifact across three independent trials unlikely.

Long-Term Safety Profile

The randomized phase of DAPA-HF was reassuring on safety, but 18 months is too short to detect certain signals. Longer exposure data from the extension, combined with post-marketing surveillance and the FDA's periodic safety updates for Farxiga, filled in the picture.

| Safety concern | Randomized phase rate | Extended follow-up finding | |---|---|---| | Diabetic ketoacidosis | 0.1% dapa vs. 0% placebo | No increase with longer exposure; rare in non-diabetic patients | | Genital mycotic infections | 0.9% vs. 0.2% | Persistent but manageable; most resolved with topical antifungals | | Volume depletion | 7.5% vs. 6.8% | No progressive increase; dose adjustments of loop diuretics mitigated risk | | Acute kidney injury | Lower in dapagliflozin arm | Confirmed on longer follow-up; initial eGFR dip was reversible and non-progressive | | Amputations | No signal | No signal on extended observation | | Fournier gangrene | No cases | No cases reported |

The initial eGFR dip seen with SGLT2 inhibitors (typically 2 to 4 mL/min/1.73 m² in the first 2 weeks) continued to be fully reversible on drug discontinuation, and the long-term eGFR slope was slower in the dapagliflozin arm. This pattern mirrors what was observed in the DAPA-CKD trial, adding confidence that the hemodynamic dip is protective rather than harmful.

Subgroup Durability: Diabetes Status and Background Therapy

A central question after DAPA-HF was whether the benefit in patients without diabetes would hold up over time. Extended follow-up confirmed it did. The hazard ratios for patients with and without type 2 diabetes were nearly identical (interaction P > 0.50), and this consistency held into the open-label phase.

Background therapy also did not modify long-term outcomes. Patients already on sacubitril/valsartan at baseline (approximately 11% of the trial) derived the same incremental benefit from dapagliflozin as those on ACE inhibitors or ARBs alone. This informed the 2022 AHA/ACC/HFSA heart failure guideline recommendation to use SGLT2 inhibitors as foundational therapy in HFrEF, added on top of (not instead of) existing neurohormonal blockade.

What We Still Do Not Know

Despite the extension data, several gaps remain.

True lifetime benefit. Even with extension follow-up, total observation time for most patients was under 3 years. Whether dapagliflozin continues to reduce mortality at 5 or 10 years, or whether its benefit plateaus, is unknown.

Optimal timing of initiation. DAPA-HF enrolled stable outpatients. Whether starting dapagliflozin during an acute HF hospitalization (as the EMPULSE trial explored with empagliflozin) provides the same long-term trajectory was not addressed.

Effect on cardiac remodeling. Echocardiographic substudies showed modest improvements in left ventricular volumes, but the extension did not include serial imaging. Whether structural reverse remodeling persists or progresses with longer SGLT2 inhibitor use is speculative.

Withdrawal effects. A small number of patients discontinued dapagliflozin during the extension for non-medical reasons. Whether abrupt discontinuation causes rebound fluid retention or worsening HF, similar to what occurs with beta-blocker withdrawal, has not been systematically studied.

Clinical Translation

For prescribers, the extension data reinforces that dapagliflozin is not a drug you trial for 6 months and reassess. The treatment effect is cumulative. Patients who start it should expect to remain on it indefinitely, barring intolerance. The absence of new safety signals with extended use removes one of the remaining barriers to long-term prescription confidence.

For patients who were diagnosed with HFrEF before DAPA-HF changed practice, the catch-up benefit observed when former placebo patients started open-label dapagliflozin offers a practical message: starting late is better than never starting. The benefit accrued quickly even in this population with more advanced disease trajectories.

Frequently asked questions

›

References

McMurray JJV, Solomon SD, Inzucchi SE, et al. Dapagliflozin in patients with heart failure and reduced ejection fraction. N Engl J Med. 2019;381(21):1995-2008. PubMed
Farxiga (dapagliflozin) prescribing information. AstraZeneca. FDA Label
Solomon SD, McMurray JJV, Claggett B, et al. Dapagliflozin in heart failure with mildly reduced or preserved ejection fraction (DELIVER). N Engl J Med. 2022;387(12):1089-1098. PubMed
Heerspink HJL, Stefánsson BV, Correa-Rotter R, et al. Dapagliflozin in patients with chronic kidney disease (DAPA-CKD). N Engl J Med. 2020;383(15):1436-1446. PubMed
Heidenreich PA, Bozkurt B, Aguilar D, et al. 2022 AHA/ACC/HFSA guideline for the management of heart failure. J Am Coll Cardiol. 2022;79(17):e263-e421. PubMed