Heart Failure Outcomes Most Certain With SGLT2 Inhibitors

The clearest benefits are heart-failure outcomes

A 2026 systematic review and meta-analysis of 59 randomized controlled trials reinforces a hierarchy of evidence for sodium-glucose cotransporter 2, or SGLT2, therapy in heart failure. The most dependable benefits are fewer heart-failure hospitalizations and a lower risk of cardiovascular death. The pooled all-cause mortality result is less certain because analyses designed to detect or account for publication bias weakened that signal.

That distinction matters when clinicians explain treatment value. Avoiding hospitalization is not a surrogate or merely a laboratory change; it is a clinically important outcome for patients and health systems. Cardiovascular death is also disease-specific and directly relevant. A possible all-cause mortality benefit may be discussed, but it should not be presented with the same confidence as the established heart-failure benefits.

The conclusion is consistent with the major placebo-controlled outcome trials. In heart failure with reduced ejection fraction, dapagliflozin reduced the composite of worsening heart failure or cardiovascular death in DAPA-HF. Empagliflozin produced a similar reduction in EMPEROR-Reduced. In populations with mildly reduced or preserved ejection fraction, EMPEROR-Preserved and DELIVER showed that the benefit extended across a broader range of left ventricular ejection fraction, although these composite results were driven predominantly by fewer heart-failure events.

Individual trials also illustrate why endpoint-specific interpretation is important. DAPA-HF reported a cardiovascular-death hazard ratio of 0.82 (95% CI 0.69 to 0.98). In EMPEROR-Reduced, the corresponding estimate was 0.92 (0.75 to 1.12). Trials in preserved or mildly reduced ejection fraction likewise showed clearer reductions in heart-failure events than in cardiovascular death considered alone.

Pooling increases statistical precision and can reveal a cardiovascular mortality effect that individual trials were not powered to establish.

How the review assembled the evidence

The systematic review combined 59 randomized trials evaluating SGLT2 inhibitors or sotagliflozin in people with heart failure. Sotagliflozin inhibits both SGLT2 and SGLT1 and is therefore not pharmacologically identical to the selective SGLT2 agents. Comparators were generally placebo or control therapy added to background heart-failure treatment, while clinical setting, ejection fraction, diabetes status, and follow-up varied across trials.

Because the evidence base consists of randomized trials, the pooled treatment comparisons are better suited to causal inference than observational associations. Randomization does not, however, eliminate problems introduced during evidence synthesis. Differences among trial populations and endpoint definitions can generate heterogeneity, and selective publication or selective reporting can make an overall effect appear more favorable than the complete evidence would support.

Publication-bias methods commonly assess whether smaller or less favorable studies may be missing from the available literature. Funnel-plot asymmetry and related statistical tests are warning signals, not proof that investigators or journals withheld results. Sensitivity analyses then estimate how the pooled result might change under assumptions about missing studies. In this review, those checks were consequential for all-cause mortality but did not displace the more stable conclusions for heart-failure hospitalization and cardiovascular death.

All-cause mortality is often perceived as the hardest endpoint because it avoids disagreement about the cause of death. Yet it can be statistically difficult in a heart-failure program. Trials may have limited follow-up, deaths unrelated to cardiovascular disease dilute a treatment effect, and relatively few deaths reduce precision. A result can therefore be clinically plausible and conventionally significant in a primary analysis while still becoming uncertain when small-study effects or potentially missing evidence are modeled.

Translating certainty into clinical discussions

The evidence supports presenting prevention of worsening heart failure—especially hospitalization—as the most predictable outcome of SGLT2 treatment. This benefit has appeared across multiple large trials, with and without diabetes and across reduced, mildly reduced and preserved ejection fraction. The consistency across different agents and populations is more informative than any single composite endpoint.

Cardiovascular mortality can also be described as reduced in the pooled randomized evidence, while acknowledging that several individual trials did not show statistically conclusive reductions in cardiovascular death alone. That is not contradictory. Meta-analysis combines information and improves precision, whereas each trial reflects its own event rate, duration, and eligibility criteria.

All-cause mortality warrants more restrained language: the therapy may lower it, but the magnitude and certainty are less established. This framing avoids two errors. It does not convert a sensitivity-analysis concern into a claim that there is no mortality benefit, and it does not use a vulnerable pooled estimate as the principal reason to start treatment.

For policy and quality measurement, hospitalization outcomes may offer the strongest basis for expected population benefit. However, translate relative effects into absolute terms whenever baseline risk is available. A similar hazard ratio can correspond to a substantially larger absolute benefit in a recently hospitalized, high-risk patient than in a stable outpatient with fewer expected events.

Limits that remain important

The meta-analysis spans trials with different ejection-fraction thresholds, acute and chronic clinical settings, background therapies, follow-up periods and definitions of worsening heart failure. Aggregate pooled estimates may not identify how effects differ within older adults, patients with advanced kidney disease, those with very low blood pressure, or groups underrepresented in the original trials.

Publication-bias analyses also depend on assumptions and can be unstable when studies differ substantially in size or design. They lower confidence in the all-cause mortality estimate but cannot determine whether unpublished neutral studies truly exist. Finally, trial efficacy and safety reflect protocolized follow-up and eligibility criteria; real-world tolerability, access, and competing illness still affect the benefit an individual patient may experience.

Questions clinicians ask

Which outcome should lead the treatment conversation?

Prevention of worsening heart failure, particularly hospitalization, has the most consistent support across large randomized trials and the 59-trial synthesis. Pooled evidence also shows reduced cardiovascular mortality. All-cause mortality can be mentioned as possible, but it should not be framed as the most certain or primary benefit.

Does publication-bias sensitivity negate the cardiovascular benefit?

No. The concern is outcome-specific. Sensitivity analyses weaken confidence in the all-cause mortality signal, but they do not erase the reproducible reductions in heart-failure events or the pooled reduction in cardiovascular death. Funnel asymmetry should prompt calibrated language rather than wholesale dismissal of the randomized evidence.

Do these findings apply across ejection-fraction categories?

The major outcome trials enrolled patients with reduced, mildly reduced, and preserved ejection fraction, and each category showed fewer composite heart-failure events with dapagliflozin or empagliflozin. The relative contribution of cardiovascular death was less consistent, especially in preserved ejection fraction, where fewer worsening-heart-failure events drove much of the benefit.

Should all-cause mortality influence treatment selection?

It may contribute to the overall discussion, but it should not carry the same evidentiary weight as heart-failure hospitalization or cardiovascular death. Treatment selection should rest on the established cardiovascular and heart-failure outcomes, applicable labeling, safety considerations, kidney function, comorbidities, and the patient’s baseline risk and priorities.

References

1. Efficacy and Safety of SGLT2 Inhibitors and Sotagliflozin in Heart Failure: A Systematic Review and Meta-Analysis of 59 Randomized Controlled Trials — PubMed Central, 2026 2. Dapagliflozin in Patients with Heart Failure and Reduced Ejection Fraction — The New England Journal of Medicine, 2019 3. Cardiovascular and Renal Outcomes with Empagliflozin in Heart Failure — The New England Journal of Medicine, 2020 4. Empagliflozin in Heart Failure with a Preserved Ejection Fraction — The New England Journal of Medicine, 2021 5. Dapagliflozin in Heart Failure with Mildly Reduced or Preserved Ejection Fraction — The New England Journal of Medicine, 2022