Did empagliflozin work better in patients with diabetes than without?

No. Hazard ratios were nearly identical (0.79 with diabetes, 0.78 without). The benefit is independent of glycemic control, consistent with non-glycemic mechanisms of action including natriuresis and reduced cardiac preload.

Was the benefit weaker in patients with very high ejection fractions?

Numerically, patients with EF above 60% showed a smaller point estimate (HR 0.87) compared to those with EF 41 to 49% (HR 0.71). The interaction was not statistically significant (p = 0.21), so the difference could be due to chance.

Did older patients tolerate empagliflozin as well as younger patients?

Yes. Patients above and below the median age of 72 had similar efficacy signals and comparable adverse event profiles, including similar rates of volume depletion.

How did women respond compared to men in this trial?

Women had a hazard ratio of 0.82 (95% CI 0.67 to 1.00), compared to 0.77 for men. The interaction p-value was 0.65, meaning no statistically significant difference by sex. The slightly wider female CI reflects fewer events, not reduced efficacy.

Does baseline kidney function affect the benefit of empagliflozin in HFpEF?

No. Patients with eGFR above and below 60 mL/min/1.73 m² both benefited. The FDA label permits initiation at eGFR as low as 20 for heart failure indications.

Were Black patients adequately represented in this trial?

No. Black patients made up roughly 3.8% of enrollment (225 of 5,988). The confidence interval for this subgroup was wide (HR 0.68 to 95% CI 0.35 to 1.31), preventing any firm conclusions about differential efficacy.

Does a low NT-proBNP mean empagliflozin won't help?

Not necessarily. Patients below the median NT-proBNP had a hazard ratio of 0.88 (95% CI 0.72 to 1.07). The point estimate still favors treatment, though the confidence interval crosses 1.0. Low BNP in HFpEF sometimes reflects diagnostic uncertainty rather than mild disease.

Did obesity affect response to empagliflozin in this trial?

Obese patients (BMI ≥30) showed a numerically larger benefit (HR 0.76) than non-obese patients (HR 0.82), but the interaction was not significant. The trend is mechanistically plausible given the metabolic drivers of obesity-related HFpEF.

Did background atrial fibrillation change the treatment effect?

No. A post-hoc analysis found consistent benefit regardless of atrial fibrillation status at baseline (interaction p = 0.42), despite AF being present in over half the trial population.

Should clinicians restrict empagliflozin to certain HFpEF subgroups based on these data?

The data do not support restricting use to specific subgroups. The 2022 AHA/ACC/HFSA guideline recommends SGLT2 inhibitors broadly in HFpEF. Clinicians may reasonably prioritize patients with stronger congestion signals, but no analyzed subgroup showed clear evidence of non-response.

EMPEROR-Preserved Subgroup Analyses: Who Responded Most and Least

By HealthRX Medical Team

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

At a glance

| Parameter | Detail | |---|---| | N | 5,988 | | Intervention | Empagliflozin 10 mg once daily | | Comparator | Placebo | | Duration | Median 26.2 months | | Primary endpoint | Composite of cardiovascular death or hospitalization for heart failure | | Key result | HR 0.79 (95% CI 0.69, 0.90); p <0.001 |

Why subgroup analyses matter here

Before EMPEROR-Preserved, SGLT2 inhibitors had proven efficacy in heart failure with reduced ejection fraction (HFrEF). The open question was whether this benefit extended across the full spectrum of left ventricular function. The primary trial publication answered "yes" at the population level. But heart failure with preserved ejection fraction is not one disease. It is a syndrome with enormous phenotypic heterogeneity, spanning lean elderly women with stiff ventricles, obese younger patients with metabolic syndrome, and everyone between. Subgroup analyses tell us whether empagliflozin works uniformly or whether certain phenotypes carry the signal.

Pre-specified subgroup results

The EMPEROR-Preserved investigators pre-specified 15 subgroups in the statistical analysis plan. The forest plot that accompanied the primary publication showed consistent point estimates favoring empagliflozin across nearly every partition. None of the interaction p-values for the primary composite crossed the 0.05 threshold.

Ejection fraction strata

This was the most scrutinized subgroup. Patients were stratified by baseline EF:

| EF Range | n | HR (95% CI) | Interaction p | |---|---|---|---| | 41 to 49% (HFmrEF) | ~1,983 | 0.71 (0.57, 0.88) |, | | 50 to 59% | ~2,058 | 0.80 (0.64, 0.99) |, | | ≥60% | ~1,947 | 0.87 (0.69, 1.10) | 0.21 |

The pattern is visually graded: benefit appears most pronounced in the midrange and attenuates as EF climbs above 60%. But the interaction p-value of 0.21 means we cannot conclude these are statistically different effects. A pooled analysis of EMPEROR-Reduced and EMPEROR-Preserved later confirmed a continuous relationship across the EF spectrum, with the hazard ratio trending toward 1.0 only at the very highest EFs (above 65%).

This distinction matters for prescribing. Clinicians sometimes hesitate to start SGLT2 inhibitors in patients with EF of 55% or above, citing weaker evidence. The trial data do not support that threshold. The confidence intervals overlap substantially across strata.

Diabetes status

Roughly 49% of enrolled patients had type 2 diabetes. Results by glycemic status:

| Subgroup | HR (95% CI) | |---|---| | With diabetes | 0.79 (0.67, 0.94) | | Without diabetes | 0.78 (0.64, 0.97) |

Point estimates are nearly identical. This finding was clinically important because it confirmed that empagliflozin's cardiac benefit operates independently of glucose lowering, consistent with proposed mechanisms involving natriuresis, reduced preload, and improved myocardial energetics. The 2022 AHA/ACC/HFSA heart failure guideline subsequently recommended SGLT2 inhibitors for HFpEF regardless of diabetes status, a class 2a recommendation.

Age

Patients were divided at the median age of approximately 72 years.

| Subgroup | HR (95% CI) | |---|---| | <72 years | 0.78 (0.64, 0.95) | | ≥72 years | 0.80 (0.67, 0.95) |

No meaningful divergence. Older patients tolerated the drug well, with similar rates of volume depletion and urinary tract infection across age groups.

Sex

Women made up 44.7% of enrollment, a higher proportion than most HFrEF trials. Results:

| Subgroup | HR (95% CI) | |---|---| | Male | 0.77 (0.65, 0.92) | | Female | 0.82 (0.67, 1.00) |

The female confidence interval just touches 1.0, but this reflects smaller event counts rather than a biological signal of non-response. The interaction p-value was non-significant (p = 0.65). Women with HFpEF have historically been underrepresented in trials that shaped treatment guidelines, so the fact that EMPEROR-Preserved enrolled them at a proportion closer to real-world prevalence is itself notable.

Body mass index

Obesity is central to many HFpEF phenotypes. Patients were stratified by BMI:

| BMI Category | HR (95% CI) | |---|---| | <30 kg/m² | 0.82 (0.69, 0.97) | | ≥30 kg/m² | 0.76 (0.63, 0.92) |

No significant interaction. The numerically larger effect in obese patients aligns with the hypothesis that SGLT2 inhibitors address volume and metabolic drivers that are more prominent in the obese HFpEF phenotype. A secondary analysis published in Circulation explored the relationship between BMI and treatment effect in greater detail, finding consistent benefit across the full BMI distribution without a clear inflection point.

Baseline renal function

Chronic kidney disease is common in HFpEF. eGFR subgroups:

| eGFR Stratum | HR (95% CI) | |---|---| | ≥60 mL/min/1.73 m² | 0.78 (0.66, 0.93) | | <60 mL/min/1.73 m² | 0.80 (0.66, 0.97) |

Again, no interaction. The empagliflozin FDA label permits initiation at eGFR as low as 20 mL/min/1.73 m² for heart failure indications, a threshold supported by renal subgroup data from the broader EMPEROR program.

Baseline NT-proBNP

This is where things get interesting. Patients with higher baseline natriuretic peptides (above the median of approximately 970 pg/mL) showed a clearer benefit signal:

| NT-proBNP Level | HR (95% CI) | |---|---| | Above median | 0.73 (0.63, 0.86) | | Below median | 0.88 (0.72, 1.07) |

The interaction p-value was approximately 0.09. Not formally significant, but the trend is clinically coherent. Patients with lower NT-proBNP values may have less hemodynamic congestion at baseline, giving empagliflozin's diuretic and preload-reducing mechanisms less room to operate. In practice, a low BNP in someone labeled "HFpEF" sometimes reflects diagnostic misclassification rather than mild disease.

Race and geographic region

The trial enrolled patients across 23 countries. Reported race/ethnicity subgroups:

| Subgroup | HR (95% CI) | |---|---| | White | 0.81 (0.70, 0.93) | | Asian | 0.69 (0.49, 0.96) | | Black | 0.68 (0.35, 1.31) |

Asian patients showed a numerically strong benefit, though the sample was smaller. The Black subgroup (n = 225, roughly 3.8% of the trial) had wide confidence intervals that preclude firm conclusions. This underrepresentation is a limitation the authors explicitly acknowledged.

Post-hoc explorations worth noting

Several post-hoc analyses expanded on pre-specified subgroups. A 2022 analysis in the European Heart Journal examined the effect of empagliflozin across the full HFpEF phenotypic spectrum, including patients stratified by atrial fibrillation status, number of comorbidities, and background medication use.

Key findings from post-hoc work:

Atrial fibrillation at baseline (present in ~52% of patients): benefit was consistent regardless of AF status (interaction p = 0.42).
Mineralocorticoid receptor antagonist use: no significant interaction, though patients already on MRAs had slightly higher event rates, complicating interpretation.
Loop diuretic dose: patients on higher diuretic doses (a proxy for more advanced congestion) trended toward larger absolute benefit, consistent with the NT-proBNP signal described above.

Limitations of the subgroup data

Several caveats apply. Subgroup analyses in any single trial are hypothesis-generating, not confirmatory. EMPEROR-Preserved was powered for the overall population, not for individual subgroups. Multiple comparisons inflate the chance of false positives and false negatives simultaneously.

The EF distribution deserves scrutiny. The trial enrolled patients with EF above 40%, meaning a substantial proportion (roughly one-third) had EF between 41% and 49%. European and American guidelines now classify this range as HFmrEF, a distinct category. Including these patients boosted event rates and likely strengthened the overall result. Whether the benefit in "true" HFpEF (EF ≥50%) is as strong remains debated, though the point estimates still favor treatment.

Enrollment demographics also limit generalizability. Black patients comprised under 4% of the trial. Hispanic/Latino ethnicity was not separately reported. The median age was 72, so data in younger HFpEF patients (particularly those with obesity-driven disease in their 50s and 60s) are sparse.

What this means for prescribing

The subgroup data from EMPEROR-Preserved support a simple clinical message: empagliflozin benefits most identifiable HFpEF subgroups, and there is no clearly defined patient profile that should be excluded from consideration. The 2022 AHA/ACC/HFSA guideline reflected this by recommending SGLT2 inhibitors broadly in HFpEF without restrictive qualifiers.

Clinicians may reasonably prioritize initiation in patients with more convincing phenotypic markers of congestion: elevated natriuretic peptides, higher BMI, active loop diuretic use, and EF in the 40 to 55% range. These are the patients where the point estimates were most favorable and the mechanistic rationale is strongest. But withholding empagliflozin from a patient with an EF of 62% and borderline NT-proBNP would not be well supported by these data either.

Frequently asked questions

›

References

Anker SD, Butler J, Filippatos G, et al. Empagliflozin in Heart Failure with a Preserved Ejection Fraction. N Engl J Med. 2021;385(16):1451-1461. PubMed
Heidenreich PA, Bozkurt B, Aguilar D, et al. 2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure. Circulation. 2022;145(18):e895-e1032. PubMed
Packer M, Butler J, Zannad F, et al. Empagliflozin and Major Renal Outcomes in Heart Failure. N Engl J Med. 2021;385(16):1531-1533. PubMed
Empagliflozin (Jardiance) Prescribing Information. U.S. Food and Drug Administration. FDA Label
Packer M, Anker SD, Butler J, et al. Influence of Ejection Fraction on the Effects of Empagliflozin in Heart Failure. Eur Heart J. 2022;43(suppl):ehac544. PubMed