HFrEF Management

Definition

HFrEF is the one heart failure phenotype in which both quadruple foundational therapy and device therapy have randomised mortality evidence. This page owns the treatment pathway: the four pillars and their titration, the agents ranked behind them, revascularisation, ICD and CRT, and the drug choices that comorbidity forces. What the phenotype is — its definition, pathophysiology and diagnosis — is on entities/HFrEF, and the shared syndrome is on entities/Heart-Failure.

The structural fact that organises the page: two of the four pillars are phenotype-specific and two are not. SGLT2 inhibitors and MRAs are Class I, Level A at any ejection fraction, so they also appear on concepts/HFpEF-Management. The beta-blocker and the renin-axis inhibitor are HFrEF-specific, and they are the whole reason the two pathways diverge once the first two drugs are in. (sources/HF-ESC-2026, rating: very high; sources/HF-AHA-2022)

Key Concepts

The Four Pillars

(Table: sources/HF-AHA-2022, sources/arni-paradigm-hf-nejm-2014, sources/mra-hfref-rales-nejm-1999, sources/dapagliflozin-hfref-nejm-2019, sources/empagliflozin-hfref-nejm-2020, rating: very high)

ARNi. Component endpoints were CV death HR 0.80 (NNT=32), HF hospitalisation HR 0.79 and all-cause death HR 0.84. Against enalapril it caused less cough, less creatinine rise ≥2.5 mg/dL and fewer renal-impairment discontinuations (0.7% vs 1.4%; P=0.002), with more symptomatic hypotension but rarely leading to discontinuation. Switching requires a 36-hour washout of the ACE inhibitor to avoid angioedema. (sources/arni-paradigm-hf-nejm-2014; sources/Handbook-HA-2024, rating: high)

MRA. Eplerenone established benefit in NYHA II later, in EMPHASIS-HF. In RALES the secondary endpoints moved in the same direction — progressive HF death RR 0.64, sudden cardiac death RR 0.71 (P=0.02), HF hospitalisation RR 0.65 — and the result held across all six prespecified subgroups. Serious hyperkalaemia 2% vs 1% (NS); gynaecomastia or breast pain 10% vs 1% in men. (sources/mra-hfref-rales-nejm-1999)

SGLT2i. Fourth pillar; effect independent of diabetes status and of ischaemic versus nonischaemic aetiology (ischaemic patients are ~50% of both trials).

(Table: sources/dapagliflozin-hfref-nejm-2019, sources/empagliflozin-hfref-nejm-2020, rating: very high)

The EMPEROR-Reduced renal result suggests SGLT2i protection is disease-modifying rather than haemodynamic: the eGFR difference persisted after drug withdrawal (−0.93 vs −4.21 mL/min/1.73 m² at 23–45 days off treatment). (sources/empagliflozin-hfref-nejm-2020)

Sequencing and Titration

• No trial has tested a sequence, so no guideline specifies one. STRONG-HF calls the traditional sequential approach "suboptimal" and favours simultaneous implementation, but the 2026 ESC guideline declines to name a number or order of drug classes for that reason. Rapid pre-discharge initiation with close early follow-up is Class I/B2. (sources/HF-ESC-2026, rating: very high; sources/HF-update-ESC-2023)
• All four are titrated to target dose, with the loop diuretic titrated instead to the lowest dose that maintains euvolaemia. (sources/Handbook-HA-2024, rating: high)
• Hyperkalaemia is the practical brake on quadruple therapy, not hypotension: MRA and renin-axis inhibition both raise potassium, and the eGFR-dependent ceilings determine how far each can be pushed. Those thresholds, and what applies irrespective of phenotype, are tabulated on entities/Heart-Failure. See also concepts/Cardiorenal-Syndrome.
• Discharge is the decision point, not the clinic three months later. The six-week post-discharge window carries the highest readmission and death risk, which is where the guideline locates up-titration. Detail: concepts/Decompensated-HF-Management. (sources/HF-ESC-2026)

Additional Agents

(Table: sources/HF-ESC-2021, sources/HF-AHA-2022, sources/vericiguat-victoria-nejm-2020, sources/digitoxin-hfref-digithf-nejm-2025, sources/omecamtiv-galactichf-nejm-2021, rating: very high; high)

• Vericiguat works by soluble guanylate cyclase stimulation, which overlaps pharmacodynamically with nitrates and PDE5 inhibitors. It enrolled 2016–2018, so its incremental value on modern quadruple GDMT is untested. The handbook's entry criteria — LVEF <45%, recent hospitalisation or IV diuretics, raised natriuretic peptides — are broader than VICTORIA's. See entities/Vericiguat. (sources/vericiguat-victoria-nejm-2020; sources/Handbook-HA-2024)
• Ivabradine is one of the few agents whose rate criterion is printed on the label: sinus rhythm with heart rate ≥70 on a maximally tolerated beta-blocker. (sources/Handbook-HA-2024)
• DIGIT-HF used digitoxin rather than the guideline-listed digoxin, enrolled 1,212 of a planned 2,190, and had median treatment of 18 months. Both individual components were non-significant despite a significant composite. (sources/digitoxin-hfref-digithf-nejm-2025, rating: high)
• Omecamtiv mecarbil binds cardiac myosin to increase force-generating heads without altering calcium transients — distinct from classical inotropes — but was not approved. See entities/Omecamtiv-Mecarbil. (sources/omecamtiv-galactichf-nejm-2021, rating: high)

Iron Deficiency

Most evidence is from LVEF ≤45%. Diagnose by ferritin <100 ng/mL, or ferritin 100–300 ng/mL with TSAT <20%.

• To improve symptoms and quality of life: ESC Class I, Level A — upgraded from IIa in 2021 on the basis of FAIR-HF, CONFIRM-HF and EFFECT-HF. (sources/HF-update-ESC-2023, rating: very high)
• To reduce HF hospitalisation: ESC Class IIa, Level A — based on AFFIRM-AHF and IRONMAN. (sources/HF-update-ESC-2023)
• HEART-FID (n=3,065; ambulatory, LVEF ≤40%) — the largest IV iron trial — failed its hierarchical primary composite (win ratio 1.10; 99% CI 0.99–1.23; P=0.02 against a required p<0.01), and CV death or HF hospitalisation was non-significant (HR 0.93). The divergence from AFFIRM-AHF (acute HF at discharge, rate ratio 0.79) is most often attributed to the lower-risk ambulatory population, higher baseline TSAT, pandemic-era hospitalisation attenuation, and the stringent alpha. (sources/fe-hf-heartfid-nejm-2023, rating: high)
• IV iron is not recommended in HFpEF — see concepts/HFpEF-Management. Full evidence synthesis: concepts/Iron-Deficiency-in-HF.

Drugs to Avoid

AHA COR 3 Harm: non-dihydropyridine calcium channel blockers (negative inotropy), Class IC antiarrhythmics and dronedarone (pro-arrhythmic in structural disease), thiazolidinediones (fluid retention), saxagliptin and alogliptin (HF hospitalisation signal), and NSAIDs. (sources/HF-AHA-2022, rating: very high)

Antiarrhythmic Agents

• Amiodarone is the only antiarrhythmic without evidence of net mortality harm in HFrEF overall, but that framing is too generous. SCD-HeFT (n=2,521; LVEF ≤35%; median 45.5 months) found overall HR 1.06 (P=0.53) — neutral, not beneficial — with a prespecified NYHA interaction (P=0.004): NYHA II HR 0.85 (NS); NYHA III HR 1.44 (97.5% CI 1.05–1.97), a statistically significant harm signal. The "acceptable in HFrEF" designation is defensible chiefly in NYHA II, and amiodarone never substitutes for an ICD in primary prevention. (sources/icd-hfref-scdheft-nejm-2005, rating: very high; sources/amiodarone-cvdrug-2020, rating: high)
• Class IC agents (flecainide, propafenone) and dronedarone are actively harmful in structural heart disease — COR 3 Harm. (sources/HF-AHA-2022)
• Full pharmacology, dosing and toxicity monitoring: entities/Amiodarone.

Blood Pressure in HFrEF

• Uptitrate GDMT first. ARNi/ACEi/ARB, beta-blocker, MRA and SGLT2i at maximally tolerated doses serve both the HF and the blood-pressure indication. (sources/HT-AHA-2025, rating: very high)
• Add a dihydropyridine calcium channel blocker (e.g. amlodipine) only if pressure remains elevated on optimised GDMT. Non-dihydropyridine agents remain COR 3 Harm. (sources/HT-AHA-2025)
• Target SBP <130 mmHg. SPRINT reduced incident HF by 38% (HR 0.62) with an SBP target below 120 mmHg. (sources/HT-AHA-2025)

Comorbidity — COPD and Beta-Blocker Choice

• Beta-blockers remain Class I in HFrEF + COPD — the inverse of the HFpEF recommendation on concepts/HFpEF-Management. Use β1-selective agents: bisoprolol is preferred, reducing both HF events and COPD exacerbations in a dose-dependent manner and outperforming carvedilol and metoprolol in this population. Carvedilol's β2 blockade lowers FEV1/FVC and DLCO — avoid it in hypoxic or reduced-DLCO patients, or use it at low dose with pulmonary function monitoring where the VE/VCO₂ slope is exaggerated. See concepts/HF-COPD-Comorbidity. (sources/hf-copd-hfreview-2025, rating: high)

Revascularisation in Ischaemic Cardiomyopathy

• CABG for LVEF ≤35%: AHA COR 1/B-R. STICH (n=1,212) showed lower CV death (28% vs 33%; P=0.05) and lower all-cause death at 10 years versus medical therapy alone. Selection requires multivessel disease, viable myocardium and acceptable surgical risk. (sources/CCS-AHA-2023, rating: very high)
• PCI for LVEF ≤35%: no survival benefit. REVIVED-BCIS2 (n=700; LVEF ≤35%; BCIS jeopardy score ≥6; viability in ≥4 dysfunctional segments; median 41 months) found no difference in all-cause death or HF hospitalisation (HR 0.99; P=0.96; 37.2% vs 38.0%). The early KCCQ benefit (+6.5 points at 6 months) eroded to non-significance by 24 months as the medical arm improved. Angiography retains a diagnostic role in defining aetiology and directing lipid therapy. (sources/pci-hf-revived-bcis2-nejm-2022, rating: very high)
• Viability imaging does not identify who benefits. LGE-CMR, PET-FDG, SPECT and dobutamine echo all predict segmental contractile recovery, but no trial has shown that viability-guided revascularisation improves survival — including the STICH viability substudy (n=618), in which viability did not discriminate CABG benefit. Full imaging performance data: concepts/Myocardial-Viability. (sources/imaging-viability-aha-2020, rating: very high)
• SGLT2i benefit is aetiology-independent — ischaemic patients were ~50% of DAPA-HF and EMPEROR-Reduced, with no interaction. (sources/HF-AHA-2022)

Device Therapy

ICD.

• ESC 2026: secondary prevention I/A; primary prevention in ischaemic HFrEF at LVEF ≤35% after ≥3 months of foundational medical therapy I/B1; non-ischaemic IIa/B1. Full recommendation table, AUC 2025 scenarios, generator replacement and S-ICD: entities/ICD and entities/S-ICD. (sources/HF-ESC-2026, rating: very high; sources/icd-crt-auc-2025)
• The LVEF ≤35% threshold comes from SCD-HeFT (HR 0.77; 23% RRR; no aetiology interaction), not MADIT-II, which used ≤30% (HR 0.69). (sources/icd-hfref-scdheft-nejm-2005, sources/icd-icm-maditii-nejm-2002)
• In non-ischaemic disease the arrhythmic signal is clear but the mortality signal is not. DEFINITE halved sudden death (HR 0.20; P=0.006) with all-cause mortality non-significant (HR 0.65) because only a third of deaths were arrhythmic; DANISH then found no all-cause benefit (HR 0.87) despite SCD HR 0.50. That pair is why ESC rates non-ischaemic IIa rather than Class I. (sources/icd-nicm-definite-nejm-2004, sources/icd-nicm-danish-nejm-2016)
• High-risk pathogenic variants (LMNA, FLNC, desmosomal proteins, phospholamban): consider ICD even with LVEF >35%, because arrhythmic risk is disproportionate to EF. (sources/HF-AHA-2022)
• ICD in CKD: benefit is established above eGFR 60 (HR 0.49) but not below (HR 0.80); registry data suggest benefit at 30–59 but not <30, with more complications. (sources/cardiorenal-aha-2019)

CRT.

• ESC 2026: LBBB ≥150 ms in sinus rhythm at LVEF ≤35% is I/A; LBBB 130–149 ms and non-LBBB ≥150 ms are IIa/C; non-LBBB 130–149 ms is IIb/C; QRS <130 ms is III/A. Full table and CSP comparison: entities/CRT. (sources/HF-ESC-2026, rating: very high)
• The QRS ≥150 ms threshold comes from MADIT-CRT's prespecified interaction, not its entry criterion — QRS ≥150 ms HR 0.48 versus 130–149 ms HR 1.06 (P=0.001). CARE-HF (CRT-P, HR 0.64, NNT ≈9) left 7% residual sudden deaths despite reverse remodelling, which is the argument for CRT-D over CRT-P in ICD-eligible patients. (sources/crt-maditcrt-nejm-2009, sources/crt-carehf-nejm-2005)
• CRT in CKD: functional improvement persists at eGFR 30–59, but all-cause mortality after CRT is higher below 60 (HR 1.66). (sources/cardiorenal-aha-2019)

TEER for secondary mitral regurgitation.

• ESC 2025 Class I A (upgraded from IIb): recommended for haemodynamically stable symptomatic patients with LVEF 20–50%, persistent severe ventricular SMR despite optimised GDMT and CRT, LVESD ≤70 mm, ≥1 HF hospitalisation in the past year or elevated BNP, SPAP ≤70 mmHg, no severe RV dysfunction, no advanced HF and no revascularisation-requiring CAD. (sources/vhd-esc-2025, rating: very high)
• Selection is the whole result: benefit was confined to disproportionate MR (COAPT), with no benefit in proportionate MR (MITRA-FR). See concepts/Secondary-Mitral-Regurgitation. (sources/vhd-esc-2025)

Contradictions / Open Questions

Foundational and Additional Therapy

• PARADIGM-HF stopped early and enriched by run-in. Early termination overestimates effect size, and the double run-in excluded about 20% of screened patients for intolerance — so the enrolled population is more tolerant and higher-risk than real-world HFrEF. (sources/arni-paradigm-hf-nejm-2014, rating: very high)
• PARADIGM-HF quality-of-life gain is below the MCID. KCCQ improved by +1.64 points, statistically significant but under the ~5-point threshold for clinical importance, so the population-level symptomatic benefit may be modest. The trial also predates SGLT2i, so the incremental advantage of ARNi over contemporary quadruple GDMT is unknown.
• RALES overstates MRA benefit and understates its risk. It was stopped at the efficacy boundary after a mean 24 months, which inflates the 30% relative risk reduction; background therapy was ACE inhibitor plus loop diuretic with no beta-blocker requirement; and it was manufacturer-funded. Critically, the 2% hyperkalaemia rate was not reproduced in practice: spironolactone prescribing rose 20-fold and hyperkalaemia hospitalisation 6.8-fold in elderly patients after publication, with excess mortality — a failure to apply the trial's creatinine ≤2.5 mg/dL and potassium ≤5.0 mmol/L exclusions. (sources/mra-hfref-rales-nejm-1999, rating: very high)
• DAPA-HF and EMPEROR-Reduced disagree on mortality. DAPA-HF reduced both CV death (HR 0.82) and all-cause death (HR 0.83); EMPEROR-Reduced, despite enrolling sicker patients, reduced neither (both HR 0.92). Whether this reflects the shorter follow-up, the higher background event rate or a real population difference is unresolved; the mortality pillar of the evidence rests more firmly on DAPA-HF. (sources/dapagliflozin-hfref-nejm-2019, sources/empagliflozin-hfref-nejm-2020)
• Background therapy limits modern applicability. Only ~10% of DAPA-HF patients were on sacubitril-valsartan; the post hoc ARNi subgroup (HR 0.75; 0.50–1.13) is consistent but underpowered. Adding SGLT2i to full quadruple GDMT is extrapolated, not tested.
• DAPA-HF's only prespecified heterogeneity signal was NYHA III/IV attenuation. Other severity markers — worse LVEF, higher NT-proBNP, lower eGFR — showed no attenuation, and guidelines recommend SGLT2i across NYHA II–IV regardless. (sources/dapagliflozin-hfref-nejm-2019)
• EMPEROR-Reduced's renal benefit rests on a 26% subgroup. The post-discontinuation eGFR data (n=966 with paired measurements) are the strongest evidence that SGLT2i renal protection is structural rather than haemodynamic, but selective measurement invites survivorship bias, and no renal-primary SGLT2i trial exists in HFrEF. (sources/empagliflozin-hfref-nejm-2020)
• VICTORIA predates SGLT2i. Vericiguat enrolled 2016–2018 with no SGLT2i co-administration, so its incremental benefit on top of modern quadruple GDMT is untested. (sources/vericiguat-victoria-nejm-2020)
• DIGIT-HF cannot be attributed to digoxin. It used digitoxin, which is hepatically metabolised rather than renally cleared, was underpowered (1,212 of a planned 2,190), had median treatment of only 18 months, and only 19.3% SGLT2i use. The significant composite also rests on two non-significant components. (sources/digitoxin-hfref-digithf-nejm-2025, rating: high)
• GALACTIC-HF's primary result contradicts its own entry criterion. Omecamtiv mecarbil reduced the composite (HR 0.92; P=0.03) with every secondary outcome null, and the prespecified LVEF interaction showed benefit confined to LVEF ≤28% (HR 0.84) with none above 28% (HR 1.04) — a population narrower than the ≤35% entry criterion. (sources/omecamtiv-galactichf-nejm-2021, rating: high)

Device Therapy

• CRT-for-AV-block fell from Class I to IIa with no new trial behind the change. The 2021 recommendation was I/A; the 2026 is IIa/B1. Nothing was published in between showing harm — the downgrade reads as a re-grading of certainty, which is exactly the kind of change that is invisible if you only track trial results. (sources/HF-ESC-2026, rating: very high)
• MADIT-CRT: benefit at QRS ≥150 ms only, no mortality benefit, and not an LBBB trial. The QRS ≥150 ms Class I threshold comes from the prespecified interaction, not the entry criterion. Annual mortality was 3% in both arms at 2.4 years. LBBB was not a primary subgroup, so MADIT-CRT does not test the modern LBBB-first guideline; that rests on later meta-analyses. Enrolled 2004–2008, pre-ARNi/SGLT2i, single manufacturer. (sources/crt-maditcrt-nejm-2009, rating: very high)
• CARE-HF: CRT-P mortality evidence is limited to NYHA III/IV and did not require LBBB. The current LBBB requirement comes from post-hoc analyses showing attenuated non-LBBB response. The 7% residual sudden death rate is the argument for CRT-D in ICD-eligible patients. NYHA II benefit was established by MADIT-CRT and REVERSE, not CARE-HF, and modern quadruple GDMT independently promotes reverse remodelling, potentially narrowing the absolute CRT benefit. (sources/crt-carehf-nejm-2005)
• MADIT-II threshold extrapolation and early stopping. MADIT-II used LVEF ≤30%; the current ≤35% threshold extends beyond its direct evidence and is partly supported by SCD-HeFT. It was stopped at the efficacy boundary with a mean 20-month follow-up. HF hospitalisation trended higher in the ICD arm (19.9% vs 14.9%; P=0.09), consistent with a rescue effect — patients saved from sudden death accumulate progressive HF. (sources/icd-icm-maditii-nejm-2002, rating: very high)
• DANISH predates contemporary GDMT, and its Class IIa rating reflects that. Enrolled 2008–2014 before SGLT2i and ARNi were standard; 31% of deaths were non-cardiovascular, so competing mortality limits any all-cause benefit. Benefit was significant only under age 68 (HR 0.64; P=0.01). (sources/icd-nicm-danish-nejm-2016, rating: high)
• SCD-HeFT found no ICD benefit in NYHA III (HR 1.16; NS) despite a prespecified interaction (P<0.001) favouring NYHA II (HR 0.54). Investigators advised against withholding ICDs in NYHA III, and guidelines do not stratify by NYHA class. The shock-only single-lead design and higher competing HF mortality in NYHA III offer partial explanations. (sources/icd-hfref-scdheft-nejm-2005, rating: very high)
• The amiodarone safety framing is not supported by its own trial. SCD-HeFT's prespecified NYHA interaction showed NYHA III HR 1.44 (1.05–1.97), yet no guideline stratifies amiodarone safety by NYHA class. (sources/icd-hfref-scdheft-nejm-2005, sources/amiodarone-cvdrug-2020)

Revascularisation and Iron

• HEART-FID versus AFFIRM-AHF: the acute-versus-ambulatory window. The leading hypothesis is that the peri-discharge window is a high-risk, iron-responsive state not reproduced in stable outpatients. But HEART-FID's p<0.01 threshold (an FDA special protocol agreement) means its P=0.02 would have counted as significant in any standard trial. The ESC recommendations therefore rest on AFFIRM-AHF, IRONMAN and older symptom trials, not on the largest study. Incremental benefit on contemporary quadruple GDMT plus SGLT2i is untested. (sources/fe-hf-heartfid-nejm-2023, rating: high; sources/HF-update-ESC-2023, rating: very high)
• STICH versus REVIVED-BCIS2: CABG and PCI diverge. Whether the discrepancy reflects completeness of surgical revascularisation, open-label bias, or a genuine biological difference in the LVEF ≤35% setting is unresolved. The 2023 guideline preserves CABG at COR 1 and gives PCI no survival indication. (sources/pci-hf-revived-bcis2-nejm-2022, sources/CCS-AHA-2023, rating: very high)

Connections

• Part of entities/HFrEF — the phenotype: definition, pathophysiology and diagnosis
• Part of entities/Heart-Failure — the parent overview
• The other phenotype's pathway: concepts/HFpEF-Management — two shared pillars, then divergence
• Decompensation and the discharge window: concepts/Decompensated-HF-Management
• Device detail: entities/ICD · entities/S-ICD · entities/CRT · concepts/Conduction-System-Pacing
• Drug pages: entities/Sacubitril-Valsartan · entities/Vericiguat · entities/Omecamtiv-Mecarbil · entities/Amiodarone · entities/Cardiac-Glycosides · concepts/Beta-Blockers
• Aetiology and substrate: entities/DCM · entities/TTN · entities/LMNA · concepts/Myocardial-Viability · concepts/Electrical-Remodeling
• Kidney limits: concepts/Cardiorenal-Syndrome · concepts/Diuretic-Resistance
• Comorbidity: concepts/HF-COPD-Comorbidity · entities/COPD · concepts/Iron-Deficiency-in-HF
• Valves: concepts/Secondary-Mitral-Regurgitation

Sources

• sources/HF-ESC-2026
• sources/HF-AHA-2022
• sources/HF-ESC-2021
• sources/HF-update-ESC-2023
• sources/Handbook-HA-2024
• sources/arni-paradigm-hf-nejm-2014
• sources/mra-hfref-rales-nejm-1999
• sources/dapagliflozin-hfref-nejm-2019
• sources/empagliflozin-hfref-nejm-2020
• sources/vericiguat-victoria-nejm-2020
• sources/digitoxin-hfref-digithf-nejm-2025
• sources/omecamtiv-galactichf-nejm-2021
• sources/amiodarone-cvdrug-2020
• sources/HT-AHA-2025
• sources/fe-hf-heartfid-nejm-2023
• sources/CCS-AHA-2023
• sources/pci-hf-revived-bcis2-nejm-2022
• sources/imaging-viability-aha-2020
• sources/icd-hfref-scdheft-nejm-2005
• sources/icd-icm-maditii-nejm-2002
• sources/icd-nicm-definite-nejm-2004
• sources/icd-nicm-danish-nejm-2016
• sources/crt-maditcrt-nejm-2009
• sources/crt-carehf-nejm-2005
• sources/icd-crt-auc-2025
• sources/cardiorenal-aha-2019
• sources/vhd-esc-2025
• sources/hf-copd-hfreview-2025