Reviews/SGLT2 Inhibitors/SGLT2 inhibitors reduce all-cause mortality in HFrEF
· SG-HF-02 · v2
Across every usable study in this corpus, people with heart failure and a weak pump who took an SGLT2 inhibitor died less often than those who did not. Three large real-world databases from Denmark, a multinational network and US Medicare all point the same way [1][2][3], and a randomised slice of a diabetes trial that happened to include such patients found the same [16]. The catch is that most of this evidence watches what happened to patients in ordinary care rather than randomly assigning the drug, so sicker or older patients may simply have been less likely to receive it. The single clean randomised number aimed at the whole heart-failure population did not reach statistical significance [5], and the one randomised result that was significant came from a small pre-planned subgroup, not a trial built to answer this question [16]. The panel treated the direction as consistent and well-supported but held back from the top tier because the strongest design is missing from the corpus.
Every included source points in the same direction: lower all-cause mortality with SGLT2i in HFrEF. The three large cohorts converge tightly, with hazard ratios of 0.70 [1], 0.75 [2] and 0.76 [3], each against a non-SGLT2i or alternative-drug comparator. A prespecified randomised HFrEF subgroup of a type-2-diabetes trial reported a significant all-cause mortality reduction (HR 0.59) [16], and a small single-centre study in ischaemic heart failure agreed (HR 0.27) [4]. The one broad randomised all-cause-death estimate was directionally concordant but non-significant (HR 0.82, 95% CI 0.59-1.14) [5]. No dedicated HFrEF randomised trial with all-cause mortality as its own endpoint sits in the corpus as primary data. The contested point: The divide is between the consistent direction and the design carrying it: the significant randomised signal comes only from a 671-patient subgroup of a diabetes trial [16], while the sole broad-population randomised all-cause-death estimate crossed 1 [5]. The bulk of support is observational [1][2][3][4], where confounding by indication remains possible even after adjustment.
All 23 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from SGLT2 Inhibitors, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
Four of seven included sources are observational (cohort/registry) and subject to confounding by indication and healthy-user bias; [2] explicitly reports SGLT2i users were younger and more often male, the direction of bias that favours the treated arm even after propensity adjustment.
The sole placebo-controlled randomised all-cause-mortality estimate in HFrEF [16] is a prespecified subgroup (N=671) of a type-2-diabetes trial, not a dedicated HFrEF trial; subgroup results are hypothesis-supporting rather than confirmatory.
Where it remains uncertain. The one broad-population randomised all-cause-death estimate was non-significant (HR 0.82, 95% CI 0.59-1.14) [5], and the significant randomised result is a subgroup of a diabetes trial [16].
Aleatory
The significant randomised estimate rests on 671 patients [16], so sampling variability alone leaves a wide plausible range.
Epistemic
The corpus holds no randomised trial built to test all-cause mortality in HFrEF as its own endpoint; the trials that would appear only inside excluded meta-analyses [11][13][18].
Inferential
Reading a diabetes-trial HFrEF subgroup [16] as evidence for HFrEF generally extrapolates beyond the trial's design.
The load-bearing papers from a corpus of 23, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
National HFrEF registry; lower all-cause mortality, though the comparator is MRA rather than no-SGLT2i.
“In weighted analysis comparing SGLT2 inhibitors to MRAs, the wHR was 0.70 (95% CI 0.57-0.86; absolute risk difference -2.1 per 100 person-years, 95% CI -0.9 to -3.2)”
Panel auditInternal validity: medium · scope of inference: Danish HFrEF patients (LVEF ≤40%, aged ≥45) already on RASI and beta-blockers, SGLT2i vs MRA as third-line.Key limitations: active comparator is MRA not placebo · observational confounding by indication · single-country registry
Large multinational real-world HFrEF cohort; significant all-cause mortality reduction vs non-users.
“Sodium-glucose cotransporter-2 inhibitor users vs non-users had a lower risk of 1-year all-cause mortality [HR, 95% confidence interval (CI) = 0.75, 0.69-0.83]”
Panel auditInternal validity: medium · scope of inference: Newly diagnosed HFrEF patients after first HF hospitalization in a multinational EHR network.Key limitations: retrospective EHR data · residual/unmeasured confounding · immortal-time/adherence not addressed in abstract
Independent US Medicare cohort of hospitalised older HFrEF; concordant significant reduction, consistent across EF subgroups.
“SGLT2i initiation was independently associated with lower all-cause mortality (adjusted hazard ratio [HR], 0.76 [95% CI, 0.67-0.86])”
Panel auditInternal validity: medium · scope of inference: US Medicare patients aged ≥65 hospitalized for HFrEF and SGLT2i-eligible.Key limitations: observational · confounding by indication · older-only population
Only a composite win ratio is reported; all-cause mortality is not isolated so its direction cannot be read from the abstract.
Panel auditInternal validity: medium · scope of inference: Patients hospitalized for acute de novo or decompensated chronic HF regardless of ejection fraction.Key limitations: all-cause mortality not isolated (composite win ratio) · population not HFrEF-specific · 90-day follow-up
4 of 23 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
Direction is robust and convergent across independent datasets, but the corpus lacks a dedicated HFrEF randomised all-cause-mortality endpoint and leans on observational data, warranting Likely rather than Established.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
No primary HFrEF randomised all-cause mortality endpoint. Every randomised datapoint in the corpus is either a subgroup [16] or a non-significant broad-population estimate [5]; the direct confirmatory design is absent as primary data. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: Include, as primary research, the dedicated HFrEF SGLT2i outcome trials whose all-cause mortality results currently reach the corpus only through excluded meta-analyses [11][13][18].
Open question: Does SGLT2i reduce all-cause mortality in an HFrEF population when all-cause death is a prespecified randomised endpoint rather than a subgroup result? Why it is still open: The only significant randomised estimate in the corpus is a 671-patient subgroup of a diabetes trial [16] and the one broad randomised all-cause-death estimate was non-significant [5].
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 23 Oct 2026 · protocol v0.19