Independent · Evidence-led · We don't sell peptides
EU / NordicsUpdated weeklyEN

Explore this article's sources with AI

ChatGPTClaudePerplexityGeminiGrokGoogle AI

Follow PeptideMethods on Google

First published

GLP-1 versus SGLT2: a confounded cohort

A matched records study favored GLP-1 drugs in MASLD with heart failure, but early separation makes treatment channeling likely.

Why we wrote this. A striking records association needs its early timing and residual confounding stated up front.

In this article (5 sections)
  1. The recorded outcome differences
  2. The first 90 days raise the main doubt
  3. Matching did not make this a randomized trial
  4. Randomized heart-failure evidence points elsewhere
  5. What this study is useful for

A US records study found lower observed mortality and fewer coded acute heart-failure diagnoses among adults with metabolic dysfunction-associated steatotic liver disease and heart failure who received a GLP-1 receptor agonist rather than empagliflozin or dapagliflozin. The comparison was observational. The unusually large early difference and incomplete new-user verification limit it. Remaining baseline imbalances also mean it cannot establish that GLP-1 drugs caused better outcomes or should be preferred[1].

The recorded outcome differences

Investigators used the TriNetX US Collaborative Network and an index eligibility period from 1 January 2018 through 2 June 2026. The GLP-1 group included semaglutide, liraglutide, or dulaglutide. The active-comparator group included empagliflozin or dapagliflozin. Propensity matching produced 8,550 people per group for the main analysis[1].

Recorded mortality was 7.0% in the GLP-1 group and 10.1% in the SGLT2 inhibitor group. The hazard ratio was 0.607. Coded acute heart failure occurred in 14.2% and 29.5%, respectively, with a hazard ratio of 0.398. Those estimates describe associations in coded records after matching, not randomized treatment effects[1].

Among matched patients with type 2 diabetes and no recorded type 1 diabetes, the hazard ratio for coded acute heart failure was 0.431. The estimate for ketoacidosis was 0.683, but its confidence interval crossed 1. No analysis showed an advantage for acute pancreatitis[1].

The first 90 days raise the main doubt

The association was strongest immediately after treatment selection. In separately matched interval analyses, the hazard ratio for coded acute heart failure was 0.262 during days 1 to 90 and 0.507 during days 91 to 1,825. The later interval still favored the GLP-1 group in the records, but the gap was much smaller[1].

A treatment usually needs time to alter long-term clinical risk. A very large separation near the index date can signal treatment channeling: clinicians may have steered people with greater illness burden toward one drug class. The authors identify this as a plausible explanation but correctly state that their analysis cannot prove it[1].

The proportional-hazards assumption was also not met for the acute-heart-failure outcome. That matters because one hazard ratio then compresses a changing relationship into a single number. Reporting early and later windows made that problem visible rather than hiding it[1].

Matching did not make this a randomized trial

Propensity matching balances variables that are recorded and included in the model. It cannot balance severity, treatment intent, adherence, access, or clinical details that are missing or poorly coded. Albumin, bilirubin, and body mass index remained imbalanced after matching, in a direction favoring the GLP-1 receptor agonist group[1].

The study describes itself as initiation-based, but a true new-user design was not verified. Prior exposure can distort baseline risk and the timing of outcomes. Diagnoses and deaths were obtained from routine records, so coding practices and incomplete capture can also differ between systems or treatment groups[1].

The GLP-1 exposure was a class mixture. The results do not isolate semaglutide from liraglutide or dulaglutide in the primary comparison, and they do not test tirzepatide. A class-level records association should not be rewritten as a molecule-specific effect.

Randomized heart-failure evidence points elsewhere

DAPA-HF randomly assigned 4,744 patients with heart failure and reduced ejection fraction to dapagliflozin or placebo on top of recommended therapy. Dapagliflozin reduced the primary composite of worsening heart failure or cardiovascular death, with consistent results in patients with and without diabetes[2].

EMPEROR-Reduced randomized 3,730 patients with reduced-ejection-fraction heart failure to empagliflozin or placebo. Empagliflozin reduced the combined risk of cardiovascular death or hospitalization for heart failure, an effect driven mainly by fewer hospitalizations[3].

Those trials answer a treatment-effect question because treatment was randomized. The TriNetX analysis asks whether outcomes recorded after two different prescribing choices were associated after statistical matching. Its authors explicitly state that their findings do not challenge randomized evidence supporting SGLT2 inhibitors in heart failure and do not justify treatment-preference conclusions[1].

What this study is useful for

The cohort identifies a population worth studying directly: people who have both metabolic liver disease and heart failure. It also shows why timing checks matter in real-world evidence. When an apparent benefit is largest in the first 90 days and weakens later, a credible report should test channeling and residual confounding before using causal language.

A future comparative trial could define heart-failure phenotype, liver disease severity, diabetes status, and prior drug exposure at baseline, then prespecify cardiovascular and liver outcomes. Until then, readers can use our semaglutide evidence overview and semaglutide regulation overview without treating this cohort as a prescribing contest.

Treatment choice for heart failure, diabetes, obesity, kidney disease, or metabolic liver disease depends on individual clinical details and current guidelines. This study does not support switching between drug classes without a qualified clinician or a review of the full treatment plan[1].

**Medical disclaimer:** This article is for educational and journalistic purposes only and does not constitute medical advice. Peptides discussed may be classified as prescription medicines or research chemicals depending on your jurisdiction. Always consult a qualified healthcare professional before using any peptide product. PeptideMethods.com does not sell, distribute, or facilitate the sale of any peptide product.

Frequently asked

Did GLP-1 drugs outperform SGLT2 inhibitors in a randomized trial?

No. This was a retrospective propensity-matched records study. It cannot establish superiority or a preferred treatment.

Why is the first 90-day result suspicious?

The very large early separation weakened later. That timing is compatible with treatment channeling or residual baseline differences rather than a rapid drug effect.

Does the result overturn heart-failure evidence for SGLT2 inhibitors?

No. Randomized trials support dapagliflozin and empagliflozin in heart failure, and the cohort authors explicitly said their exploratory analysis does not challenge that evidence.

Sources

  1. [1]GLP-1 receptor agonists versus SGLT2 inhibitors in MASLD with heart failure (PMID 42969666)Tier 1 · primary↩
  2. [2]DAPA-HF randomized trial (PMID 31535829)Tier 1 · primary↩
  3. [3]EMPEROR-Reduced randomized trial (PMID 32865377)Tier 1 · primary↩

No revisions yet. First published .

About the editorial team

PeptideMethods is written and edited by the PeptideMethods Editorial Team and published by Digital Compass Group Ltd. The team is not made up of medical professionals; every health, regulatory or dosage claim on the site is tied to a primary source and is not a substitute for advice from a qualified clinician.

See our editorial policy and methodology for how we research, source and verify.

Read the pillars