Glucagon-Like Peptide-1 Receptor Agonists and Cardiovascular Outcomes in Patients With Atherosclerotic Cardiovascular Disease and Obesity Without Diabetes
Akbar UA, Mondal A, Vorla M, Alruwaili W, Lacoste J, Thyagaturu H
The American Journal of Cardiology (2026) · n=14,844
In a propensity-matched real-world cohort of 14,844 non-diabetic adults with established atherosclerotic cardiovascular disease and BMI ≥27, GLP-1RA initiation was associated with a 32% relative reduction in 5-year all-cause mortality (HR 0.68; 95% CI 0.53-0.88; p=0.003) — extending the SELECT trial signal across multiple GLP-1 agents in real-world practice.
This is a propensity-matched target-trial-emulation study using the TriNetX US Collaborative Network electronic health record network, designed to test whether the SELECT-trial cardiovascular benefit of semaglutide in non-diabetic obesity generalises to the broader class of GLP-1 receptor agonists prescribed in real-world US practice.
The investigators identified adults aged ≥45 with established atherosclerotic cardiovascular disease (history of myocardial infarction, stroke, or coronary or peripheral revascularisation), BMI ≥27 kg/m², and no type 2 diabetes who initiated any GLP-1 RA (liraglutide, semaglutide, dulaglutide, or exenatide) between January 2010 and December 2025, and matched them 1:1 to non-initiators. 14,844 propensity-matched patients (7,422 per group; median age 63, IQR 55-71; 64% women) were followed for up to 5 years.
The primary outcome was all-cause mortality; secondary outcomes were acute myocardial infarction, stroke, and heart failure hospitalisation, analysed with Cox proportional hazards and Fine-Gray competing-risk models. GLP-1 RA initiation was associated with lower all-cause mortality (HR 0.68; 95% CI 0.53-0.88; p=0.003), acute MI (sHR 0.63; 95% CI 0.41-0.98; p=0.040), and HF hospitalisation (sHR 0.61; 95% CI 0.39-0.95; p=0.028). Stroke risk did not reach significance (sHR 0.76; 95% CI 0.52-1.10; p=0.146).
This is observational data from electronic health records, not a randomised trial — propensity matching reduces but does not eliminate residual confounding, and the result should be interpreted as supportive of SELECT rather than independent confirmation. TriNetX captures dispensing and claims-level data; medication adherence, dose, and treatment duration are imperfectly measured, and patients who initiated GLP-1RAs may differ systematically from non-initiators in unmeasured ways (motivation, healthcare engagement, lifestyle change capacity) — the classic healthy-adherer bias.
The study aggregates across heterogeneous GLP-1 agents (liraglutide, semaglutide, dulaglutide, exenatide) at heterogeneous doses for heterogeneous indications, so the headline number is a class-average effect rather than a per-drug effect. Stroke did not reach significance, and the confidence intervals on the MI and HF outcomes are wide enough that the effects could be substantially smaller than the point estimates.
The trial-emulation framework is sound but cannot rule out reverse causation (sicker patients less likely to be initiated). Sensitivity analysis including diabetic patients (n=31,910 matched pairs) replicated the direction.
07·Member discussion
No member discussion yet.
Member-only conversation lives here — cycle notes, practitioner commentary, pattern-matching. Be the first paying member to start the thread.