Global uptake of GLP-1 receptor agonists in obese, non-diabetic patients with cardiovascular disease

Graphical Abstract

Glucagon-like peptide-1 receptor agonists (GLP-1RA) have transitioned from their traditional role in glycemic control and weight loss to therapies offering significant cardiovascular benefit. The 2023 Semaglutide Effects on Cardiovascular Outcomes in People with Overweight or Obesity (SELECT) trial¹ demonstrated significant cardiovascular benefits with GLP-1 RA use in obese, non-diabetic patients with established cardiovascular disease (CVD). These findings align with evidence from bariatric surgery, which similarly reduces CV events and mortality,² supporting weight loss, via caloric deficit, as a key driver of observed benefits, whether achieved surgically or pharmacologically. However, despite this paradigm-shifting evidence, insurance coverage for non-diabetic indications remains restrictive. Widespread reimbursement for obesity or cardiovascular prevention would entail substantial healthcare costs, and access remains uneven. Thus, many patients who might benefit from these therapies are unable to receive them, creating a growing disconnect between trial evidence and real-world practice. Previous studies have shown persistently low prescription rates of GLP-1 RA for both diabetes and obesity,^3,4 likely driven by structural inequities and limited access.⁵ Given the landmark findings of SELECT, we aimed to assess its impact on real-world uptake of GLP-1 RA in obese, non-diabetic patients with established CVD.

We conducted a retrospective cohort study utilizing the TriNetX global collaborative

Network database comprised of de-identified electronic health records from 148 healthcare organizations (HCOs). The network includes data on over 160 million patients, with 70 HCOs and approximately 119 million individuals based in the United States. TriNetX harmonizes data using the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM), RxNorm, Anatomical Therapeutic Chemical classification, and Current Procedural Terminology codes. Demographic variables include self-reported sex, race, and ethnicity.⁶ This database is compliant with the Health Insurance Portability and Accountability Act as it only provides de-identified aggregate-level patient data.⁶ Therefore, the study was exempt from institutional review board approval. TriNetX has been employed in numerous cardiovascular outcome research.^7,8

We identified a ‘SELECT-equivalent’ population based on the ICD-10-CM codes.

We included adults aged ≥45 years with a first recorded obesity [ICD-10-CM: E66 or a body mass index (BMI) ≥ 27 kg/m²] and established CVD but without diabetes (ICD-10-CM: E08–13), between 11 November 2017, and 6 May 2025. Established CVD included myocardial infarction (MI) (ICD-10-CM: I21–22), any stroke (ICD-10-CM: I60–63 or I67.82), or peripheral arterial disease (PAD) (ICD-10-CM: I73.9), consistent with SELECT’s CVD inclusion criteria. Patients were excluded if they received any glucose-lowering medications (metformin, dipeptidyl peptidase 4 inhibitor, sulfonylurea, sodium–glucose cotransporter-2 inhibitors, thiazolidinediones, or insulin) or known contraindications to GLP-1 RA, including end-stage renal disease (ICD-10-CM: N18.6), defined as a eGFR of <15 mL/min/1.73m², or dialysis dependence (ICD-10-CM: Z99.2), pancreatitis, or multiple endocrine neoplasia syndromes. Although semaglutide was the only agent studied in SELECT, we also investigated tirzepatide, dulaglutide, and liraglutide in our analysis because these agents are commonly prescribed off-label for CV risk reduction based on extrapolated data from diabetes trials (Dulaglutide [NCT01394952]; Liraglutide [NCT01179048]) and are often used in real-world settings due to insurance coverage limitations or medication availability. Additionally, tirzepatide is currently under investigation in an ongoing CV outcome trial (NCT04255433).

We quantified the annual incidence of incident GLP-1 RA prescriptions (semaglutide, tirzepatide, dulaglutide, and liraglutide) across seven intervals. The first six were consecutive 1-year periods commencing November 11, from 2017 through 2022, namely the pre-SELECT period. The final interval spanned the post-SELECT publication period: 11 November 2023, to 6 May 2025. Incidence was defined as the proportion of eligible, previously untreated individuals initiating a GLP-1 RA during each interval. Uptake trends were stratified by age (age ≥75 vs. <75), sex, and race. Baseline characteristics of users vs. non-users were compared using independent t-tests for continuous variables and χ² tests for binary variables, with two-sided P < 0.05 indicating statistically significant. All analyses were conducted using the TriNetX Analytics Platform. The analysis was conducted on 6 May 2025.

Of 298 344 SELECT-equivalent patients (mean age, 66.7 ± 12.2 years; females, 46.6%; White,74.5%; BMI, 32.1 kg/m²; MI, 55.4%; stroke, 38.8%; PAD, 20.7%), 10 128 (3.4%) initiated GLP-1RA (mean age, 57.5 ± 9.7 years; age ≥75 years, 14.6%; females; 61.8%; Black 13.8%). Among incident prescriptions, semaglutide (68.9%) and tirzepatide (27.9%) were the most frequently prescribed agents. Compared with non-users, GLP-1 RA users were younger (57.5 ± 9.7 vs. 67.0 ± 12.1;P < 0.0001), more likely to be female (61.8%vs45.8%; P < 0.0001), had higher BMI (36.4 ± 7.2 vs. 32.0 ± 5.1;P < 0.0001), more prevalent obstructive sleep apnea (OSA) (32.5% vs. 16.9%; P < 0.0001), less prevalent CKD (6.8% vs. 11.2%; P < 0.0001), and similar HbA1c (5.6 ± 0.4 vs. 5.6 ± 0.4). Baseline CVD was comparable between two groups.

Annual incidence trends (Figure 1) revealed a marked increase in GLP-1 RA prescriptions within the SELECT-equivalent population, with a notable doubling in the post-SELECT era (2.5%) when compared with the pre-SELECT period (1.2%). This post-SELECT rise was predominantly driven by uptake in semaglutide (1.9%) and tirzepatide (1.3%), whereas liraglutide use declined (0.1%) and dulaglutide use plateaued (0.2%). The increase in uptake was consistent across subgroups but was most pronounced among patients aged <75 years (3.6%) and females (3.2%) in the post-SELECT period.

Figure 1.

Annual incidence of GLP-1 RA global uptake in the SELECT-equivalent population.

In this real-world analysis, we observed a marked acceleration in global GLP-1 RA uptake following the publication of SELECT in non-diabetic obese individuals with established CVD, primarily driven by semaglutide and, increasingly, tirzepatide. The growing adoption of tirzepatide post-SELECT may partly reflect its recent approval for OSA in December 2024, following the pivotal SURMOUNT-OSA phase 3 trial (NCT05412004). Notably, nearly one-third of GLP-1 RA users had a diagnosis of OSA, which may have influenced prescribing patterns.

Despite a doubling uptake post-SELECT, overall GLP-1RA initiation remained low (<4%), highlighting a persistent treatment gap. This underutilization in this high-risk population is concerning and consistent with previous findings in broader populations with obesity and diabetes.^3,4 Several factors may contribute to this gap, including restrictive insurance coverage and high out-of-pocket costs, limited provider awareness of expanded indications, socioeconomic disparities, and ongoing medication shortages.⁵ Compounding these challenges, the European Medicines Agency recently reported widespread shortages of GLP-1 RAs across the EU, driven by surging demand, off-label use, and manufacturing constraints.⁹ Notably, although the rising uptake of GLP-1 RA was also observed in older adults (age ≥75), only 1% of this subgroup initiated therapy. Factors such as sarcopenia, falls, and nutritional deficiencies may contribute to clinician reluctance to prescribe GLP-1 RAs in older adults.¹⁰ This prescribing pattern highlights an important area for future research to evaluate the safety, efficacy, and cost-effectiveness of GLP-1 RAs in geriatric populations.

This study has several limitations. First, the use of de-identified data precluded investigations of off-label prescriptions. Second, prescription data were limited to encounters within participating HCOs within TriNetX, potentially underestimating uptake. Additionally, the predominance of academic institutions in the dataset may limit the generalizability of findings to community-based or non-academic settings. Third, we were unable to evaluate the impact of social determinants of health, patient beliefs, and insurance restrictions, due to limitations of the database.

In conclusion, GLP-1RA uptake substantially increased among obese, non-diabetic patients with established CVD following SELECT. However, significant gaps persist, and efforts are urgently needed to ensure equitable use of these therapies and improve cardiovascular outcomes in this high-risk population.

Data availability

The data supporting the study findings are available from the TriNetX Analytics Network (www.trinetx.com) which includes subscription only access to deidentified data.