Glucagon‐like peptide‐1 receptor agonists and venous thromboembolism: an unresolved question for Australian prescribers

Since their Pharmaceutical Benefits Scheme (PBS) listing for type 2 diabetes mellitus (T2DM), glucagon‐like peptide‐1 receptor agonists (GLP‐1RAs) have become one of Australia's fastest‐growing drug classes, with nearly 2.4 million PBS prescriptions dispensed in 2023 alone. ¹ With the Pharmaceutical Benefits Advisory Committee recently recommending PBS listing of semaglutide for adults with established cardiovascular disease and obesity, this population will expand further still. ²

The cardiovascular benefits of GLP‐1RAs are well established, with meta‐analyses demonstrating consistent reductions in major adverse cardiovascular events and stroke. ³ These benefits are mediated partly through attenuation of platelet activation, reduction in plasminogen activator inhibitor‐1 and improved endothelial function. ⁴ , ⁵ However, whether these mechanisms extend meaningfully to venous thromboembolism (VTE) remains unresolved. Obesity is an established VTE risk factor, and any apparent protective association could reflect metabolic improvement rather than a direct drug effect. ⁶ Observational data are nonetheless encouraging: a large target trial emulation of over 540 000 patients with T2DM demonstrated approximately 20% lower VTE incidence with GLP‐1RAs compared with DPP‐4 inhibitors, independent of baseline body mass index. ⁷ Whether this reflects a direct drug effect or is mediated by weight loss or glycaemic improvement cannot be determined without causal mediation analysis, ⁸ and residual confounding cannot be excluded.

Against this, a 2025 meta‐analysis of 27 placebo‐controlled randomised trials encompassing over 84 000 patients found no statistically significant reduction in overall VTE risk, though a signal toward lower pulmonary embolism (PE) risk was observed. ⁹ A separate meta‐analysis of 39 trials reported an increase in deep vein thrombosis (DVT) risk (odds ratio (OR) 1.64, 95% confidence interval (CI) 1.14–2.36), which was amplified with treatment duration exceeding 78 weeks (OR 2.32, 95% CI 1.49–3.60), but no change in PE risk. ¹⁰

The discrepancy between the two meta‐analyses likely reflects several methodological differences: while Chiang et al. restricted their analysis to placebo‐controlled trials, Liu et al. included active comparators, with the DVT signal driven largely by cardiovascular outcome trials, which had longer follow‐up and more rigorous adverse event ascertainment. ⁹ , ¹⁰ Neither analysis distinguished proximal from distal DVT, an important omission given that distal events are clinically heterogeneous and variably treated and carry lower embolic risk. ¹¹ One potential explanation for the increased DVT risk but not that of PE is that GLP‐1RAs can induce hypodipsia leading to haemoconcentration as well as sarcopenia, which may impair the calf muscle pump. ¹² , ¹³ Together, these effects would act principally in the low‐shear venous environment, generating predominantly distal thrombi less likely to propagate or embolise. While speculative, this hypothesis warrants further examination.

A search of the Therapeutic Goods Administration's (TGA's) Database of Adverse Event Notifications in March 2026 identified 26 suspected thromboembolic events across four GLP‐1RAs in Australia since their respective PBS listings – 11 pulmonary emboli and 15 venous thromboses – though the PE‐to‐DVT ratio does not mirror the divergent signals observed in the meta‐analyses. ¹⁴ Adverse event underreporting to the TGA is well recognised, with fewer than 5% of reactions estimated to be captured. ¹⁵

No adequately powered trials have evaluated VTE as a pre‐specified outcome. VTE risk is not routinely considered when initiating GLP‐1RAs, even in patients with established VTE risk factors. The mixed evidence makes it difficult to counsel patients being commenced on GLP‐1RAs about VTE risk, and even modest relative risk can translate to significant event numbers at the population level. We call for VTE to be incorporated as a pre‐specified endpoint in future trials, including those of dual and triple receptor agonists. Given the rarity of VTE, however, even large trials may be underpowered, and observational studies using targeted PBS and hospital morbidity data linkage may offer the most informative real‐world estimates in this rapidly expanding patient population, provided confounding is appropriately addressed.

Data availability statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.