Tirzepatide compared to GLP-1 drugs and the risk of blood clots in people with obesity

Venous thromboembolism (VTE), encompassing deep vein thrombosis (DVT) and pulmonary embolism (PE), is a major global health concern. In Western populations, the annual incidence of VTE is estimated at approximately 1–2 cases per 1,000 person-years (1). The clinical impact of VTE extends beyond the acute event, as it is linked not only to a substantially increased risk of mortality but also to a higher incidence of subsequent arterial cardiovascular events (2). VTE may also result in serious long-term complications, including post-thrombotic syndrome and recurrent events, highlighting the importance of effective preventive strategies (3, 4).

Prior studies have shown that obesity confers more than a twofold higher risk of a first episode of VTE compared with normal weight (5), probably attributable to chronic inflammation and a hypercoagulable state (6). Weight reduction has also been associated with a lower risk of VTE (7, 8). In recent years, glucagon-like peptide-1 receptor agonists (GLP-1 RAs), originally developed for diabetes management, have also demonstrated substantial weight loss benefits (9). Moreover, GLP-1 RAs have been shown to exert anti-inflammatory effects (10), which may contribute to their cardiovascular and renal benefits (11). Observational studies have further suggested that GLP-1 RAs use, compared with dipeptidyl peptidase-4 inhibitors (DPP4is), may be associated with a reduced risk of VTE in individuals with type 2 diabetes (T2D) (12). More recently, tirzepatide, a dual glucose-dependent insulinotropic polypeptide (GIP) and GLP-1 receptor agonist, has been shown in a randomized controlled trial to achieve greater weight reduction than semaglutide (13). In another observational study, tirzepatide demonstrated superior cardiovascular outcomes compared with GLP-1 RAs in patients with T2D (14).

However, evidence remains limited regarding whether tirzepatide use is associated with a reduced incidence of VTE compared with GLP-1 RAs among patients with obesity. To address this knowledge gap, we conducted a real-world cohort study using the TriNetX database to evaluate the comparative effectiveness of tirzepatide versus GLP-1 RAs in reducing the 1-year incidence of VTE in patients with obesity.

In this large, real-world cohort study of 701,374 patients with obesity, initiation of tirzepatide was associated with a significantly lower 1-year risk of VTE compared with GLP-1 RAs. The benefit was mainly driven by reductions in PE and all-cause mortality, whereas the risk of DVT did not differ significantly between groups. These associations were robust across multiple sensitivity analyses, including negative control outcomes, and were consistent across clinically relevant subgroups, with more pronounced effects in patients aged 18–64 years and those without comorbid conditions such as heart failure. Collectively, these findings suggest a potential association between tirzepatide use and a lower risk of VTE compared with GLP-1 RAs.

In a large real-world cohort study including 540,258 patients with T2D, GLP-1 RAs use was associated with a significantly lower risk of VTE compared with DPP-4 inhibitors (HR 0.78; 95% CI 0.73–0.83). Similarly, a recent meta-analysis of 27 randomized controlled trials involving 84,003 patients reported a trend toward reduced VTE risk with GLP-1 RAs (RR 0.70; 95% CI, 0.46–1.07), with a more pronounced benefit observed for pulmonary embolism (RR, 0.60; 95% CI, 0.39–0.94) in patients with T2D and obesity (25). Our findings align with these observations: tirzepatide was associated with a significant reduction in PE compared with GLP-1 RAs, whereas no significant difference was observed for DVT, although the directional trend was consistent. These findings are broadly consistent with prior literature suggesting a more pronounced association for PE than for DVT. Taken together, these parallels suggest that the protective effect of incretin-based therapies on thromboembolic risk may be driven primarily by reductions in PE rather than DVT. Our findings may be partly explained by the superior cardiometabolic benefits of tirzepatide compared with GLP-1 RAs, including greater weight loss and improved cardiovascular risk profiles (13, 14). However, while prior trials mainly focused on glycemic and cardiovascular outcomes, evidence regarding VTE risk has been limited. Our study therefore provides novel real-world evidence suggesting that tirzepatide may confer additional protection against thromboembolic events beyond that offered by GLP-1 RAs.

The observed association between tirzepatide use and lower VTE risk may be influenced by differences in weight reduction and metabolic improvements between treatment groups (26). However, these factors were not fully captured in the database and may represent sources of residual confounding. Obesity is a well-established risk factor for VTE, conferring more than a twofold higher risk compared with normal weight individuals, largely mediated through chronic inflammation and a hypercoagulable state (27, 28). Weight reduction itself has been associated with a significantly lower risk of VTE (7, 8). Beyond weight loss, tirzepatide has been shown in prior studies to have anti-inflammatory and metabolic effects (29), improve endothelial function, and may attenuate platelet activation (30, 31); however, whether these mechanisms translate into reduced thromboembolic risk remains uncertain.

Prior evidence suggests that the potential association between incretin-based therapies and VTE may be more pronounced for PE than for DVT (12). In a recent meta-analysis of randomized controlled trials, GLP-1 RAs were associated with a greater reduction in PE compared with DVT (25). Several factors may contribute to this difference. PE is often a clinically acute and severe condition, which may be more consistently diagnosed and recorded in electronic health records. In contrast, DVT may be underdiagnosed, particularly in asymptomatic or mildly symptomatic cases, potentially attenuating observed associations in real-world data.

Importantly, this differential ascertainment may introduce detection bias, which could partly explain the observed discrepancy between pulmonary embolism and deep vein thrombosis. As such, this inconsistency may limit the internal validity of the findings and should be interpreted with caution. In addition, differences in underlying pathophysiology between PE and DVT have been proposed, although these mechanisms remain incompletely understood. Therefore, the observed discrepancy should be interpreted with caution and requires further investigation.

This study leveraged a large, multicenter real-world database and employed propensity score matching to strengthen methodological rigor and reduce the influence of measured confounders. In patients with obesity and no prior history of VTE, tirzepatide use was associated with a significantly lower 1-year incidence of VTE, along with reduced risks of PE and all-cause mortality compared with GLP-1 RAs therapy. These findings provide encouraging evidence that tirzepatide may be associated with a lower risk of thromboembolic events. However, given the inherent limitations of observational research, confirmation through randomized controlled trials and long-term outcome studies is essential to validate these associations and to better define their clinical implications.

In the absence of established pharmacologic strategies for the primary prevention of VTE in patients with obesity, our findings suggest that tirzepatide may provide an additional protective benefit beyond its established effects on glycemic control and weight reduction. Current clinical guidelines primarily recommend GLP-1RAs for patients with obesity and T2D based on cardiovascular and metabolic benefits (32, 33); however, their role in VTE prevention has not been defined. By demonstrating a significant reduction in PE and overall VTE risk, our study highlights the potential of tirzepatide to address an important unmet clinical need in this high-risk population. Overall, the findings of this study should be interpreted in the context of several limitations, including residual confounding, potential exposure misclassification, and modest effect sizes. Therefore, the results should be considered hypothesis-generating rather than definitive evidence of a causal relationship. These findings may help generate hypotheses for future studies and may inform the design of prospective trials. Notably, the absolute risk reduction observed in this study was small, which further limits the immediate clinical applicability of these findings.

This study has several limitations. First, as TriNetX is a registry-based database, the risk of misidentification and underrepresentation exists, particularly among individuals with milder disease or limited healthcare utilization, which may affect the generalizability of our findings. In addition, the use of diagnostic codes to define exposures, covariates, and outcomes raises the possibility of misclassification bias. In particular, differences in treatment-related weight loss and metabolic improvements between tirzepatide and GLP-1 RA could not be fully accounted for, which may have influenced the observed associations. To mitigate this concern, we performed negative control outcome analyses with unrelated clinical conditions, which revealed no significant differences between treatment groups, suggesting that bias from diagnostic coding was likely minimal. Additionally, subgroup analyses were exploratory, and formal interaction testing was not performed. Therefore, potential effect modification across patient subgroups cannot be definitively established.

Second, the duration of tirzepatide and GLP-1 RA therapy could not be reliably ascertained from the database, limiting our ability to evaluate the effects of treatment duration or long-term exposure. In addition, calendar time was not explicitly adjusted for in the propensity score model. Given that tirzepatide is a relatively new therapy, prescribing patterns and patient characteristics may have evolved over the study period. Although an active-comparator, new-user design was applied within a shared time frame, residual temporal confounding cannot be excluded and may have influenced the observed associations. Third, despite the inclusion of multiple clinically relevant covariates, some potentially important variables may not have been available or fully captured in the database, which may result in residual confounding. To further explore this possibility, we calculated E-values. The E-values observed in this study were modest in magnitude, suggesting that the observed associations could potentially be explained by relatively weak unmeasured confounding. Therefore, these findings should be interpreted with caution. Fourth, the distribution of continuous variables could not be formally assessed, as the TriNetX platform provides summary statistics primarily as means and standard deviations. Therefore, non-normal distributions may not have been fully characterized. Fifth, death was not treated as a competing risk. Given that all-cause mortality differed slightly between groups, this may have influenced the estimated risk of VTE.

Sixth, residual imbalance in HbA1c persisted after propensity score matching, indicating potential differences in baseline metabolic status between groups. Although sensitivity analyses stratified by HbA1c levels yielded consistent results, residual confounding related to glycemic control and overall metabolic health cannot be excluded. Seventh, GLP-1RAs were analyzed as a class, which may introduce heterogeneity due to differences in pharmacologic profiles and clinical indications across agents. Although sensitivity analyses restricted to semaglutide yielded consistent results, residual heterogeneity and confounding by indication cannot be excluded. Eighth, exposure misclassification is also possible, as treatment adherence, discontinuation, and switching could not be directly assessed within the database. Although an on-treatment analysis was performed, this approach relied on indirect definitions and may not fully reflect actual treatment patterns. Ninth, the PSM approach was subject to limitations related to data availability within the TriNetX platform. Missing data were not explicitly imputed, and analyses were conducted based on available information, which may have influenced cohort selection and matching performance. In particular, laboratory variables such as HbA1c were not available for all patients, and incomplete data may have introduced selection bias or residual confounding. Furthermore, although covariates included in the propensity score model were selected based on clinical relevance and prior literature, the availability of certain variables may have been inconsistent across patients. As a result, the matching process may not have fully accounted for all clinically important differences between groups.

Finally, owing to inherent limitations of the database, specific causes of mortality were unavailable, which may have led to misclassification of clinical endpoints.

Tirzepatide use was associated with a modestly lower 1-year incidence of VTE compared with GLP-1 receptor agonists in individuals with obesity. However, the absolute risk difference was small, and the clinical significance of this finding remains uncertain.

These results should be interpreted as associative and hypothesis-generating rather than indicative of a causal or clinically actionable effect. Further prospective studies are required to confirm these findings and to determine their potential clinical relevance.

This research was conducted without external funding or financial support, ensuring full independence in study design, data collection, analysis, interpretation, and manuscript preparation.

The author(s) declared that financial support was not received for this work and/or its publication.

The data analyzed in this study is subject to the following licenses/restrictions: the data are subject to TriNetX data use agreements and privacy regulations. Individual-level patient data are de-identified and cannot be downloaded or shared. Only aggregated results generated within the TriNetX platform are accessible to investigators. Requests to access these datasets should be directed to https://trinetx.com/↗.

The studies involving humans were approved by Institutional Review Board of Chi Mei Hospital (IRB No. 11402-E02). The studies were conducted in accordance with the local legislation and institutional requirements. The ethics committee/institutional review board waived the requirement of written informed consent for participation from the participants or the participants’ legal guardians/next of kin because the requirement for informed consent was waived owing to the retrospective design and the use of de-identified, aggregated data.

J-YW: Writing – original draft, Conceptualization, Data curation. K-WL: Investigation, Methodology, Writing – original draft. S-CH: Writing – original draft, Writing – review & editing. H-YC: Writing – original draft, Conceptualization. Y-ML: Writing – original draft, Writing – review & editing.

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmed.2026.1820366/full#supplementary-material↗