Dual Therapy vs. Single Anticoagulation After Acute Deep Vein Thrombosis Iliac Stenting: A Doubly Robust Multicenter Study

Abstract

Background

The optimal antithrombotic strategy after iliac vein stenting for acute iliofemoral deep vein thrombosis (IFDVT) remains uncertain. We compared dual antithrombotic therapy (DAT) with single anticoagulation (SAC) in patients with May-Thurner syndrome-related IFDVT.

Methods

This multicenter retrospective cohort included 450 patients undergoing successful iliac vein stenting between 2018 and 2022. At discharge, 186 received DAT and 264 received SAC. Inverse probability of treatment weighting, doubly robust estimation, and time-varying exposure modeling were used. The primary outcome was primary stent patency. Secondary outcomes included primary-assisted and secondary patency, chronic post-thrombotic symptoms requiring treatment, recurrent ipsilateral DVT, and bleeding.

Results

During a median follow-up of 46 months, DAT was associated with a lower risk of primary patency loss than SAC (sHR, 0.58; 95% CI, 0.41–0.82; p = 0.002); estimated 5-year primary patency was 81.8% versus 67.5%. Similar associations were observed for primary-assisted (sHR, 0.52; 95% CI, 0.35–0.78; p = 0.001) and secondary patency loss (sHR, 0.45; 95% CI, 0.28–0.71; p < 0.001). DAT was also associated with fewer chronic post-thrombotic symptoms requiring treatment (sHR, 0.61; p = 0.005) and recurrent DVT (sHR, 0.48; p = 0.019). Major bleeding did not differ significantly (HR, 1.28; p = 0.624), whereas clinically relevant non-major bleeding was more frequent with DAT (HR, 2.35; p < 0.001).

Conclusions

During anticoagulant-treated follow-up, DAT exposure was associated with higher long-term stent patency and fewer clinically treated chronic post-thrombotic symptoms and recurrent DVT than SAC exposure, but with more clinically relevant non-major bleeding. The study does not determine the optimal duration of anticoagulation or the effectiveness of subsequent antiplatelet monotherapy.

1. Introduction

Deep vein thrombosis (DVT) remains a major global health burden, with an annual incidence of approximately 50–120 cases per 100,000 person-years in high-income countries and a marked increase with advancing age.^1,2 Among lower-extremity DVT presentations, iliofemoral deep vein thrombosis (IFDVT) is associated with severe acute symptoms and the highest long-term risk of post-thrombotic syndrome (PTS), which may affect up to 50% of patients despite anticoagulation.^3,4 May-Thurner syndrome (MTS), caused by compression of the left common iliac vein by the overlying right common iliac artery, is an important anatomical substrate for acute left-sided IFDVT. ⁵

Management of acute IFDVT has evolved from anticoagulation alone toward early thrombus removal in selected patients. ⁶ Catheter-directed thrombolysis and pharmacomechanical thrombectomy, followed by iliac vein stenting when significant MTS-related obstruction is present, may improve venous patency and reduce moderate-to-severe PTS.^7,8 Current Society for Vascular Surgery and American Venous Forum guidelines support early thrombus removal in patients with acute IFDVT and a low bleeding risk. ⁹ However, the optimal antithrombotic regimen after successful iliac vein stenting remains uncertain 10.

Although anticoagulation is the cornerstone of post-stenting therapy, the additional benefit of antiplatelet treatment remains unclear. International consensus statements and Delphi studies have not reached agreement regarding the optimal anticoagulant, treatment duration, or role of adjunctive antiplatelet therapy.^10,11 The ARIVA randomized trial reported comparable 6-month patency with aspirin plus rivaroxaban and rivaroxaban alone in patients with post-thrombotic syndrome, ¹² while subsequent commentary emphasized that optimal post-stent medical therapy remains unresolved. ¹³ These findings may not be directly applicable to patients treated for acute MTS-related IFDVT, and long-term comparative data on dual antithrombotic therapy (DAT) versus single anticoagulation (SAC) remain limited.

This multicenter retrospective cohort study evaluated the associations of DAT and SAC with long-term primary, primary-assisted, and secondary stent patency, clinically treated chronic post-thrombotic symptoms, ipsilateral DVT recurrence, and bleeding outcomes after iliac vein stenting for acute MTS-related IFDVT. Doubly robust estimation with time-varying exposure modeling was used to account for baseline differences and treatment changes during follow-up.

2. Methods

2.1. Study Design

This multicenter, retrospective cohort study was conducted to evaluate the impact of DAT versus SAC on long-term outcomes following iliac vein stenting for acute DVT. The research utilized real-world electronic medical records (EMR) aggregated from participating tertiary referral centers between January 2018 and December 2022. The study protocol adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Ethical approval was granted by the Institutional Review Board (IRB) of the principal investigator’s institution and all corresponding participating centers (No: HM-2026-011). In accordance with the Declaration of Helsinki and local ethical regulations, the requirement for written informed consent was waived by the IRBs due to the purely retrospective, non-interventional nature of the study. All patient data were de-identified prior to analysis.

2.2. Study Population

The study cohort was identified through a systematic query of the clinical data warehouses across all participating institutions. Structured data filters were applied using International Classification of Diseases, Tenth Revision (ICD-10) diagnostic codes specific to acute deep vein thrombosis of the lower extremity (I80.2) and iliac vein compression syndrome (I87.1). The initial screening included all adult patients evaluated between January 1, 2018, and December 31, 2022. Subsequently, potential participants underwent a rigorous validation process beginning with algorithmic filtering based on predefined baseline characteristics, followed by a detailed manual review of electronic medical records, operative notes, and angiographic imaging by two independent vascular specialists. Any discrepancies were resolved through consensus or adjudication by a senior vascular surgeon.

Inclusion criteria comprised all of the following: (1) age of 18 years or older; (2) confirmed diagnosis of May-Thurner syndrome concomitant with acute lower extremity deep vein thrombosis, defined as symptom onset within 14 days prior to admission, verified via duplex ultrasound or contrast venography according to the current European Society for Vascular Surgery (ESVS) guidelines ¹⁴ ; (3) receipt of one-stop endovascular treatment, consisting of catheter-directed thrombolysis or pharmacomechanical thrombectomy (CDT/PMT) combined with iliac vein stenting, achieving “technical success” in accordance with the Society of Interventional Radiology (SIR) reporting standards, ¹⁵ which was defined as complete stent coverage of the diseased segment, adequate anchoring in healthy venous segments proximally and distally, residual stenosis of less than 30%, and brisk in-line flow without contrast stagnation; (4) explicit prescription of either a SAC regimen or DAT upon hospital discharge; (5) availability of at least one routine imaging follow-up, such as duplex ultrasound or venography, within 12 months post-discharge to mitigate surveillance bias; (6) a minimum of 3 years of complete clinical follow-up data, unless the primary endpoint or death was reached earlier. Exclusion criteria encompassed any of the following: (1) prior history of ipsilateral deep vein thrombosis or the presence of established severe post-thrombotic syndrome (PTS); (2) isolated non-thrombotic iliac vein lesions without acute thrombosis; (3) limited life expectancy of less than 1 year, such as in patients with advanced metastatic malignancies; (4) absolute contraindications to therapeutic anticoagulation or antiplatelet agents, including active gastrointestinal bleeding, recent severe intracranial hemorrhage, or severe coagulopathy, in accordance with the American College of Chest Physicians (CHEST) guidelines ¹⁶ ; (5) missing crucial outcome or exposure data, or missing baseline procedural documentation that could not be handled by multiple imputation.

2.3. Exposures, Interventions, and Time Zero

To ensure the accuracy of the survival analysis and minimize immortal time bias, the index date (time zero) was strictly defined as the day the patient was discharged from the hospital with a prescribed antithrombotic regimen following successful stent implantation. This specific date served as the starting point for all subsequent time-to-event analyses. Rather than employing a rigid baseline grouping that assumes lifetime adherence and fails to capture real-world clinical dynamics, this study utilized a dynamic exposure tracking approach.

Patient medication trajectories were reconstructed using structured follow-up assessments at 1, 3, 6, and 12 months after discharge, supplemented by the exact dates of treatment changes documented between scheduled visits. ¹⁷ Medication exposure status was determined through a detailed review of clinic notes and outpatient prescription records within the participating centers’ systems. At each assessment, treatment status was classified as DAT, SAC, antiplatelet therapy alone, or no antithrombotic treatment. DAT was defined as concurrent administration of a therapeutic-dose oral anticoagulant and a single antiplatelet agent, whereas SAC was defined as oral anticoagulation without an antiplatelet agent. Antiplatelet therapy alone and no antithrombotic treatment were recorded as separate time-varying treatment states and were not classified as either DAT or SAC. These states were retained as separate categories in the time-varying models, while the prespecified primary treatment contrast remained DAT versus SAC. Accordingly, when anticoagulation was discontinued while aspirin or clopidogrel was continued, subsequent person-time was assigned to the antiplatelet-therapy-alone state rather than to DAT or SAC. Antiplatelet monotherapy was not combined with either anticoagulant-treated group, and the prespecified primary treatment contrast remained restricted to DAT versus SAC exposure.

A regimen was considered initiated, modified, or discontinued on the date explicitly documented in the outpatient clinical record or on the date a revised prescription was issued. Person-time was assigned according to the treatment actually received during each interval. When a patient transitioned from DAT to SAC, person-time before the transition was assigned to DAT and subsequent person-time was assigned to SAC. Similarly, intervals during which a patient received antiplatelet therapy alone or no antithrombotic treatment were assigned to those respective categories and were not carried forward under the preceding DAT or SAC exposure.

Changes in antithrombotic treatment following target-lesion reintervention were handled using the same time-updated approach. Patients initially receiving SAC who were escalated to DAT after balloon venoplasty, catheter-directed thrombolysis, mechanical thrombectomy, or additional stent placement contributed person-time to SAC until the documented date of treatment escalation and to DAT thereafter. Treatment escalation initiated during or after reintervention was not retrospectively assigned to the exposure period preceding the procedure.

SAC exposure was defined as the exclusive use of standard-dose oral anticoagulants, including direct oral anticoagulants (DOACs, such as rivaroxaban or dabigatran) or vitamin K antagonists (VKAs, primarily warfarin). Dual antithrombotic exposure was defined as the concurrent administration of a therapeutic-dose oral anticoagulant and a single antiplatelet agent. The antiplatelet regimen consisted of aspirin 100 mg once daily or clopidogrel 75 mg once daily, as documented in the discharge prescription and subsequent follow-up records. No routine antiplatelet loading dose was included in the post-discharge exposure definition. Furthermore, subgroup definitions were established within the DAT cohort according to the class of oral anticoagulant used, comprising DOAC-based DAT and VKA-based DAT, and the specific antiplatelet agent was recorded. ¹⁰ Comparisons between these subgroups were prespecified as exploratory analyses and included both bleeding and stent patency outcomes. Because of the limited subgroup sample size and number of patency-loss events, these analyses were not considered confirmatory efficacy comparisons.

2.4. Data Collection and Endpoints Definitions

Baseline clinical, laboratory, and procedural data were extracted from the institutional electronic health records by two independent clinical investigators using standardized electronic case report forms, with any discrepancies resolved by a senior vascular surgeon. A comprehensive array of candidate covariates was collected based on established clinical relevance and potential confounding effects on stent patency. Demographic and clinical history variables encompassed age, sex, body mass index, smoking history, prior documented hypercoagulable state (thrombophilia), preoperative D-dimer levels, pre-existing chronic venous insufficiency, and prior bleeding history. Prior bleeding history was defined as documented gastrointestinal, intracranial, genitourinary, or other clinically relevant bleeding requiring medical evaluation within 12 months before the index admission. Procedural variables included the extent of venous outflow tract involvement, specifically noting whether the stent extended distally across the inguinal ligament into the common femoral vein (CFV; representative pre- and post-stent fluoroscopic venographic images are shown in Supplemental Figure S1), and inflow tract involvement, defined as the proximal extension of the stent into the inferior vena cava. Furthermore, intraoperative technical details such as the utilization of adjunctive catheter-directed thrombolysis or pharmacomechanical aspiration, prophylactic balloon angioplasty, the specific type of stent deployed (dedicated venous versus generic stents), stent diameter, stent length, and the total number of stents implanted were systematically recorded.

The primary efficacy endpoints of this study were structured around the classic patency triad in accordance with the SVS and the Society of Interventional Radiology (SIR) reporting standards. ¹⁵ Primary patency was strictly defined as uninterrupted venous stent patency maintained without the need for any target lesion reintervention following the initial procedure. Primary-assisted patency was defined as the continuous patency maintained after a secondary prophylactic endovascular intervention, such as balloon venoplasty, was performed to address severe in-stent restenosis or impending occlusion discovered during surveillance, prior to complete total occlusion. Secondary patency was defined as the restoration and maintenance of patency following a successful reintervention, including thrombolysis or additional stenting, for a completely occluded stent. For the time-to-event analyses, each qualifying reintervention was attributed to the antithrombotic regimen being received immediately before the procedure. A prophylactic intervention for severe in-stent restenosis or impending occlusion was therefore attributed to the regimen in use immediately before the primary-assisted patency intervention, whereas recanalization of an occluded stent was attributed to the regimen in use immediately before the secondary patency intervention. Any escalation from SAC to DAT initiated during or after the procedure affected only subsequent person-time and did not retrospectively alter the exposure attribution of the qualifying event. The secondary efficacy endpoints included ipsilateral DVT recurrence, verified by duplex ultrasound, and clinically treated chronic post-thrombotic symptoms. Clinically treated chronic post-thrombotic symptoms were defined as persistent or newly documented ipsilateral lower-extremity swelling, heaviness, pain, venous skin changes, or venous ulceration occurring at least 3 months after the index DVT and resulting in continued prescription of class II medical compression stockings (20–30 mmHg) or venoactive medication. The outcome was identified through longitudinal review of outpatient records, follow-up documentation, and treatment prescriptions. Because standardized Villalta scores were not systematically available across the participating centers, this outcome was considered a treatment-based indicator of chronic post-thrombotic symptom burden rather than a formally adjudicated diagnosis of post-thrombotic syndrome.

Safety endpoints comprised hemorrhagic events classified according to the criteria set forth by the International Society on Thrombosis and Haemostasis (ISTH). ¹⁸ Major bleeding was defined as fatal bleeding, symptomatic bleeding occurring in a critical anatomical site or organ, or clinically overt bleeding resulting in a hemoglobin drop of 2.0 g/dL or more, or requiring the transfusion of two or more units of whole blood or red cells. All fatal hemorrhagic events were strictly adjudicated as major safety endpoints. Clinically relevant non-major bleeding (CRNMB) was defined as any overt bleeding event that did not meet the criteria for major bleeding but required medical intervention, prompted an unscheduled contact with a healthcare professional, or necessitated temporary cessation, delay, or dose adjustment of the antithrombotic therapy. ¹⁹ Potential bleeding events were identified through inpatient, emergency department, and outpatient records; laboratory and transfusion data; medication interruption or dose-adjustment records; and follow-up documentation of bleeding-related care received outside the participating centers. Events treated outside the participating centers were included only when sufficient supporting documentation was available for classification. Two clinical investigators independently adjudicated all suspected bleeding events according to the ISTH criteria, without access to the results of the primary patency analysis. Disagreements were resolved by a senior clinician. Major bleeding and CRNMB were analyzed separately as the time to the first event of each category.

2.5. Sample Size Calculation

The statistical power and adequacy of the sample size were evaluated based on established event rates from prior large-scale cohorts, which reported a 3-year stent occlusion rate ranging from 10% to 15% following iliac vein stenting for acute deep vein thrombosis. ²⁰ Assuming that DAT could confer a clinically meaningful relative risk reduction corresponding to a Hazard Ratio (HR) of approximately 0.60 to 0.65, a post-hoc power analysis was conducted to justify the retrospective cohort size. Utilizing the available multicenter cohort of approximately 450 eligible patients, and setting a two-sided type I error rate (alpha) of 0.05, the study possessed greater than 80% statistical power to detect this prespecified magnitude of effect. Furthermore, to account for the design effect and variance inflation inherent to inverse probability of treatment weighting (IPTW), as well as the event-per-variable (EPV) requirements for modeling multiple competing risks, a minimum initial sample size of 450 was confirmed to be necessary. This ensured model stability and prevented overfitting of the Fine-Gray models and DRE. All power and sample size estimations were performed utilizing the ‘powerSurvEpi’ package in R software.

2.6. Statistical Analysis

Missing baseline data, restricted to variables with less than 5% missingness, were addressed utilizing multiple imputation by chained equations to preserve the full analytical cohort and maximize statistical power. An inverse probability of treatment weighting (IPTW) approach was used to reduce treatment-selection bias. The propensity score model included all prespecified baseline demographic, clinical, anatomical, and procedural variables available before discharge, together with the participating center and treatment year. Covariates were included irrespective of their univariable statistical significance. Stabilized weights were calculated and truncated at the 1st and 99th percentiles to limit the influence of extreme weights. Covariate balance was evaluated using standardized mean differences (SMDs), with an absolute SMD < 0.10 indicating acceptable balance. A doubly robust estimation strategy was subsequently implemented by fitting weighted multivariable outcome models. To reduce model overfitting, candidate covariates for the outcome model were screened using LASSO regression. Variables retaining non-zero coefficients and prespecified clinical prognostic factors were considered for inclusion in the final doubly robust model. To account for immortal time bias and treatment modifications, the antithrombotic regimen was modeled as a time-varying covariate, ensuring that a patient’s contributed person-years dynamically shifted between exposure pools precisely at documented transition milestones. For the evaluation of efficacy endpoints, including the patency triad and recurrence, a Fine-Gray subdistribution hazard model was utilized to robustly account for non-thrombosis-related mortality as a competing risk. Safety outcomes were analyzed as time to the first major bleeding event and time to the first CRNMB event using time-varying Cox proportional hazards models adjusted for age, sex, BMI, and prior bleeding history. Within the DAT cohort, exploratory comparisons of primary, primary-assisted, and secondary patency between DOAC-based and VKA-based regimens were performed using multivariable Fine-Gray models, with non-thrombosis-related mortality treated as a competing event. The models were adjusted for age, thrombophilia, distal stent extension, adjunctive CDT/PMT, total stent length, and duration of DAT exposure. Bleeding outcomes in this subgroup were analyzed using multivariable time-varying Cox models. Sensitivity analyses included 1:1 propensity score matching, overlap weighting, and restriction to the region of common support by excluding patients with propensity scores < 0.05 or > 0.95. The primary Fine–Gray model for loss of primary patency was repeated under each specification. An E-value analysis was additionally performed to estimate the minimum strength of association that an unmeasured confounder would need to have with both treatment allocation and primary patency loss to fully explain the observed association. Because standardized Villalta scores were unavailable, a sensitivity analysis was performed using a stricter definition of chronic post-thrombotic symptoms. This definition required symptoms to be documented at least 6 months after the index DVT and either to be recorded at two or more clinical assessments separated by at least 3 months or to include documented venous skin changes or venous ulceration. The association between time-varying antithrombotic exposure and this stricter outcome was estimated using the IPTW-adjusted Fine–Gray model. All statistical analyses were executed using R software, version 4.5.0, and a two-sided p-value of less than 0.05 was considered statistically significant.

3. Results

3.1. Baseline Characteristics, Dynamic Therapy, and IPTW Results

Between January 2018 and December 2022, a systematic query of the multicenter EMR database identified an initial pool of potentially eligible patients, which, after the rigorous application of prespecified inclusion and exclusion criteria, resulted in a final analytical cohort of 450 patients (Figure 1). Among them, 186 (41.3%) patients were prescribed DAT and 264 (58.7%) were prescribed SAC at hospital discharge. In the original unweighted cohort, several procedural and anatomical factors were significantly imbalanced between the two treatment arms; notably, patients in the DAT group exhibited a higher proportion of stents extending distally across the inguinal ligament into the CFV (40.9% vs. 24.6%, p < 0.001) and had significantly longer total stent lengths (108.5 ± 22.4 mm vs. 92.3 ± 18.6 mm, p < 0.001) compared to the SAC group, reflecting the clinical reality that dual therapy is frequently reserved for more complex anatomical presentations (Table 1). Following the implementation of IPTW and supplemental 1:1 PSM, the baseline demographic characteristics, clinical histories, and operative parameters achieved excellent equilibrium between the cohorts, with all SMDs for the 16 predefined covariates successfully compressed below the rigorous 0.1 threshold (Table 1). After stabilized-weight truncation, the IPTW weights ranged from 0.41 to 3.26, and the weighted effective sample size was 401.7. All measured covariates, including treatment center and treatment year, achieved an absolute SMD below 0.10 after weighting. The longitudinal trajectory of antithrombotic regimens throughout the first year post-discharge was visualized utilizing a Sankey diagram (Figure 2), which delineated a pronounced step-down pattern within the DAT cohort. Among the 186 patients initially discharged on DAT, 153 (82.3%), 86 (46.2%), and 22 (11.8%) remained on DAT at 3, 6, and 12 months, respectively. At 12 months, 101 patients (54.3%) had transitioned to SAC, 42 (22.6%) were receiving antiplatelet therapy alone after discontinuation of anticoagulation, and 21 (11.3%) were receiving no antithrombotic treatment. Among the 164 patients no longer receiving DAT, 148 (90.2%) underwent planned de-escalation according to the intended antithrombotic schedule, including transition to SAC or antiplatelet monotherapy, whereas treatment modification was attributed to CRNMB in 7 patients and major bleeding in 2 patients.

Figure 1.

Study flow diagram outlining the patient selection process

Table 1.

Comparison of Baseline Characteristics and Procedural Parameters before and after IPTW Weighting and PSM Matching

Variables	DAT (n = 186)	SAC (n = 264)	Statistic (t/χ2)	P-value	Raw SMD	IPTW SMD	PSM SMD
Demographics
Age, years	54.2 ± 12.1	55.8 ± 11.8	-1.417	0.157	0.134	0.032	0.021
Male, n (%)	91 (48.9)	135 (51.1)	0.218	0.641	0.044	0.015	0.012
BMI, kg/m2	25.4 ± 3.8	24.9 ± 4.1	1.332	0.184	0.126	0.028	0.019
Smoking history, n (%)	39 (21.0)	51 (19.3)	0.228	0.633	0.045	0.018	0.014
Clinical History
Thrombophilia, n (%)	29 (15.6)	37 (14.0)	0.203	0.652	0.042	0.011	0.009
Pre-existing CVI, n (%)	52 (28.0)	65 (24.6)	0.584	0.445	0.072	0.024	0.018
Baseline D-dimer, mg/L	4.8 ± 2.2	4.6 ± 2.5	0.893	0.372	0.085	0.031	0.022
Prior bleeding history, n (%)	14 (7.5)	24 (9.1)	0.351	0.554	0.056	0.018	0.012
Anatomy & Stenting
Distal extension (CFV), n (%)	76 (40.9)	65 (24.6)	13.298	< 0.001	0.347	0.054	0.042
Proximal extension (IVC), n (%)	17 (9.1)	21 (8.0)	0.134	0.714	0.036	0.012	0.008
Dedicated venous stent, n (%)	135 (72.6)	180 (68.2)	1.054	0.305	0.098	0.027	0.015
Stent diameter, mm	12.8 ± 1.4	12.6 ± 1.2	1.624	0.105	0.153	0.048	0.033
Total stent length, mm	108.5 ± 22.4	92.3 ± 18.6	8.358	< 0.001	0.785	0.061	0.052
Number of stents, n	1.4 ± 0.6	1.2 ± 0.4	4.218	< 0.001	0.392	0.045	0.038
Intraoperative Details
Adjunctive CDT/PMT, n (%)	165 (88.7)	227 (86.0)	0.957	0.328	0.091	0.022	0.016
Balloon angioplasty, n (%)	178 (95.7)	251 (95.1)	0.165	0.685	0.039	0.014	0.010

Notes. Values are expressed as mean ± standard deviation for continuous variables and as number (percentage) for categorical variables. P-values refer to the comparisons of the original (unweighted) cohort. Abbreviations: BMI, body mass index; CFV, common femoral vein; IVC, inferior vena cava; CDT, catheter-directed thrombolysis; PMT, pharmacomechanical thrombectomy; DAT, dual antithrombotic therapy; SAC, single anticoagulation; SMD, standardized mean difference; IPTW, inverse probability of treatment weighting; PSM, propensity score matching.

Figure 2.

Sankey diagram illustrating the dynamic evolution of antithrombotic therapy over the 12-month follow-up period. The diagram depicts treatment transitions from hospital discharge to the 3-, 6-, and 12-month follow-up milestones. Nodes represent dual antithrombotic therapy (DAT), single anticoagulation (SAC), antiplatelet therapy alone, or no antithrombotic treatment, and the width of each connecting flow is proportional to the number of patients transitioning between treatment states. The diagram demonstrates the predominant step-down from DAT to SAC over time, as well as less frequent transitions to antiplatelet therapy alone or no antithrombotic treatment. Treatment status was updated according to the documented date of regimen modification

Among the 264 patients initially discharged on SAC, 39 underwent target-lesion reintervention, including 24 interventions for severe restenosis or impending occlusion and 15 recanalization procedures for complete stent occlusion. Eighteen patients subsequently transitioned to DAT, including 11 after a primary-assisted patency intervention and 7 after a secondary patency intervention.

3.2. Long-Term SVS Patency Triad and Efficacy

Over a median follow-up period of 46 months (interquartile range, 38–55 months), the associations between antithrombotic exposure and clinical outcomes were evaluated using an IPTW-adjusted Fine-Gray subdistribution hazard model, with the medication regimen modeled as a time-varying covariate and non-thrombosis-related mortality treated as a competing event. DAT exposure was associated with a lower risk of patency loss across all three SVS patency outcomes compared with SAC exposure (Table 2). The estimated 1-, 3-, and 5-year primary patency rates were 93.5%, 86.2%, and 81.8%, respectively, during DAT exposure and 88.4%, 75.1%, and 67.5%, respectively, during SAC exposure, corresponding to an sHR of 0.58 for primary patency loss (95% CI, 0.41–0.82; p = 0.002). Similar associations were observed for primary-assisted patency loss (sHR, 0.52; 95% CI, 0.35–0.78; p = 0.001) and secondary patency loss (sHR, 0.45; 95% CI, 0.28–0.71; p < 0.001) (Figure 3). DAT exposure was also associated with lower cumulative incidences of clinically treated chronic post-thrombotic symptoms (13.4% vs. 23.1%; sHR, 0.61; 95% CI, 0.43–0.86; p = 0.005) and ipsilateral DVT recurrence (4.8% vs. 10.6%; sHR, 0.48; 95% CI, 0.26–0.89; p = 0.019). Under the stricter sensitivity definition requiring later or repeated symptom documentation, the corresponding cumulative incidences were 10.8% and 18.7%, respectively, and the association remained directionally consistent (sHR, 0.64; 95% CI, 0.43–0.95; p = 0.027) (Supplemental Table S1).

Table 2.

Estimated 1-, 3-, and 5-Year SVS Patency Rates and Clinical Efficacy Outcomes Based on the IPTW-Adjusted Time-Varying Fine-Gray Model

Efficacy endpoints	DAT exposure	SAC exposure	Statistic (sHR)	95% CI	P-value
Primary Patency, %
1-Year	93.5	88.4	—	—	—
3-Year	86.2	75.1	—	—	—
5-Year	81.8	67.5	—	—	—
Overall loss of primary patency	—	—	0.58	0.41–0.82	0.002
Primary-Assisted Patency, %
1-Year	96.2	91.1	—	—	—
3-Year	91.5	80.6	—	—	—
5-Year	88.0	74.2	—	—	—
Overall loss of primary-assisted patency	—	—	0.52	0.35–0.78	0.001
Secondary Patency, %
1-Year	98.0	94.2	—	—	—
3-Year	95.5	87.8	—	—	—
5-Year	92.4	82.5	—	—	—
Overall loss of secondary patency	—	—	0.45	0.28–0.71	< 0.001
Secondary Clinical Outcomes
Clinically treated chronic post-thrombotic symptoms, %	13.4	23.1	0.61	0.43–0.86	0.005
Ipsilateral DVT recurrence, %	4.8	10.6	0.48	0.26–0.89	0.019

Notes. Patency rates and cumulative incidences are derived from the inverse probability of treatment weighting (IPTW) pseudo-cohort. The subdistribution hazard ratios (sHR) and corresponding P-values were calculated using the Fine-Gray competing risk regression model, treating antithrombotic regimens as time-varying covariates and non-thrombosis-related mortality as a competing risk event. An sHR < 1.0 indicates a protective effect (lower risk of patency loss, clinically treated chronic post-thrombotic symptoms, or recurrence) associated with DAT exposure compared to SAC. Abbreviations: DAT, dual antithrombotic therapy; SAC, single anticoagulation; sHR, subdistribution hazard ratio; CI, confidence interval; DVT, deep vein thrombosis; SVS, Society for Vascular Surgery.

Figure 3.

Cumulative incidence plots of the SVS patency triad. The step-function curves illustrate the IPTW-adjusted cumulative incidence of loss of primary patency (A), loss of primary-assisted patency (B), and loss of secondary patency (C) over the 5-year follow-up period. The analyses rigorously account for dynamic therapeutic crossovers via time-varying covariates and incorporate non-thrombosis-related mortality as a competing risk, showing lower adjusted cumulative incidences of patency loss during DAT exposure. The numbers at risk shown below each panel are unweighted observed patient counts classified according to the antithrombotic regimen being received at each time point. Because treatment exposure was time-varying, patients could contribute to different exposure rows after a documented regimen change. Patients receiving antiplatelet therapy alone or no antithrombotic treatment were not included in the DAT or SAC risk counts

3.3. Real-World Safety Outcomes

The incidence of hemorrhagic events during the follow-up period was evaluated utilizing a time-varying Cox proportional hazards model to accurately attribute bleeding episodes to the concurrent antithrombotic exposure status at the time of the event. All major bleeding and CRNMB events included in the analysis had sufficient clinical, laboratory, transfusion, or treatment documentation for classification according to the ISTH criteria. Overall, the occurrence of major bleeding was infrequent and demonstrated no statistically significant difference between the DAT and SAC exposure periods. Specifically, major bleeding occurred at an incidence rate of 1.24 per 100 person-years during DAT exposure compared to 0.98 per 100 person-years during SAC exposure, yielding a non-significant HR of 1.28 (95% CI, 0.48–3.42; p = 0.624). None of the major bleeding events in either exposure pool were fatal; the majority were gastrointestinal hemorrhages that resolved following temporary discontinuation of antithrombotic agents and, in some cases, blood transfusion. Conversely, the dynamic exposure to DAT was associated with a substantially elevated risk of CRNMB. The incidence rate of CRNMB was 5.57 per 100 person-years under DAT, which was significantly higher than the 2.45 per 100 person-years observed under SAC (HR, 2.35; 95% CI, 1.48–3.74; p < 0.001). These CRNMB events were predominantly mucosal in origin, including epistaxis, gingival bleeding, and menorrhagia, which necessitated medical consultation but rarely led to permanent cessation of anticoagulation (Figure 4). Furthermore, an exploratory subgroup analysis conducted exclusively within the DAT exposure pool revealed distinct safety profiles based on the type of oral anticoagulant utilized. Patients receiving a direct oral anticoagulant in combination with a single antiplatelet agent (DOAC-based DAT) exhibited a significantly lower incidence rate of CRNMB than those receiving VKA-based DAT (4.15 vs. 8.78 per 100 person-years; HR, 0.44; 95% CI, 0.22–0.89; p = 0.022), alongside a numerically lower, although statistically non-significant, rate of major bleeding (Table 3). In the exploratory patency analysis, DOAC-based DAT was associated with a lower risk of primary patency loss than VKA-based DAT (sHR, 0.58; 95% CI, 0.35–0.96; p = 0.034). Associations for primary-assisted patency loss (sHR, 0.62; 95% CI, 0.34–1.13; p = 0.119) and secondary patency loss (sHR, 0.60; 95% CI, 0.29–1.24; p = 0.168) were directionally similar but not statistically significant (Supplemental Table S2).

Figure 4.

Simon-Makuch plots (time-varying adjusted survival curves) estimating freedom from major bleeding (A) and clinically relevant non-major bleeding (CRNMB) (B). The survival curves depict the cumulative freedom from hemorrhagic events dynamically stratified by the time-updated antithrombotic exposure status. While freedom from major bleeding remained statistically comparable between the DAT and SAC regimens, the DAT cohort exhibited a significantly steeper decline in freedom from CRNMB, highlighting the mucosal bleeding trade-off associated with the addition of antiplatelet agents

Table 3.

Overall Bleeding Events and Exploratory Subgroup Safety Analysis Based on Time-Varying Exposure

Bleeding endpoints	DAT exposure (PY = 485)	SAC exposure (PY = 1224)	HR (95% CI)	P-value
Primary Safety Endpoints
Major Bleeding, n (IR)	6 (1.24)	12 (0.98)	1.28 (0.48–3.42)	0.624
- Gastrointestinal	4	8	—	—
- Intracranial	0	1	—	—
- Other requiring transfusion	2	3	—	—
CRNMB, n (IR)	27 (5.57)	30 (2.45)	2.35 (1.48–3.74)	< 0.001
- Epistaxis	8	9	—	—
- Gingival/mucosal	7	8	—	—
- Genitourinary/menorrhagia	8	7	—	—
- Cutaneous/subcutaneous	4	6	—	—
Subgroup Analysis within DAT	DOAC-based DAT (PY = 337)	VKA-based DAT (PY = 148)	Subgroup HR (95% CI)	P-value
Major Bleeding, n (IR)	3 (0.89)	3 (2.03)	0.43 (0.09–2.14)	0.306
CRNMB, n (IR)	14 (4.15)	13 (8.78)	0.44 (0.22–0.89)	0.022

Notes. Event frequencies (n) indicate the number of patients who experienced at least one event of the specified category while receiving the corresponding antithrombotic regimen. The Cox models evaluated time to the first event of each category. Incidence rates (IR) are expressed per 100 person-years (PY) of exposure. Hazard ratios (HR) and corresponding P-values were derived from the multivariable time-varying Cox proportional hazards model adjusting for baseline age, sex, BMI, and prior bleeding history. An HR > 1.0 indicates an increased risk of bleeding associated with DAT exposure relative to SAC, or VKA-based DAT relative to DOAC-based DAT. Abbreviations: DAT, dual antithrombotic therapy; SAC, single anticoagulation; PY, person-years; IR, incidence rate per 100 person-years; HR, hazard ratio; CI, confidence interval; CRNMB, clinically relevant non-major bleeding; DOAC, direct oral anticoagulant; VKA, vitamin K antagonist.

3.4. Dimensionality Reduction of Predictors, Competing Risk Regression, and E-Value Analysis

To optimize model parsimony and statistical stability, the 16 candidate baseline and procedural covariates were subjected to LASSO regression with 10-fold cross-validation. At the penalty parameter (λ) that minimized the cross-validation error, three covariates retained non-zero coefficients: distal stent extension across the inguinal ligament, adjunctive CDT/PMT, and total stent length (Figure 5). Within the doubly robust estimation framework, the final multivariable Fine-Gray competing-risk model included these selected covariates together with the primary exposure of DAT versus SAC. DAT exposure remained associated with a lower risk of primary patency loss (sHR, 0.58; 95% CI, 0.41–0.82; p = 0.002). Distal stent extension into the CFV (sHR, 1.84; 95% CI, 1.25–2.71; p = 0.002), absence of adjunctive CDT/PMT (sHR, 2.12; 95% CI, 1.34–3.35; p = 0.001), and greater total stent length (per 10-mm increase: sHR, 1.12; 95% CI, 1.04–1.21; p = 0.005) were associated with a higher risk of primary patency loss (Table 4 and Figure 6).

Figure 5.

Covariate selection using the least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation. (A) The cross-validation error curve plotted against the penalization parameter (Log-λ). The vertical dashed line indicates the optimal Lambda value that minimizes the mean squared error (λ.min). (B) The coefficient path plot demonstrates the shrinkage of regression coefficients for the 16 candidate baseline covariates toward zero as the penalty increases. At the optimal penalty threshold, 13 non-contributory variables were penalized to zero (grey dashed lines), resulting in the absolute retention of three core prognostic predictors (colored solid lines)

Table 4.

Independent Predictors of Primary Patency Loss and E-Value Analysis Based on the LASSO-Selected and DRE-Weighted Fine-Gray Model

Predictor	sHR	95% CI	P-value
Primary Exposure
DAT exposure (Ref: SAC)	0.58	0.41–0.82	0.002
Selected Covariates
Distal stent extension (Ref: No extension)	1.84	1.25–2.71	0.002
Absence of adjunctive CDT/PMT (Ref: Performed)	2.12	1.34–3.35	0.001
Total stent length (per 10-mm increase)	1.12	1.04–1.21	0.005
Robustness Assessment	Value	Lower 95% CI	​
E-value for DAT protective effect	2.84	1.74	—

Notes. The multivariable regression was performed on the inverse probability of treatment weighting (IPTW) pseudo-population using a Fine-Gray subdistribution hazard model. An sHR < 1.0 indicates a protective factor, whereas an sHR > 1.0 indicates a risk factor for primary patency loss. The E-value quantifies the minimum strength of association that an unmeasured confounder would require with both the exposure and the outcome to completely nullify the observed treatment effect. Abbreviations: DAT, dual antithrombotic therapy; SAC, single anticoagulation; sHR, subdistribution hazard ratio; CI, confidence interval; CDT, catheter-directed thrombolysis; PMT, pharmacomechanical thrombectomy.

Figure 6.

Forest plot of independent predictors for the loss of primary patency. The plot visualizes the subdistribution hazard ratios (sHR) and 95% confidence intervals derived from the doubly robust estimation (DRE) Fine-Gray model. DAT exposure was associated with a lower risk of patency loss, while anatomical complexity (distal extension across the inguinal ligament) and technical limitations (absence of adjunctive CDT/PMT) significantly increase the risk

The association between DAT exposure and lower primary patency loss remained consistent across alternative propensity score specifications. In the 1:1 propensity score-matched cohort, the sHR was 0.61 (95% CI, 0.42–0.89; p = 0.010). Similar estimates were obtained using overlap weighting (sHR, 0.60; 95% CI, 0.43–0.84; p = 0.003) and after restriction to patients within the common-support region (sHR, 0.63; 95% CI, 0.45–0.88; p = 0.007) (Supplemental Table S3). The E-value for the primary estimate was 2.84, with a lower confidence-limit E-value of 1.74.

4. Discussion

The principal finding of this multicenter retrospective cohort study was that exposure to DAT, consisting of an oral anticoagulant combined with a single antiplatelet agent, was associated with higher long-term stent patency following endovascular intervention for acute iliofemoral DVT in the setting of May-Thurner syndrome. Using doubly robust estimation with time-varying exposure modeling, dynamic exposure to DAT was associated with a lower risk of primary patency loss than SAC. This association remained consistent across propensity score matching, overlap weighting, and restriction to the common-support population. This robust benefit was consistently mirrored across all tiers of the SVS patency triad, ¹⁵ including primary-assisted and secondary patency, and was accompanied by lower cumulative incidences of clinically treated chronic post-thrombotic symptoms and ipsilateral DVT recurrence. These findings underscore that the addition of an antiplatelet agent to standard anticoagulation may address a critical pathophysiological gap in the prevention of in-stent thrombosis and restenosis within the venous system.

The chronic symptom outcome in this study should be distinguished from formally adjudicated PTS. Standardized Villalta scores were not routinely documented across the participating centers; therefore, the analysis captured persistent post-thrombotic symptoms that resulted in continued clinical treatment. The association remained consistent when a stricter definition requiring later or repeated symptom documentation was applied. Nevertheless, treatment prescription may be influenced by physician practice, center-specific follow-up protocols, and patient adherence, and the observed outcome should not be directly compared with Villalta-defined PTS rates reported in prospective studies.

The observation of a substantial patency benefit with DAT aligns with and substantiates a growing body of literature challenging the traditional paradigm of anticoagulation monotherapy after venous stenting. Our findings are consistent with the multicenter study by Qiu et al, which found that anticoagulation plus antiplatelet therapy was associated with better iliac vein stent patency than anticoagulation alone after stenting for acute or subacute iliofemoral DVT. ²¹ Earlier observational evidence also suggested that combined anticoagulant and antiplatelet therapy may be associated with favorable venous stent outcomes. In a small series of patients undergoing stenting for iliofemoral post-thrombotic venous obstruction, Langwieser et al reported maintained primary patency during follow-up with rivaroxaban plus clopidogrel, although the absence of a comparator group and the limited sample size precluded conclusions regarding comparative efficacy. ²² Similarly, Huang et al reported that anticoagulant combined with antiplatelet therapy was associated with improved iliac vein stent patency following thrombolysis and thrombectomy for subacute iliofemoral DVT in a retrospective propensity score-matched cohort, further supporting the potential benefit of adjunctive antiplatelet therapy in thrombotic iliac venous lesions. ²³ Endo et al subsequently reported an association between concomitant antiplatelet and anticoagulant therapy and iliocaval stent patency, although their cohort was also limited by its retrospective design and modest sample size. ²⁴ Furthermore, a systematic review by Notten et al underscored that the preponderance of historic post-interventional protocols relied heavily on vitamin K antagonists, with antiplatelet agents being only rarely incorporated. This highlights a critical lack of comparative evidence that our multicenter data actively addresses. ²⁵ Real-world evidence from population-based analyses, such as those by Lokken et al, further emphasizes the marked heterogeneity in post-stenting antithrombotic management. The fact that combination therapy in their analysis was not associated with increased bleeding compared to anticoagulation alone provides important external validation for our safety findings, suggesting that the clinical shift toward DAT may be achieved without compromising patient safety. ²⁶

The mechanistic rationale for the superiority of DAT over SAC in the venous stent environment is multifaceted and extends beyond the simple prevention of recurrent venous thromboembolism. While the central role of anticoagulation in inhibiting the coagulation cascade and fibrin formation is well-established, the contribution of platelets to venous stent failure is increasingly recognized. Stent implantation induces endothelial denudation and exposes the subendothelial matrix, triggering platelet adhesion, activation, and aggregation, a process that is not fully mitigated by anticoagulants alone. ²⁷ This early platelet-rich thrombus formation can serve as a nidus for subsequent fibrin deposition and neointimal hyperplasia, ultimately leading to in-stent restenosis or occlusion. Interestingly, a study by Langwieser et al suggested that venous stenting itself does not appear to chronically activate platelets, but rather that patients who develop in-stent stenosis exhibit greater levels of baseline platelet activation prior to stent placement. ²⁷ This implies that the addition of an antiplatelet agent may be particularly beneficial in a subset of patients with a pre-existing pro-thrombotic phenotype, effectively blunting the initial platelet response to the implanted foreign body.

The safety profile observed in our cohort is both reassuring and nuanced. The incidence of major bleeding was infrequent and statistically comparable between the DAT and SAC exposure periods, a finding that aligns with previous multicenter observations and underscores the relative safety of adding a single antiplatelet agent to a therapeutic anticoagulation backbone.^18,21 Conversely, a notable increase in CRNMB was observed with DAT, reflecting the expected pharmacodynamic synergy between anticoagulants and antiplatelet agents. ¹⁹ This elevated risk of predominantly mucosal bleeding episodes represents a predictable clinical trade-off for the superior patency afforded by dual therapy. Importantly, the exploratory analysis within the DAT cohort suggested that DOAC-based DAT was associated with both a lower incidence of CRNMB and a lower risk of primary patency loss than VKA-based DAT. The associations for primary-assisted and secondary patency were directionally consistent but did not reach statistical significance, likely reflecting the smaller number of advanced patency-loss events. These findings provide a preliminary benefit–risk signal favoring DOAC-based DAT but should be interpreted cautiously. Anticoagulant selection was not randomized and may have been influenced by renal function, comorbidities, anticipated adherence, physician preference, and the need for laboratory monitoring. The subgroup results therefore do not establish the superiority of DOAC-based DAT and require confirmation in adequately powered prospective comparisons. While some population-based evidence has suggested no overall increase in bleeding with combination therapy, ²⁶ our granular modeling demonstrates that the hemorrhagic risk is real but predominantly concentrated in less severe categories. Furthermore, this risk can be effectively mitigated through the preferential use of DOACs, a principle that mirrors evidence from high-bleeding-risk scenarios in cardiovascular medicine, such as patients with atrial fibrillation undergoing percutaneous coronary intervention. ²⁸ Beyond the primary exposure, our analysis identified two critical anatomical and procedural factors that independently predict the loss of primary patency: distal stent extension across the inguinal ligament and the absence of adjunctive CDT/PMT. The association between CFV extension and an increased risk of patency loss is a consistent theme in the literature. ²⁹ Stenting across the inguinal ligament places the device in a zone of dynamic flexion and external compression, which may promote neointimal hyperplasia and stent deformation. ²⁹ This finding serves as a critical reminder that while extending a stent distally may be necessary to ensure adequate inflow, it comes at a cost. Similarly, the protective effect of adjunctive CDT/PMT underscores the fundamental importance of achieving a pristine venous lumen prior to stent deployment, reinforcing the “open vein hypothesis” wherein aggressive thrombus removal is an essential prerequisite for durable outcomes.

Post-stenting antithrombotic therapy is frequently delivered as a step-down strategy, although the direction and timing of de-escalation vary across centers. Some postoperative protocols discontinue anticoagulation after approximately 3 months while continuing antiplatelet monotherapy, whereas others discontinue the antiplatelet component and continue anticoagulation. Both pathways were captured in our time-varying exposure framework. Once anticoagulation was discontinued, subsequent person-time was assigned to the antiplatelet-therapy-alone state rather than to DAT or SAC. In our cohort, the predominant de-escalation pathway among patients discharged on DAT was transition to SAC; however, 42 patients (22.6%) were receiving antiplatelet therapy alone at 12 months. Only 9 treatment modifications were directly attributed to bleeding, indicating that most transitions reflected planned treatment changes rather than adverse-event-driven withdrawal. Therefore, the reported DAT-versus-SAC estimates apply specifically to periods during which patients remained anticoagulated and do not directly evaluate whether anticoagulation can be safely discontinued at 3 months or whether antiplatelet monotherapy is sufficient thereafter. This distinction is clinically important because contemporary recommendations remain heterogeneous. A recent multidisciplinary Delphi study reached consensus that antiplatelet therapy should be continued if anticoagulation is discontinued, but did not reach consensus on whether anticoagulation beyond 3 months is necessary or whether aspirin should routinely be added to anticoagulation after venous stenting. ³⁰ The POATIVES and international Delphi statements similarly emphasize uncertainty regarding treatment duration and the role of adjunctive antiplatelet therapy.^10,11 Accordingly, our findings support an association between DAT and improved patency during anticoagulant-treated periods, but they do not establish the optimal timing of anticoagulant discontinuation or the effectiveness of subsequent antiplatelet monotherapy.

The limitations of this study warrant careful consideration. First, the retrospective observational design remains susceptible to confounding by indication. Patients receiving DAT had longer stented segments and more frequent distal extension into the common femoral vein before weighting, indicating that physicians preferentially selected dual therapy for anatomically complex cases. Treatment allocation may also have been influenced by unrecorded factors, including residual thrombus burden, inflow quality, subtle lesion morphology, perceived adherence, and individual physician or center preference. Although IPTW, doubly robust estimation, propensity score matching, overlap weighting, and common-support restriction produced consistent estimates and balanced the measured covariates, none of these approaches can eliminate bias from unmeasured clinical judgment factors. The E-value quantifies the strength of residual confounding required to explain the observed association but does not demonstrate the absence of such confounding. Accordingly, the results should be interpreted as associations rather than causal treatment effects. Second, standardized Villalta scores were not routinely available across the participating centers. We therefore used a treatment-based definition of chronic post-thrombotic symptoms, which may have been influenced by physician prescribing practices, center-specific follow-up protocols, and patient adherence. Although the association remained consistent under a stricter definition requiring later or repeated symptom documentation, outcome misclassification remains possible, and these estimates should not be interpreted as effects on formally adjudicated PTS. Third, bleeding ascertainment relied primarily on records available within the participating centers and on sufficiently documented outside-center events. Minor events not requiring medical contact and events treated elsewhere without supporting documentation may have been missed, potentially leading to underestimation of less severe bleeding. Fourth, comparisons of bleeding and patency outcomes between DOAC- and VKA-based DAT were exploratory, with limited numbers of patency-loss events and no randomized treatment allocation; consequently, these subgroup estimates remain susceptible to treatment-selection bias and should be considered hypothesis-generating. Fifth, although antiplatelet monotherapy after anticoagulation discontinuation was captured as a separate time-varying exposure state, the study was neither prespecified nor powered to compare this step-down strategy with continued SAC or DAT. Therefore, the optimal duration of anticoagulation and the comparative effectiveness and safety of subsequent antiplatelet monotherapy remain unresolved. Sixth, the cohort was derived from tertiary referral centers, which may limit generalizability. Finally, medication adherence was assessed from prescription and follow-up records and may not fully reflect actual treatment use.

5. Conclusion

In this multicenter real-world study, during periods of oral anticoagulation, time-varying DAT exposure was associated with higher long-term stent patency than SAC exposure following iliofemoral stenting for May-Thurner syndrome. DAT was also associated with lower rates of clinically treated chronic post-thrombotic symptoms and recurrent DVT, but with a higher risk of clinically relevant non-major bleeding. Antiplatelet monotherapy after anticoagulation discontinuation was not a prespecified primary comparator; therefore, the study does not determine whether anticoagulation can be safely discontinued at 3 months or whether antiplatelet monotherapy is sufficient thereafter. Given the potential for residual treatment-selection bias and outcome misclassification, prospective studies comparing prespecified step-down antithrombotic schedules are required.

ORCID iDs

Pengyu Ji https://orcid.org/0009-0000-3418-6278

Yong Wang https://orcid.org/0009-0000-0210-3984

Ethical Considerations

This multicenter, retrospective cohort study was approved by the Institutional Review Board (IRB) of Ordos Central Hospital (Approval No: OCH-IRB-2023-042) and all corresponding participating centers. The study protocol adhered to the ethical principles outlined in the Declaration of Helsinki.

Consent to Participate

The requirement for written informed consent to participate was waived by the Institutional Review Board (IRB) of Ordos Central Hospital due to the purely retrospective, non-interventional nature of the study. All patient data were strictly de-identified prior to analysis.

Data Availability Statement

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.*