Abstract
BACKGROUND: Deep vein thrombosis (DVT) is a frequent complication following endovascular thrombectomy (EVT) in patients with acute ischaemic stroke (AIS), potentially leading to fatal pulmonary embolism (PE). Identifying patients early at high risk for DVT is clinically important. This study developed and validated a nomogram combining laboratory findings and clinical characteristics to predict the risk of lower-extremity DVT after EVT in patients with AIS. METHODS: This retrospective multicentre observational study was conducted in two tertiary hospitals in China, enrolling 640 patients who underwent ultrasonography for DVT diagnosis within 10 days following EVT. Data on medical history, examination and laboratory results were collected for logistic regression analyses to develop a DVT risk nomogram. RESULTS: Logistic regression analyses identified critical predictors of DVT: lower limb National Institutes of Health Stroke Scale (NIHSS) score ≥ 2, elevated D-dimer levels (≥ 1.62 mg/L) and prolonged puncture-to-recanalization time (PRT ≥ 66 min). The nomogram demonstrated good discriminative ability (AUC 0.741-0.822) and clinical utility across internal and external validation cohorts. Additionally, the presence of DVT was significantly associated with reduced functional independence at 90 days post-EVT, highlighting the negative impact of DVT on patient recovery (OR = 3.85; 95% CI: 2.18-6.78; p < 0.001). CONCLUSION: The study provides a practical clinical tool for early detection and intervention in patients with AIS at high risk for DVT following EVT. Early identification and intervention may help improve outcomes in patients with AIS undergoing EVT. RELEVANCE TO CLINICAL PRACTICE: This nomogram helps in the early detection and proactive management of DVT in AIS patients, which can reduce severe complications and improve patient recovery outcomes. PATIENT OR PUBLIC CONTRIBUTION: No patient or public contributions were involved in this study due to its retrospective design, where data were utilised from existing medical records without direct patient interaction.