Risk prediction models for venous thromboembolism in stroke: a systematic review and meta-analysis

Objective: This study aimed to systematically evaluate risk prediction models for venous thromboembolism (VTE) in stroke patients and to provide a reference for future model development and clinical research. Methods: A systematic search was conducted across multiple databases, including PubMed, Embase, Web of Science, the Cochrane Library, China National Knowledge Infrastructure (CNKI), Wanfang Database, VIP, and SinoMed, to identify studies on VTE risk prediction models in stroke patients. Databases were searched from inception to September 1, 2025. Risk of bias and applicability of the prediction models were assessed using the PROBAST checklist. Meta-analyses were conducted to estimate pooled VTE incidence and the area under the curve (AUC) for model performance using Stata 17.0. Results: A total of 2,726 records were retrieved, and seven prediction models were included. Reported VTE incidence ranged from 9.8% to 38.9%, with AUC values between 0.781 and 0.978, indicating moderate to high apparent discriminative performance, however, this should be interpreted with caution in light of the high risk of bias and lack of external validation. None of the seven included models underwent independent external validation, and according to PROBAST, all included models were judged to be at high overall risk of bias. The pooled VTE incidence was 20.8% (95% CI: 14.7%-27.0%), and the pooled AUC across six models was 0.87 (95% CI: 0.81-0.93). Conclusion: Among the VTE risk prediction models for stroke patients included in this study, although some models demonstrated favorable predictive performance, all models were judged to be at high risk of bias according to PROBAST. Future research should prioritize the validation and refinement of existing models or the development of new models with more rigorous methodological design, in order to better support clinical decision-making for patients with stroke. Systematic Review Registration: https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42024603132, PROSPERO CRD42024603132.