A Nomogram Model to Predict Non-Retrieval of Short-term Retrievable Inferior Vena Cava Filters Short Title: Risk Factors for Non-retrieval of Retrievable Filters

Abstract

Abstract Objective To develop and validate a nomogram for predicting non-retrieval of the short-term retrievable inferior vena cava (IVC) filters. Methods In this study, univariate and multivariate logistic regression analyses were performed to identify predictive factors of short-term retrievable filter non-retrieval, and a nomogram was then established based on these factors. The nomogram was created based on data from a training cohort and validated based on data from a validation cohort. The predictive value of the nomogram was estimated using area under the curve (AUC) and calibration curve analysis (Hosmer-Lemeshow test). Results A total of 1321 patients who had undergone placement of short-term retrievable filters (Aegisy or OptEase) were included in the analysis. The overall retrieval rate was. Age, mixed type deep vein thrombosis (DVT) vs peripheral type DVT, active cancer, history of long-term immobilization, VTE was detected in the intensive care unit, active/recurrent bleeding, IVC thrombosis, and history of venous thromboembolism were independent predictive risk factors for non-retrieval of filters. Interventional therapy for DVT, fresh fracture, and interval of ≥ 14 days between filter placement and patient discharge were independent protective factors for non-retrieval of filters. The nomogram based on these factors demonstrated good ability to predict the non-retrieval of filters (training cohort AUC = 0.870; validation cohort AUC = 0.813. Conclusion This nomogram demonstrated strong predictive accuracy and discrimination capability. This model may help clinicians identify patients who are not candidates for short-term retrievable filter placement and help clinicians make timely, individualized decisions in filter choice strategies.