Evaluation of a clinical decision support tool to predict permanence of retrievable inferior vena cava filters

Objective: To evaluate the usefulness of a published clinical decision support tool to predict the likelihood of a retrievable inferior vena cava (IVC) filter being maintained as a permanent device. Methods: This multicenter retrospective cohort study included 1498 consecutive patients (852 men and 646 women; median age, 60 years; range, 18-98 years) who underwent retrievable IVC filter insertion between January 2012 and December 2019. The indications for IVC filtration, baseline neurologic disease, history of venous thromboembolism (VTE), and underlying malignancy were recorded. Accuracy, sensitivity, and specificity of a published clinical support tool were calculated to determine the usefulness of the tool. Results: The majority of filters (1271/1498 [85%]) were placed for VTE with a contraindication to anticoagulation. A history of VTE was present in 811 of 1498 patients (54%) patients; underlying malignancy in 531 of 1498 patients (35%), and neurological disease in 258 of 1498 patients (17%). Of the 1498 filters, 456 (30%) were retrieved, 276 (18%) were maintained as permanent devices on follow-up, and 766 (51%) filters were not retrieved. The accuracy of the clinical prediction model was 61%, sensitivity was 60%, and specificity was 62%. Conclusions: A previously published clinical decision support tool to predict permanence of IVC filters had modest usefulness in the examined population; this factor should be taken into account when using this clinical decision support tool outside of the original study population. Future studies are required to refine the predictive capability of IVC filter decision support tools for broader use across different patient populations.