Why Temporary Filters Are Not Removed: Clinical Predictors in 1,000 Consecutive Cases.

BACKGROUND: Compared to permanent inferior vena cava (IVC) filters, higher complication rates occur with long-term use of temporary IVC filters. We aimed to identify patient clinical factors at the time of placement that could predict failure to remove a temporary IVC filter.

METHODS: A retrospective review was performed of both vascular surgery and interventional radiology prospective databases between December 2008 and December 2013. We analyzed a total number of 1,024 consecutive, temporary IVC filters stratified by whether retrieval was attempted or made permanent. Univariate, multivariate, and prediction modeling analyses with internal validation were performed on abstracted data, which included risk factors, treatment modalities, and indications for IVC filter placement.

RESULTS: Of 1,024 temporary IVC filters, removal was attempted in 60% and no attempt at removal (kept permanent) in 40%. Of the 619 with attempted removal, the overall successful retrieval rate was 95%. The majority of filters were not attempted to be removed because of persistent filter indications (360 cases). Risk factors associated with IVC filter permanence included male sex, older age, history, or indication of venous thromboembolism (VTE) with inability to anticoagulate, malignancy, and neurologic condition. Risk factors most predictive of permanence in the multivariate model were malignancy (odds ratio [OR]: 3.0, P < 0.001) or neurologic disorder (OR: 2.69, P = 0.0005). Validation revealed our model had a sensitivity of 60.4% and specificity of 69.9%.

CONCLUSIONS: Our study shows that patients who are more likely to have a temporary IVC filter kept permanent are more likely to be older males with a history of malignancy, neurologic condition, or VTE. These factors are also predictive of permanence and can be used in our predictive model to provide insight into the significant preoperative risk factors that should play into the decision-making process.