Risk factors and a predictive model for nonfilter-associated inferior vena cava thrombosis in patients with lower extremity deep vein thrombosis

Abstract

Objective: Nonfilter-associated inferior vena cava thrombosis (IVCT) is an under-recognized but severe state of venous thromboembolism. The aims of this study were to investigate risk factors and develop a prediction model based on clinical data and imaging findings to evaluate the probability of IVCT in patients with lower extremity deep vein thrombosis (LEDVT). Methods: A single-center retrospective cohort study was conducted. We analyzed the clinical data and multimodal imaging findings of consecutive patients with confirmed LEDVT between February 2016 and January 2022. The demographics, presentation of LEDVT, laboratory examination, thrombus characteristics, comorbidities and risk factors for LEDVT, and imaging findings were analyzed using an independent t-test, Chi-square test, Fisher's exact test, and regression analysis to determine the univariable and multivariable associations and to establish a predictive model to assess the probability of IVCT. Results: A total of 267 eligible patients were included, of whom 40 were in the IVCT group and 227 were in the non-IVCT group. The incidence of nonfilter-associated IVCT was 15.0% (40/267). Age < 63.5 years [odds ratio (OR) 2.54; 95% confidence interval (CI), 1.10–5.85, p = 0.029], male sex (OR 2.82; 95% CI, 1.19–6.72, p = 0.019), proximal DVT (OR 8.21; 95% CI, 1.01–66.76, p = 0.049), bilateral DVT (OR 7.30; 95% CI, 3.28–16.21, p < 0.001), and D-dimer >4.72 μg/ml (OR 4.64; 95% CI, 1.80–11.72, p = 0.001) were risk factors for IVCT's occurrence. Then, we established a prediction model based on these risk factors. The diagnostic efficiency [area under the curve (AUC) of receiver operating characteristic (ROC) curve was 0.858] for predicting IVCT was superior to that of isolated risk factors, including age < 63.5 years (AUC of ROC curve was 0.624) or D-dimer >4.72 μg/ml (AUC of ROC curve was 0.656). Conclusion: Age < 63.5 years, male sex, proximal LEDVT, bilateral LEDVT and D-dimer >4.72 μg/ml were risk factors. The diagnostic efficiency of the predictive model for predicting IVCT was superior to that of a single risk factor alone. It may be used for predicting the probability of nonfilter-associated IVCT in patients with LEDVT.