Developing and validating risk predicting models to assess venous thromboembolism risk after radical cystectomy.

Abstract

Background: Radical cystectomy (RC) patients are at significant risk for venous thromboembolism (VTE). Current predictive models, such as the Caprini risk assessment (CRA) model, have limitations. This research aimed to create a novel predictive model for forecasting the risk of VTE after RC.

Methods: This single-center study involved RC patients treated between January 1, 2010 and December 31, 2019. The individuals were divided into training and testing groups in a random manner. Multivariate and stepwise logistic regression were utilized to create two novel models. The models' performance was compared to the commonly used CRA model, employing metrics including net reclassification improvement (NRI), integrated discrimination improvement (IDI), and receiver operating characteristic (ROC) curve analyses.

Results: A total of 272 patients were enrolled, among whom 36 were diagnosed with VTE after RC. Model A and Model B were then conducted. The area under ROC of Model A and Model B is 0.806 [95% confidence interval (CI): 0.748-0.856] and 0.833 (95% CI: 0.777-0.880), respectively, which were also determined in the testing cohorts. The two new Models were superior both in classification ability and prediction ability (NRI >0, IDI >0, P<0.01). Model A and Model B had a concordance index (C-index) of 0.806 and 0.833, respectively. In decision curve analysis (DCA), the two new models provided a net benefit between 0.02 and 0.84, suggesting promising clinical utility.

Conclusions: Regarding predictive accuracy, both models surpass the existing CRA model, with Model A being advantageous due to its fewer variables. This easy-to-use model enables swift risk assessment and timely intervention for high-risk groups, yielding favorable patient outcomes.