Development and validation of a deep vein thrombosis risk assessment tool for surgical patients aged 75 years and older

BACKGROUND: Deep vein thrombosis (DVT) is a common and severe medical condition characterized by the formation of thrombi in deep veins, primarily affecting older surgical patients. The present study aimed to identify risk factors for DVT in surgical patients aged 75 years and older and subsequently develop and validate a risk assessment tool for this patient population. METHODS: A retrospective study was conducted on surgical patients (n = 686) aged 75 years and older at a tertiary general hospital in Hefei, China, from January to December 2024. Predictors for the model were selected using Least Absolute Shrinkage and Selection Operator (LASSO) regression, followed by multivariable logistic regression. Area under the curve, calibration curve, and decision curve analysis (DCA) were used to examine the discriminative power, calibration, and clinical efficacy of the predictive models. Internal validation was performed using both bootstrap resampling and 10-fold cross-validation. RESULTS: The incidence of DVT among surgical patients aged 75 years and older was 14.7% (n = 101/686). Six predictors were identified and used to establish a nomogram: malignancy (OR: 7.590, 95% CI: 2.670–21.500), sex (OR: 0.387, 95% CI: 0.195–0.724), anesthesia duration (OR: 1.010, 95% CI: 1.006–1.014), D-dimer (OR: 1.210, 95% CI: 1.130–1.310), platelet count (OR: 1.010, 95% CI: 1.005–1.015), and pneumatic tourniquet application (OR: 2.700, 95% CI: 1.470–5.170). The nomogram demonstrated excellent discrimination (AUC = 0.786, 0.786 (95% CI, 0.738–0.834) and good calibration (Hosmer-Lemeshow test, P = 0.588). Upon interval validation, the model achieved a concordance index (C-index) of 0.791 (95% CI, 0.780–0.800). Finally, DCA demonstrated the net clinical benefit of this nomogram. CONCLUSIONS: This study constructed a practical model to predict DVT in surgical patients aged 75 years and older. This model incorporates demographic characteristics and clinical risk factors, enabling individualized prediction.