Construction of a risk prediction model for postoperative deep vein thrombosis in lung cancer patients

Abstract

Objective To investigate the independent risk factors for postoperative deep vein thrombosis in lung cancer patients and to construct a risk prediction model. Methods Clinical data of 354 inpatients who underwent thoracoscopic surgery for lung cancer in Department of Thoracic Surgery of First Affiliated Hospital of Army Medical University between May 2019 and May 2023 were retrospectively collected and analyzed. LASSO regression was used to screen potential factors, followed by multivariate logistic regression to identify risk factors, and then a nomogram prediction model was constructed. Calibration curves, receiver operating characteristic (ROC) curves, and decision curves were drawn to evaluate the model's calibration, discrimination, sensitivity, specificity, and clinical utility. The net reclassification improvement (NRI) and integrated discrimination improvement (IDI) indices were employed to compare the predictive performance of the constructed model with the Caprini score for outcome events. Results LASSO regression identified 17 potential influencing factors. Multivariate regression analysis showed that D-dimer, central venous catheter (CVC) placement, and lower extremity varicose veins were independent risk factors for postoperative DVT in lung cancer patients (P < 0.05). Calibration curve analysis showed the model had good agreement between the predicted and observed values. ROC curve analysis indicated that the sensitivity and specificity of the model was 0.812 and 0.963, respectively, with an area under the curve (AUC) value of 0.912 (95%CI: 0.840~0.983). In comparison, the Caprini model had a sensitivity and specificity of 0.625 and 0.860, respectively, with an AUC value of 0.752 (95%CI: 0.657~0.846). The NRI and IDI for the model group compared to the Caprini model were 0.709 and 0.431, respectively. Decision curve analysis showed that the net benefit of applying the model from this study was higher than that of the Caprini model. Conclusion D-dimer, CVC, and varicose veins of lower extremities are independent risk factors for DVT after thoracoscopic surgery in patients with lung cancer. Our constructed nomogram model can effectively predict the risk of DVT after thoracoscopic surgery in patients with lung cancer.

Keywords

• lung cancer
• deep venous thrombosis
• prediction model
• nomogram