Comparison of Different Prediction Models for Predicting Deep Venous Thrombosis in Patients Undergoing Chemotherapy for Colorectal Cancer

Objective To compare the predictive value of four models—Khorana, Vienna CATS, PROTECHT, and ONKOTEV—for deep venous thrombosis (DVT) formation in patients undergoing chemotherapy for colorectal cancer. Methods Data were collected from patients undergoing chemotherapy for colorectal cancer for the first time in the Department of Colorectal Surgery at a tertiary hospital in Gansu Province from January 2024 to August 2024. On the second day of admission, the risk of DVT was assessed using the four models, and lower limb vascular color Doppler ultrasound was performed to confirm the occurrence of DVT. The area under the receiver operating characteristic curve (AUROC) was used to evaluate the predictive value of each model. Results 316 colorectal cancer patients were included in this study, and 85 patients developed DVT, with an incidence rate of 26.90%. The AUROC values for the Khorana, Vienna CATS, PROTECHT, and ONKOTEV models were 0.732, 0.781, 0.772, and 0.877, respectively, with corresponding sensitivities of 0.625, 0.829, 0.625, and 1.000, respectively. Pairwise comparisons of the four models showed that ONKOTEV significantly differed from the other three models (P < 0.05), whereas no significant differences were observed among the other three models. Conclusion Four models all demonstrated good predictive value for DVT in colorectal cancer patients undergoing chemotherapy. And ONKOTEV exhibited the best predictive value, which was recommended to be preferentially used to realize the early and timely intervention, reduce the incidence of DVT, alleviate the healthcare burden, and improve the life quality of patients.