Risk Prediction Model for Lower Limb DVT in Critically Ill Orthopaedic Patients: A Focus on Spinal Surgery and Postoperative Immobility
نویسندگان
1 Department of Intensive Care Medicine, Shanghai Fourth People's Hospital, Shanghai, 200434, China
2 Department of Intensive Care Medicine, Shanghai Fourth People's Hospital, Shanghai, 200434, China
3 Department of Intensive Care Medicine, Shanghai Fourth People's Hospital, Shanghai, 200434, China
4 Department of Hematology, Huaian First People's Hospital, Huai'an, Jiangsu, 223300, China
5 Department of Intensive Care Medicine, Shanghai Fourth People's Hospital, Shanghai, 200434, China
6 Department of Intensive Care Medicine, Shanghai Fourth People's Hospital, Shanghai, 200434, China
7 Department of Intensive Care Medicine, Shanghai Fourth People's Hospital, Shanghai, 200434, China
8 Department of Intensive Care Medicine, Shanghai Fourth People's Hospital, Shanghai, 200434, China
doi
10.22034/ircmj.2025.481321.1516چکیده
Background and Objectives: Lower limb deep vein thrombosis (DVT) poses a significant risk to patients undergoing orthopedic surgeries. The aim of the present study was to develop a customized intervention plan and construct a risk prediction model for lower limb DVT in these patients. Methods: We used a convenience sampling method due to its practicality and efficiency in clinical settings to enrol orthopaedic inpatients. The samples were divided into two groups: a training set (n=183) collected between January 2020-December 2021 for model development and a validation set (n=83) collected from February-July 2022 for validation of risk prediction model. Model performance was assessed using accuracy, sensitivity, specificity, and the area under the ROC curve (AUC). Results: The model correctly identified 61 out of 83 patients, achieving 73.49% accuracy in identifying both DVT-positive and DVT-negative cases, with a sensitivity of 81.82% (9/11) for detecting DVT-positive patients and a specificity of 72.22% (52/72) for identifying non-DVT patients. The AUC (0.807, 95% CI: 0.671–0.894) reflects that the model had strong discriminatory ability, though some uncertainty remains, as reflected in the confidence interval with a Jorden index of 0.491 which balances sensitivity and specificity to optimize the model’s predictive power. The Sensitivity (83.4%) and Specificity (75.2%) of the model shows that it can effectively detect DVT cases while minimizing false positives. Multivariate regression analysis revealed that independent risk factors (occurrence rate<5%) included age (>60 years), post-surgery bedridden status (>5days), and spinal surgeries. Conclusion: This study developed a risk prediction model for lower limb DVT in critically ill orthopaedic patients. Key risk factors, including advanced age, prolonged post-surgery bedridden status, and spinal surgeries. Larger multi-center studies are needed for validation.
