Objective: This study aimed to develop and validate a risk nomogram for preoperative proximal and distal deep vein thrombosis (DVT) in geriatric patients with hip fractures.
Patients and methods: The 970 collected geriatric hip fracture patients were randomly divided into a training set (70%, n=682) and a validation set (30%, n=288). Multivariate logistic regression analyses were used to optimize the predictive risk variables for proximal and distal preoperative lower extremity DVT in the training set, respectively, and the selected variables were finally incorporated to establish preoperative DVT nomogram prediction models. Receiver operating characteristic curves (ROC), calibration plots, and decision curve analysis (DCA) were performed to validate the nomograms in the training and validation sets, respectively.
Results: Among the 970 patients, 125 (12.88%) were diagnosed with preoperative DVT. The area under the curve (AUC) for predicting preoperative proximal DVT was 0.888 in the training and 0.792 in the validation sets. The AUC for predicting preoperative distal DVT was 0.907 in the training and 0.790 in the validation sets. The calibration plots and decision curve analysis for preoperative proximal DVT performed well in the training set and slightly worse in the validation set. The calibration plots and decision curve analysis for preoperative distal DVT performed well in both the training and validation sets.
Conclusions: To construct nomograms for predicting the risk of proximal and distal preoperative lower extremity DVT in geriatric hip fracture patients. For patients at high risk, as assessed by this model, clinicians should intervene and treat them promptly before surgery.