A clinical risk prediction model for perioperative lower extremity DVT in patients undergoing spinal fracture surgery

To develop a perioperative lower-extremity deep vein thrombosis (DVT) risk prediction model for spinal fracture surgery patients using logistic regression, supporting clinical prevention strategies.

Clinical data from 249 patients undergoing spinal fracture surgery (July 2019-October 2024) were retrospectively analyzed. Participants were divided into a model group (n = 166) and a validation group (n = 83) in a 2:1 ratio. Univariate and multivariate logistic regression identified independent risk factors for perioperative DVT, and a predictive model was established. Model fit was evaluated using the Hosmer-Lemeshow test, and predictive performance was assessed via receiver operating characteristic (ROC) curve analysis.

Independent risk factors included perioperative blood transfusion, elevated C-reactive protein, D-dimer >500 μg/L, hypertension, age ≥60 years, and prolonged bed rest. The model [P = 1/(1 + e^-Z)] demonstrated a good fit (Hosmer-Lemeshow χ ² = 12.139, P = 0.807). ROC analysis showed AUC values of 0.75 (95% CI: 0.80-0.92) for the model group and 0.81 (95% CI: 0.64-0.98) for the validation group, indicating robust predictive performance.

The identified risk factors are critical predictors of perioperative DVT in spinal fracture patients. The proposed model exhibits strong clinical utility for early risk stratification and intervention guidance.