Abstract
BACKGROUND AND PURPOSE: Deep vein thrombosis (DVT) is a major complication after lower extremity traumatic fractures. Current nursing assessment is largely postoperative-centered and may overlook risk evolution during the preoperative waiting period. METHODS: In this prospective observational cohort study, 948 patients with lower extremity traumatic fractures were enrolled between January 2019 and January 2026. Serial color Doppler ultrasound examinations were performed at five timepoints: admission, preoperative, postoperative day 1, postoperative day 3, and discharge. Univariate and multivariate logistic regression analyses were used to identify predictors of preoperative DVT. A dynamic nomogram was developed and evaluated using receiver operating characteristic analysis, calibration, decision curve analysis, and SHAP-based interpretation. RESULTS: The overall incidence of perioperative DVT was 32.9% (312/948). DVT showed a distinct "twin-peak" temporal pattern, with the highest incidence during the preoperative waiting period and a second peak on postoperative day 3. Thrombi also shifted from predominantly distal veins at admission to greater proximal involvement after surgery. Independent predictors of preoperative DVT were age >65 years, preoperative waiting time >3 days, D-dimer >1.85 mg/L, albumin <35 g/L, and time from injury to admission >12 h. The nomogram showed good discrimination (AUC 0.905, 95% CI 0.882-0.928), outperforming individual predictors such as D-dimer (AUC 0.796). Decision curve analysis demonstrated clinical net benefit, and SHAP analysis identified D-dimer and albumin as the most influential variables. CONCLUSION: Perioperative DVT after lower extremity fractures follows a dynamic pattern with a critical preoperative risk window. The proposed nomogram may support stage-specific nursing assessment and targeted interventions, particularly by reducing preoperative delay and optimizing nutritional status.