Objective The aim of this study was to construct a novel nomogram prediction model and to investigate the value of the new predictors in predicting perioperative deep vein thrombosis formation in lower limb traumatic closed fracture (LTCF). Methods In this study, we reviewed data from 1,608 patients with LTCF, developed a new predictive model through analysis, and used Lasso regression to screen for the final reliable variables. Modeling was followed by using the ROC curve analysis, calculation of area under the ROC curve and Validation function. Calibration curves, decision curve analysis and clinical impact curves were used to comprehensively assess the performance of the model, and used to construct new predictors, inflammatory immune factors and traditional scores by using ROC analysis model, which was compared with the new predictive model. Results The novel nomogram model was constructed with an AUC = 0.918, 95% CI: 0.894-0.942, C-index = 0.913. The new predictors (PHR and PDR), inflammatory-immune domain factors (NLR, PLR, and SII), and traditional scoring systems (Wells and Caprini) had AUC values of 0.689, 0.838, 0.552, 0.557, 0.542, 0.897, and 0.872. Conclusion The novel nomogram model constructed in this study is feasible for predicting the occurrence of perioperative DVT in LTCF. PHR and PDR demonstrated superior predictive performance for perioperative DVT in LTCF patients, integrating platelet, metabolic, and fibrinolytic pathways. Associated inflammatory indices (NLR, PLR, SII) offered adjunctive risk assessment value. Threshold analysis indicated optimal performance at a 0.35 cutoff (F1 = 0.828), with perfect precision achievable at ≥0.65, highlighting the model's clinical adaptability through adjustable thresholds.
Zeng et al. (Thu,) studied this question.