Development and validation of a predictive model for preoperative deep vein thrombosis following traumatic thoracolumbar fractures

Abstract

Abstract Although a sequential work-up for deep vein thrombosis has reached agreement worldwide, the mysterious nature of DVT following fractures brings challenges to early diagnosis and intervention. The objective of the present study was to develop and validate a nomogram for predicting preoperative DVT risk in patients with thoracolumbar fractures using readily available clinical data. Of the 1350 patients, 930 were randomly assigned to the training cohort. A prediction model was established and visualized as a nomogram based on eight predictors related to preoperative DVT. The performance of the model was tested by the receiver operating characteristic curve, Hosmer–Lemeshow test, calibration curve, and decision curve analysis. We further verified the model in the validation cohort. The AUCs of the prediction model were 0.876 and 0.853 in training and validation cohorts, respectively. The Hosmer–Lemeshow test demonstrated good fitness in the training set (X2 = 5.913, P = 0.749) and the validation set (X2 = 9.460, P = 0.396). Calibration and decision curve analyses performed well in training and validation sets. In short, we developed a prediction model for preoperative DVT risk in patients with thoracolumbar fractures and verified its accuracy and clinical utility.