Analysis of Risk Factors and Establishment of a Predictive Model for Endplate Collapse After Posterior Short-Segment Fixation Combined with Screw Placement at the Injured Vertebra in Thoracolumbar Fractures

Abstract Purpose To identify the risk factors for endplate collapse in patients with thoracolumbar fractures undergoing posterior short-segment fixation combined with intermediate screw placement, and to develop and validate a clinical prediction model to aid in early risk stratification and prognosis optimization. Materials and methods Clinical data from 135 patients with thoracolumbar fractures treated with posterior short-segment fixation and screw placement at the injured vertebra were retrospectively analyzed. Based on postoperative imaging findings, patients were stratified into an endplate collapse group (n = 57) and a non-collapse control group (n = 78). Variables collected included demographic and fracture characteristics (age, gender, BMI, fracture location, fracture pattern, mechanism of injury, osteoporotic status), radiographic parameters (screw length at the injured vertebra, severity of endplate and intervertebral disc injury, extent of anterior and posterior complex injury, postoperative mean vertebral body height [MVBH] and Cobb angle [CA]), and postoperative outcomes (fixation duration, postoperative Pfirrmann grade). Univariate analysis identified variables with significant between-group differences. Independent risk factors for postoperative endplate collapse were determined by multivariable logistic regression analysis. A predictive model was constructed incorporating these factors, and its performance was evaluated using receiver operating characteristic (ROC) curve analysis with calculation of area under the curve (AUC), sensitivity, and specificity. Results Univariate analysis revealed significant differences in BMI, osteoporotic status, screw depth, endplate injury grade, intervertebral disc injury severity (Sander classification), load-sharing classification (LSC) score, postoperative Pfirrmann grade, and postoperative Cobb angle between groups (all P < 0.05). Multivariable logistic regression identified osteoporosis, screw depth, and endplate injury grade as independent risk factors for postoperative endplate collapse (all P < 0.05), with screw depth demonstrating the largest odds ratio and thus the strongest association. A predictive model incorporating these three factors was constructed. Receiver operating characteristic curve analysis demonstrated an area under the curve (AUC) of 0.888 (95% CI: 0.826–0.943), with sensitivity of 78.9% and specificity of 89.7%, indicating favorable predictive performance. Conclusion Postoperative endplate collapse following posterior short-segment fixation with injured vertebra screw placement for thoracolumbar fractures is multifactorial, with osteoporosis, screw depth, and endplate injury grade serving as major independent risk factors. The predictive model developed in this study effectively stratifies the risk of postoperative endplate collapse and may provide a valuable tool for preoperative risk assessment, intraoperative technical optimization, and postoperative individualized rehabilitation planning, thereby potentially reducing the incidence of endplate collapse and improving functional outcomes.