Risk Factors and Predictive Model for Early Surgical Site Infection Following Single-Level PLIF in Diabetic Patients

Xusheng Li^1,2Shuid Ahmad Nazrun¹Mohd Miswan Mohd Fairudz¹Cao Donghui²Xiao Zhang²Yanrong Tian²Haifeng Yuan^2*

• ¹Universiti Teknologi MARA Kampus Sungai Buloh, Sungai Buloh, Malaysia
• ²General Hospital of Ningxia Medical University, Yinchuan, China

Objective: This study aims to investigate the predictive value of postoperative serum biomarkers for early surgical site infection (SSI) following single-level posterior lumbar interbody fusion (PLIF) in diabetic patients, and to construct an infection risk prediction model based on key indicators. The goal is to provide a theoretical basis and tool support for precise clinical prevention and control. Methods: A retrospective analysis was conducted on 1,680 diabetic patients who underwent single-level PLIF in our Hospital, from January 2011 to December 2024. Among these, 165 patients developed early SSI. Univariate analysis was performed using–Whitney U test and the chi-square test. Subsequently, LASSO regression was employed for variable selection and dimensionality reduction, and independent risk factors were determined using multivariate logistic regression. Data were divided into training and validation sets in a 7:3 ratio, and a prediction model was constructed using 10-fold cross-validation. The model's predictive performance and clinical utility were comprehensively evaluated with calibration curves, receiver operating characteristic (ROC) curves, and decision curve analysis (DCA). Results: Univariate analysis revealed that patients in the infection group had significantly higher postoperative day 3 fasting blood glucose,C-reactive protein ,white blood cell count, and 4 other postoperative inflammatory markers compared to the non-infection group (all P<0.001). Multivariate logistic regression further identified CRP pod3 , WBC pod3 , ESR pod3 , PCT pod3 , NLR pod3 , and PLR pod3 as independent risk factors. The decision tree prediction model, constructed based on these variables, showed excellent discrimination ability with areas under the ROC curve (AUC) of 0.987 for the training set and 0.990 for the validation set. The calibration curve closely followed the ideal reference line, indicating good model fit. DCA demonstrated that the model had high clinical net benefit across all risk thresholds. Conclusion: Postoperative day 3 serum inflammatory markers have high predictive value in identifying early SSI in diabetic patients undergoing single-level PLIF. The prediction model constructed based on these markers performs excellently in terms of accuracy, stability, and clinical utility, making it an effective tool for early identification of high-risk infection patients and providing scientific evidence for individualized postoperative management strategies and interventions.