# Risk factors and predictive model for early surgical site infection following single-level PLIF in diabetic patients

## TL;DR

This study identifies post-surgery blood markers that predict infection risk in diabetic patients after spinal surgery, creating a model to help doctors prevent infections.

## Contribution

A novel prediction model using postoperative inflammatory markers to identify early surgical site infection risk in diabetic PLIF patients.

## Key findings

- CRP, WBC, and other postoperative inflammatory markers are strong predictors of early SSI in diabetic PLIF patients.
- The prediction model achieved high accuracy with AUCs of 0.987 (training) and 0.990 (validation).
- The model shows excellent clinical utility and stability for early risk identification.

## Abstract

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.

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).

Univariate analysis revealed that patients in the infection group had significantly higher postoperative day 3 fasting blood glucose (FPG pod3: 17.18 vs. 14.13 mmol/L), C-reactive protein (CRP pod3: 281.70 vs. 111.17 mg/L), white blood cell count (WBC pod3: 41.28 vs. 16.90 × 109/L), and 4 other postoperative inflammatory markers compared to the non-infection group (all P < 0.001). Multivariate logistic regression further identified CRP pod3 (OR = 1.025, 95% CI: 1.01–1.04, P < 0.001), WBC pod3 (OR = 1.27, 95% CI: 1.17–1.43, P < 0.001), Erythrocyte Sedimentation Rate (ESR) pod3 (mm/h) (OR = 1.021, 95% CI: 1.01–1.04, P = 0.007), Procalcitonin (PCT) pod3 (ng/mL) (OR = 1.503, 95% CI: 1.24–11.95, P < 0.001), Neutrophil-to-Lymphocyte Ratio (NLR) pod3 (OR = .131, 95% CI: 1.07–1.23, P < 0.001), and Platelet-to-Lymphocyte Ratio (PLR) pod3 (OR = 1.012, 95% CI: 1.01–1.02, P < 0.001) 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 (95% CI: 0.972–1.000) for the training set and 0.990 (95% CI: 0.971–1.000) 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.

Postoperative day 3 serum inflammatory markers (e.g., CRP, WBC) 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.