Risk Factors and Nursing Strategies for Postoperative Pain Management in Patients with Lumbar Spinal Stenosis Undergoing Transforaminal Lumbar Interbody Fusion: A Retrospective Study

Jingran GuoXiaoying WangLijuan WangYu WangJie LiYi Bu^*

• The Second Hospital of Tangshan, Tangshan, China

Objective: This study attempts to identify risk factors associated with postoperative pain in patients with lumbar spinal stenosis undergoing transforaminal lumbar interbody fusion (TLIF) and to propose targeted nursing strategies. Methods: We retrospectively analyzed 502 patients who underwent TLIF. Patients were grouped into mild, moderate, and severe pain groups based on postoperative pain severity. Baseline characteristics, comorbidities, sex, age, body mass index (BMI), and history of lumbar surgery were compared across groups. Preoperative serological markers such as glycated hemoglobin (HbA1c), albumin, globulin, red blood cell count (RBC), white blood cell count (WBC), platelet count (PLT), neutrophil-to-lymphocyte ratio (NLR), and C-reactive protein (CRP) were analyzed. Surgical parameters, including operative time, intraoperative blood loss, surgical segment, bone graft material, anesthesia method, drainage duration, and postoperative complications, were also assessed. Ordinal logistic regression and Extreme gradient boosting (XGBoost) models were applied to analyze risk factors influencing postoperative pain severity, with model performance assessed by Receiver Operating Characteristic (ROC) curves and calibration plots. Results: Significant differences among pain groups were observed in age, BMI, HbA1c, albumin, globulin, RBC, WBC, PLT, NLR, CRP, operative time, intraoperative blood loss, drainage duration, surgical segment, and complication rates (all P<0.05). Ordinal logistic regression identified these factors as significant predictors of severe pain, with intraoperative blood loss showing the highest odds ratio (OR=1.037, P<0.001). XGBoost analysis ranked intraoperative blood loss as the top contributor. In the test set, XGBoost achieved an AUC of 0.968 compared with 0.974 for the ordinal logistic model; however, the logistic model demonstrated superior variance explanation (R²=0.728 vs. 0.710) and prediction accuracy (RMSE=0.262 vs. 0.268; MAE=0.116 vs. 0.146). Conclusion: Intraoperative blood loss emerged as a critical factor affecting pain severity. Both ordinal logistic regression and XGBoost models provide strong predictive performance and can effectively guide individualized nursing strategies, potentially improving postoperative recovery for TLIF patients.