TITLE:
Development and Validation of a Clinical Prediction Model for Postoperative Pulmonary Complications in Patients Undergoing Spinal Tumor Surgery
AUTHORS:
Wenxuan Jiang, Yong’e Gu
KEYWORDS:
Spinal Tumors, Pulmonary Complications, Risk Factors, Logistic Regression, Random Forest, Predictive Model
JOURNAL NAME:
Open Journal of Nursing,
Vol.16 No.9,
September
18,
2026
ABSTRACT: Objective: To analyze the independent factors influencing postoperative pulmonary complications in patients with spinal tumors, develop a risk prediction model based on these factors, and evaluate its predictive performance, thereby providing a reference for the perioperative risk stratification of high-risk patients. Methods: A total of 245 patients with spinal tumors who underwent surgical treatment at our hospital between January 2024 and March 2026 were retrospectively included. Patients were divided into a complication group (n = 34) and a non-complication group (n = 211) according to the occurrence of pulmonary complications within 30 days after surgery. Nineteen clinical indicators covering the preoperative and intraoperative stages were collected as candidate predictors. Univariate analysis was first performed to screen variables that differed between groups, followed by multivariate binary logistic regression to identify independent risk factors. A random forest prediction model was then constructed using the selected variables, and its discriminative ability was assessed by the receiver operating characteristic curve and the area under the curve (AUC). Results: The overall incidence of postoperative pulmonary complications was 13.88% (34/245). Univariate analysis showed significant between-group differences in BMI (P = 0.045), operative duration (P = 0.0162), surgical approach (P Conclusion: Surgical approach is an independent high-risk factor for postoperative pulmonary complications after spinal tumor surgery, while BMI and massive transfusion show protective and risk-related trends, respectively. The constructed random forest model has limited discriminative ability and can be used as a perioperative tool for risk factor identification and auxiliary risk stratification. To achieve accurate individualized prediction, larger sample sizes and the incorporation of specialty-specific indicators such as pulmonary function are needed.