A Novel Model for Predicting Post-Craniotomy Meningitis Using Early Postoperative Risk Stratification: A Multi-Center Retrospective Study.

Purpose: Early prediction of post-craniotomy meningitis (PCM) is critical for optimizing prognosis. Given the limitations of existing prediction tools, this study sought to develop and externally validate a novel model for early postoperative risk stratification and prediction of PCM. Patients and Methods: Data were retrospectively collected from three Chinese hospitals. Least absolute shrinkage and selection operator (LASSO) and backward stepwise logistic regression were utilized to select independent predictors for nomogram construction. Nomogram performance was further evaluated by areas under the receiver operating characteristic curves (AUC) (discrimination), calibration plots (calibration) and decision curve analysis (clinical applicability). Meanwhile, pairwise comparisons of AUCs were performed to compare predictive performance between the model and the individual independent predictors. Additionally, a practical probability threshold of 30% was proposed based on the decision curve analysis to evaluate the model's performance in terms of sensitivity and specificity. Results: According to their admitted hospitals, 694 patients were allocated to the training cohort, and 309 comprised the external validation cohort. Within the training cohort, postoperative CSF leak, intraoperative/postoperative ventricular drain placement, and trans-sinusal surgery were identified as independent predictors (all Conclusion: Our novel model exhibited favorable performance and clinical applicability, and was found to outperform the individual predictors. Accordingly, this newly built model seems a practical warning tool, though better-designed studies are needed for its further refinement.