A Model for Intracranial Bacterial Infection After Neurosurgery Integrating Bacterial Culture and Genetic Testing

Abstract

Background:

Early diagnosis of intracranial bacterial infections following neurosurgical procedures remains challenging because of delayed or inconclusive pathogen identification. This study aimed to develop and validate a predictive model for early detection of post-neurosurgical procedure bacterial infections by integrating clinical features and cerebrospinal fluid (CSF) biomarkers.

Methods:

We conducted a retrospective study including 429 patients who underwent neurosurgery at Tianjin Huanhu Hospital between November 2018 and December 2024. Patients were randomly divided into training (n = 300) and validation (n = 129) cohorts. Predictive parameters were selected via uni-variable analysis and LASSO regression, followed by multi-variable logistic regression to construct a nomogram. Model performance was evaluated using receiver operating characteristic curves, calibration plots, and decision curve analysis (DCA).

Results:

Among 429 patients, 147 (34.3%) had confirmed CSF bacterial infections. Seven independent predictors were identified: CSF time to operation, CSF color, CSF white blood cell count ≥ 2000 × 10⁶/L, CSF glucose < 1.89 mmol/L, CSF lactate concentration, hyperpyrexia, and slower reaction time. The model achieved an area under the ROC curve of 0.822 (95% confidence interval [CI]: 0.771–0.873) in the training cohort and 0.715 (95% CI: 0.604–0.827) in the validation cohort. DCA demonstrated substantial clinical net benefit across a threshold probability range of 0.1–0.25.

Conclusion:

This nomogram-based model provides a practical and reliable tool for early risk stratification of post-neurosurgical procedure intracranial bacterial infections, supporting timely diagnostic and therapeutic decision-making in clinical practice.