Central Nervous System Infection After Neuroendoscopic and Microscopic Combined Hematoma Removal: Risk Factors and Construction of a Nomogram Prediction Model

Objective  To analyze the risk factors associated with central nervous system (CNS) infection in patients after neuroendoscopic hematoma removal combined with and microscopic hematoma removal, and to construct and validate a nomogram prediction model.

Methods  A total of 460 patients who underwent neuroendoscopic hematoma removal combined with microscopic hematoma removal at our hospital between January 2021 and December 2024 were retrospectively enrolled. The patients were assigned to a modeling cohort (n = 322) and a validation cohort (n = 138) in a 7∶3 ratio. Furthermore, the modeling cohort was divided into an infection group (n = 68) and a non-infected group (n = 254) according to whether CNS infection occurred. The independent predictors of central nervous system infection were identified by logistic regression analysis, and a nomogram prediction model was constructed accordingly.

Results  The overall incidence of CNS infection in the 460 patients was 20.65% (95/460). According to the logistic regression analysis, the independent risk factors for CNS infection in patients after neuroendoscopic and microscopic combined hematoma removal included a history of diabetes mellitus (odds ratio OR = 3.431, 95% CI: 1.300-9.057), the Glasgow Coma Scale (GCS) score (OR = 0.574, 95% CI: 0.462-0.711), cerebrospinal fluid leakage (OR = 4.492, 95% CI: 1.430-14.116), operation duration (OR = 1.011, 95% CI: 1.004-1.019), duration of drainage tube placement (OR = 5.452, 95% CI: 2.423-12.268) and albumin (ALB) level (OR = 0.778, 95% CI: 0.720-0.840) (P < 0.05). Based on these risk factors, a nomogram prediction model was constructed, and the area under the receiver operating characteristic curve (AUC) of the predicted events in the modeling cohort and the validation cohort was 0.928 (95% CI: 0.895-0.960) and 0.918 (95% CI: 0.885-0.951), respectively. The calibration curve fitted well with the ideal curve (Hosmer-Lemeshow test, P > 0.05), and the decision curve analysis demonstrated significant net benefit.

Conclusion  The nomogram model based on history of diabetes mellitus, GCS score, cerebrospinal fluid leakage, operation duration, duration of drainage tube placement, and ALB level demonstrates high predictive performance for CNS infection after neuroendoscopy-assisted microscopic hematoma removal.