To establish a nomogram model for risk prediction of postoperative central nervous system (CNS) infection in patients with severe traumatic brain injury (sTBI), and to provide reference for the prevention, early diagnosis and treatment of postoperative CNS infection in sTBI patients.

The clinical data of 370 patients with sTBI who underwent neurosurgery at Department of Neurosurgery, Northern Jiangsu People′s Hospital of Jiangsu Province, from January 2017 to September 2020 were retrospectively analyzed. A total of 370 patients were randomly divided into training set (n=259) and validation set (n=111) according to the ratio of 7 ∶3. Patients in the training set were divided into infection group (n=45) and non-infection group (n=214) according to whether CNS infection occurred. Univariate and multivariate logistic regression analyses (forward method) were used to screen the risk factors of postoperative CNS infection after sTBI and establish nomogram prediction models. The receiver operating characteristic (ROC) curve was drawn with the area under the curve (AUC) calculated to evaluate the efficacy of the nomogram model in prediction of CNS infection. The AUC >0.80 indicated that the efficacy was high. Internal validation of the model was carried out, and decision curve analysis (DCA) and clinical impact curve (CIC) were used to evaluate the clinical utility and net benefit of the prediction model.

Based on the comparison between infection group and non-infection group, there were significant differences in admission GCS, combined cerebrospinal fluid otorrhea, cerebrospinal fluid rhinorrhea leakage, skull base fracture, intraventricular hemorrhage, emergency operation, operation time, placement of brain-ventricle drainage, placement of intracranial pressure probe, secondary operation, intracranial rebleeding, cerebrospinal fluid sampling, tracheal intubation, incision infection, serum albumin(ALB) and lactic acid levels (all P<0.05). The results of multivariate logistic regression analysis showed that combined skull base fracture, cerebrospinal fluid otorrhea, rhinorrhea, intracranial rebleeding and the decrease of serum ALB level were risk factors of postoperative CNS infection in sTBI patients (all P<0.05). The ROC curve of the nomogram model showed that the AUC values of the training set and the validation set were 0.98 (95%CI: 0.96-0.99) and 0.95 (95%CI: 0.88-1.00), respectively, with high prediction efficiency. Bootstrap self-sampling method was used to verify the model. The results showed that the average absolute error between the predicted risk and the actual risk of the training set was 0.02 after 1 000 times of repeated sampling, indicating that the predicted value was basically consistent with the measured value, and had good prediction and identification abilities. The DCA curve showed that the nomogram model was more accurate than the single risk factor in predicting the occurrence of CNS infection, which could improve the clinical benefit rate. The CIC curve showed that the nomogram prediction model had high net clinical benefit.

The nomogram prediction model based on the risk factors of CNS infection after sTBI is simple to utilize, which is helpful for early screening of high-risk patients with CNS infection after sTBI, so as to achieve early diagnosis and early treatment, and reduce the incidence of CNS infection.

Craniocerebral trauma; Central nervous system infections; Risk factors; Nomo-grams; Forecasting