Objective: This study aims to identify risk factors for central nervous system (CNS) infection in elderly patients hospitalized with traumatic brain injury (TBI) and to develop a reliable predictive tool for assessing the likelihood of CNS infection in this population.
Method: We conducted a retrospective study on 742 elderly TBI patients treated at Tangdu Hospital, China. Clinical data was randomly split into training and validation sets (7:3 ratio). By conducting univariate and multivariate logistic regression analysis in the training set, we identified a list of variables to develop a nomogram for predicting the risk of CNS infection. We evaluated the performance of the predictive model in both cohorts respectively, using receiver operating characteristics curves, calibration curves, and decision curve analysis.
Results: Results of the logistic analysis in the training set indicated that surgical intervention (P = 0.007), red blood cell count (P = 0.019), C-reactive protein concentration (P < 0.001), and cerebrospinal fluid leakage (P < 0.001) significantly predicted the occurrence of CNS infection in elderly TBI patients. The model constructed based on these variables had high predictive capability (area under the curve-training = 0.832; area under the curve-validation = 0.824) as well as clinical utility.
Conclusions: A nomogram constructed based on several key predictors reasonably predicts the risk of CNS infection in elderly TBI patients upon hospital admission. The model of the nanogram may contribute to timely interventions and improve health outcomes among affected individuals.
Keywords: CNS infection; Elderly TBI; Logistic regression analysis; Nomogram; Risk prediction.
Copyright © 2023 Elsevier Inc. All rights reserved.