Recalibrating the Glasgow Coma Score as an Age-Adjusted Risk Metric for Neurosurgical Intervention

BACKGROUND The Glasgow Coma Scale (GCS) score is the most frequently used neurologic assessment in traumatic brain injury (TBI). The risk for neurosurgical intervention based on GCS is heavily modified by age. The objective is to create a recalibrated Glasgow Coma Scale (GCS) score that accounts for an interaction by age and determine the predictive performance of the recalibrated GCS (rGCS) compared to the standard GCS for predicting neurosurgical intervention. METHODS This retrospective cohort study utilized the National Trauma Data Bank and included all patients admitted from 2010-2015 with TBI (ICD9 diagnosis code 850-854.19). The study population was divided into 2 subsets: a model development dataset (75% of patients) and a model validation dataset (remaining 25%). In the development dataset, logistic regression models were used to calculate conditional probabilities of having a neurosurgical intervention for each combination of age and GCS score, to develop a point-based risk score termed the rGCS. Model performance was examined in the validation dataset using area under the receiver operating characteristic (AUROC) curves and calibration plots. RESULTS There were 472,824 patients with TBI. The rGCS ranged from 1-15, where rGCS 15 denotes the baseline risk for neurosurgical intervention (4.4%) and rGCS 1 represents the greatest risk (62.6%). In the validation dataset there was a statistically significant improvement in predictive performance for neurosurgical intervention for the rGCS compared to the standard GCS (AUROC: 0.71 versus 0.67, difference, -0.04, P