Citation

Lesko MM, Jenks T, O'Brien S, Perel P, Childs C, Bouamra O, Lecky F. J. Neurotrauma 2013; 30(24): 2021-2030.

Affiliation

The Trauma Audit and Research Network (TARN), Clinical Sciences Building , Salford Royal Hospital , Stott lane , Salford , United Kingdom, M6 8HD ; MehdiMLesko@gmail.com.

Copyright

(Copyright © 2013, Mary Ann Liebert Publishers)

DOI

10.1089/neu.2013.2988

PMID

23865489

Abstract

Currently available prognostic models in Traumatic Brain Injury (TBI) are derived from historical datasets or from heterogenous datasets in terms of the trauma care delivered. The objective of our study was to develop models to predict survival in a recent cohort of TBI patients within a relatively homogeneous trauma care system. Records of patients with brain injury were extracted from the Trauma Audit and Research Network (TARN) database. The relationship of the variables (i.e. age, Glasgow Coma Score (GCS), pupillary reactivity, Injury Severity Score (ISS), Computed Tomography (CT) classifications, classification of various intracranial pathologies, systolic and mean blood pressure, oxygen saturation and the presence of extracranial injury) to survival at discharge were determined. Stepwise logistic regression analysis was performed to determine the best prognostic model. Two models were derived from data of 802 patients (models A and B). Age, GCS, pupillary reactivity, hypoxia and brain stem injury are significant predictors in both. However, model A contains ISS in contrast to model B with the presence of brain swelling and major extracranial injury instead. Both models have good predictive performance (Model A; Area Under the ROC Curve (AUC) = 0.92 (95% CI: 0.90-0.95), Nagelkerke R2: 0.62, Model B; AUC = 0.93 (95% CI: 0.91-0.95), Nagelkerke R2: 0.63). Two accurate and reliable prognostic models were developed from a recent cohort of TBI population.

Language: en