Neurosurgery Blog: Outcome Prediction in Moderate and Severe Traumatic Brain Injury: A Focus on Computed Tomography Var

Abstract

Background

With this study we aimed to design validated outcome prediction models in moderate and severe traumatic brain injury (TBI) using demographic, clinical, and radiological parameters.

Methods

Seven hundred consecutive moderate or severe TBI patients were included in this observational prospective cohort study. After inclusion, clinical data were collected, initial head computed tomography (CT) scans were rated, and at 6 months outcome was determined using the extended Glasgow Outcome Scale. Multivariate binary logistic regression analysis was applied to evaluate the association between potential predictors and three different outcome endpoints. The prognostic models that resulted were externally validated in a national Dutch TBI cohort.

Results

In line with previous literature we identified age, pupil responses, Glasgow Coma Scale score and the occurrence of a hypotensive episode post-injury as predictors. Furthermore, several CT characteristics were associated with outcome; the aspect of the ambient cisterns being the most powerful. After external validation using Receiver Operating Characteristic (ROC) analysis our prediction models demonstrated adequate discriminative values, quantified by the area under the ROC curve, of 0.86 for death versus survival and 0.83 for unfavorable versus favorable outcome. Discriminative power was less for unfavorable outcome in survivors: 0.69.

Conclusions

Outcome prediction in moderate and severe TBI might be improved using the models that were designed in this study. However, conventional demographic, clinical and CT variables proved insufficient to predict disability in surviving patients. The information that can be derived from our prediction rules is important for the selection and stratification of patients recruited into clinical TBI trials.

Content Type Journal Article
Category Original Research
Pages 1-11
DOI 10.1007/s12028-012-9795-9
Authors
- Bram Jacobs, Department of Neurology (935), Radboud University Nijmegen Medical Centre (RUNMC), P.O. Box 9101, 6500 HB Nijmegen, The Netherlands
- Tjemme Beems, Department of Neurosurgery (931), Radboud University Nijmegen Medical Centre, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands
- Ton M. van der Vliet, Department of Radiology, RUNMC, Nijmegen, The Netherlands
- Arie B. van Vugt, Department of Emergency Medicine, RUNMC, Nijmegen, The Netherlands
- Cornelia Hoedemaekers, Department of Intensive Care Medicine (632), Radboud University Nijmegen Medical Centre, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands
- Janneke Horn, Department of Intensive Care Medicine, Academic Medical Centre, University of Amsterdam, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands
- Gaby Franschman, Department of Anesthesiology, VU University Medical Centre, Amsterdam, De Boelelaan 1117, 1081 HV Amsterdam, The Netherlands
- Ian Haitsma, Department of Neurosurgery, Erasmus Medical Centre, Rotterdam, P.O. Box 2040, 3000 CA Rotterdam, The Netherlands
- Joukje van der Naalt, Department of Neurology, University Medical Centre Groningen, University of Groningen, P.O. Box 30.001, 9700 RB Groningen, The Netherlands
- Teuntje M. J. C. Andriessen, Department of Neurology (935), Radboud University Nijmegen Medical Centre (RUNMC), P.O. Box 9101, 6500 HB Nijmegen, The Netherlands
- George F. Borm, Department of Epidemiology, Biostatistics and HTA (133), Radboud University Nijmegen Medical Centre, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands
- Pieter E. Vos, Department of Neurology (935), Radboud University Nijmegen Medical Centre (RUNMC), P.O. Box 9101, 6500 HB Nijmegen, The Netherlands