Related Experiment Videos
Early prediction of mortality in isolated head injury patients: a new predictive model
Demetrios Demetriades1, Eric Kuncir, Carlos V R Brown
1Division of Trauma and Critical Care, Department of Surgery, LAC+USC Medical Center, Keck School of Medicine University of Southern California, Los Angeles, California 90033, USA. demtria@usc.edu
Background:
To construct a predictive model of survival in isolated head injury patients, on the basis of easily available parameters that are independent risk factors for survival outcome.
Methods:
Trauma registry-based study of head injury patients who had no other major extracranial injuries and were not hypotensive at admission. A predictive model of probability of death was constructed using discriminant analysis, on the basis of admission Glasgow Coma Scale (GCS) score, head Abbreviated Injury Score (AIS), age, and mechanism of injury.
Results:
The study included 7,191 patients with head trauma. The overall correct classification rate of the proposed predictive model was 94.2% as compared with 89.0% of the admission GCS score (p < 0.05) and 92.8% of the head AIS (p < 0.05). The correct classification rate of the predictive model developed for the severe head trauma (GCS score 4-8) patients was 79.9%, as compared with 72.6% using the admission GCS score alone or 75.1% (p < 0.05). A one-page, easy to use table summarizing the predicted mortality on the basis of GCS score, head AIS, mechanism of injury, and age was developed.
Conclusions:
The proposed model has a significantly better predictive power, especially in severe head trauma, than the extensively used GCS and head AIS. A simple table on the probability of death of a particular patient based on admission GCS score, head AIS, mechanism of injury and age of patient can provide instant information.