Predictive factors of mortality in severe craniocerebral trauma

BACKGROUND: The prediction of mortality in severe head injury is of interest for the evaluation of patient prognosis and to also permit therapeutic measures to be considered and improve knowledge of this problem. METHODS: A multiple logistic regression model was developed and validated from the data obtained from a series of 231 patients hospitalized for the above mentioned condition. Seventy-five percent of the patients were used to define the model with the remaining 25% validating the same. RESULTS: The variables included in the model were: intraventricular hemorrhage, odds ratio = 20.4 (confidence interval 95%: 3.56-116); compression of basal cistern and/or of the III ventricle, odds ratio = 11.5 (4.43-29.8); mydriatic pupils in both eyes, odds ratio = 5.71 (1.32-24.6); age, odds ratio = 1.03 (1.01-1.05) and Glasgow scale upon admission, odds ratio = 0.57 (0.43-0.75). The maximum global value of the model (84.9%) corresponded to a sensitivity of 84.5% and specificity of 85.2%. The cut-off point of probability of death was found to be 0.475. In the validation of the model the highest global value (84.2%) was also observed at the cut-off point of 0.475 with a sensitivity of 84.2% and specificity of 84.4%. CONCLUSIONS: The predictive factors of mortality in severe head injury are the evidence of hemorrhage in the cerebral ventricles, mydriasis and a low score on the Glasgow scale. The model presented is useful and valid for carrying out the prediction of mortality at the time of admission and is also easy to apply since the variables used are obtained in the initial examination of the patients with severe head injury.