Predicting survival from head trauma 24 hours after injury: a practical method with therapeutic implications

To develop a method to predict long-term outcome after head injury and determine if outcome can be accurately predicted 24 hours after injury.

A retrospective review was performed on a study cohort of 672 head-injured patients admitted in coma (Glascow Coma Scale score < or = 8) who remained comatose for at least 6 hours, survived more than 24 hours, and had 6-month outcome data available. Stepwise logistic regression analysis was used to determine which clinical variables predicted 6-month outcome. Statistically significant clinical predictors were combined into a single examination variable (MPX score), which reflected a rank-ordering of examinations from worst to best, which was then further weighted by patient age. The relation between 6-month outcome and MPX score at admission and 24 hours was plotted and analyzed.

Age, best motor score, and pupillary reactivity at admission and 24 hours were significant predictors of outcome; extraocular motility was predictive at 24 hours only. Age was the most important independent predictor, followed by best motor score, pupillary reactivity, and extraocular motility. Combining these predictors into MPX score resulted in a set of graphs that reliably predicted long-term outcome. The 24-hour MPX data were better predictors of 6-month outcome and were more specific in predicting negative outcomes than admission data.

The method is simple to use, relying on bedside neurologic examination and a single graph, but appears to predict long-term outcome accurately as early as 24 hours after head injury. If validated on other large series of patients, this method could provide an objective and practical basis for terminating care in patients unlikely to survive a head injury.