Prospective evaluation of a predictive model of mortality in patients with isolated head injury

In a previous retrospective study, we developed a predictive model of survival in isolated head injuries based on easily available parameters such as age, mechanism of injury, Glasgow Coma Scale, and head Abbreviated Injury Scale (AIS). The purpose of the present study is to prospectively evaluate this predictive model.

Isolated head injuries admitted to a Level I urban trauma center were prospectively accrued from May 1, 2006 through April 30, 2007. Age, mechanism of injury, Glasgow Coma Scale, head AIS, and survival status were recorded for each patient. Patients with extracranial AIS >3, head AIS = 6, or hypotension were excluded. These data were entered into our previously developed predictive model and the percentage of correct classification was used to measure how well the predictive model predicted outcome. Sensitivity, specificity, positive and negative predictive values, and their 95% confidence intervals were calculated and compared with values obtained from our original, retrospective study.

Seven hundred eighty-six patients met the criteria for inclusion in the study with an overall mortality of 5.8% (46 patients). When entered into our predictive model, the percentage of correct classification rate was 92% compared with the 94% rate seen in the original study, which is better than other available predictive tools based on combined scoring systems such as the Trauma and Injury Severity Score methodology.

When evaluated prospectively, our predictive model has similar accuracy in predicting survival of all patients with head trauma as our original retrospective study and performs better than other predictive models such as the Trauma and Injury Severity Score methodology. This study demonstrates that a simple table based on easily obtained admission patient characteristics can rapidly provide information about the probability of survival in patients with head injuries.