A prospective study of outcome predictors after severe brain injury in children

Objective: To directly compare the predictive powers of somatosensory evoked potentials (SEPs) to those of motor and pupillary responses. Design and setting: Prospective clinical study in a paediatric intensive care unit. Patients and participants: 102 severely brain-injured children less than 15 years of age. Measurements and results: SEPs and motor and pupillary responses were serially recorded during the first 9 days after admission. Initial, last and those tests performed on or after day 2 were analysed. Outcome was assessed 5 years after injury. SEPs had equal or superior predictive statistics and ROC curves compared to the other tests with few exceptions. Pupillary responses had higher sensitivity for favourable outcome prediction while for unfavourable outcome prediction the last motor responses had higher sensitivity, and the last pupillary responses had slightly higher specificity. Combining SEPs and motor responses provided the best combination for predicting unfavourable outcome. Conclusions: SEPs are the best overall predictor of outcome while motor and pupillary responses have advantages in some specific areas. The routine use of SEPs should be considered in the prediction of outcome of severely brain-injured patients.