Prediction of minimally conscious state with somatosensory evoked potentials in long-term unconscious patients after traumatic brain injury.

BACKGROUND

To evaluate the predictive power of somatosensory evoked potentials (SEPs) for minimally conscious state (MCS) in long-term unconscious patients after traumatic brain injury (TBI).

METHODS

SEPs were recorded in 58 patients with duration of unconsciousness >30 days after TBI. SEPs were classified into three grades. Predictors including age, sex, Glasgow Coma Scale (GCS), and cause of injury were also analyzed, respectively. The outcome was divided into two groups including unconscious group and MCS group. The outcome was assessed at 12 months after TBI.

RESULTS

In 58 patients, 22 of 58 were minimally conscious, 3 of 58 dead, and 33 of 57 were still in vegetative state at 12 months after TBI. SEPs grade (p = 0.001) and GCS (p = 0.010) were significantly associated with the outcome. The area under the receiver operator characteristic curve of SEPs was 0.891 ± 0.048 (p < 0.001; 95% confidence interval, 0.798-0.984) for predicting outcome, and of GCS score was only 0.746 ± 0.066 (p = 0.002; 95% confidence interval, 0.616-0.876). The accuracy of the whole model for predicting unconscious and MCS was 91.7% and 86.4%, respectively. The overall correct prediction was as high as 89.7% (p < 0.001).

CONCLUSIONS

SEPs are excellent in predicting the outcome of long-term unconscious patients after TBI. SEPs should be considered more often and more routinely used after TBI.