Objective: Somatosensory evoked potentials (SEP) have been recommended for use as a prognostic tool in the evaluation of patients with severe brain injury. Since the pathophysiology in traumatic injury differs from hypoxic ischemic injury, the findings and prognostic relevance of SEP may also differ. We critically appraise the evidence evaluating the predictive value of bilateral absence of SEPs in traumatic brain injured patients.

Methods: Studies were identified through MEDLINE, conference proceedings, citation review of relevant articles, and contact with expert informants. Inclusion criteria were: cohort study of traumatic head injury patients, minimum follow-up of 3 months post injury, use of a valid functional outcome score, and presence of patients with absent cortical (N20/P25) SEP in the presence of cervical (N13) response. In duplicate, independently, we abstracted data on the population, test characteristics, outcomes, and methodologic quality.

Results: The search yielded 44 prospective trials evaluating the predictive value of SEP in patients with brain injury; 12 trials allowed extraction of the results for patients with traumatic brain injury. Eight trials included pediatric patients but did not allow separate extraction of their data. The SEP recording was repeated at least twice in all studies to increase reliability. The timing of the initial SEP test varied from 24 hours to 33 day's post injury. Patients were followed for at least 6 months post injury in 83% of the studies. Three of 164 patients with absent cortical SEP on initial study had a good outcome (mild to moderate neurological disability). A 13 year old patient with a primary brain stem injury showed return of SEP response on day 5. A patient with a basilar skull fracture had an absent cortical SEP on day 2 post-injury but was withdrawing to painful stimuli. And a 3 year old with GOS=3 at 6 months improved on later follow-up. Bilateral absence of cortical SEP had a positive predictive value for poor outcome (severe disability, persistent vegetative state, or death) of 98.2% (95% Confidence Interval 96.2%, 100%).

Conclusions: Across studies, bilateral absence of cortical SEP was highly predictive of poor outcome in patients with traumatic brain injury. However, use in clinical decision making must be guarded because the prediction may be falsely pessimistic in 3.8% of cases, many studies had insufficient follow-up, the timing of the initial SEP and subsequent SEP tests varied, and experience in young children is minimal.