Objective: To examine the utility of somatosensory evoked potential (SEP) p
eaks and CSF creatine kinase BE isoenzyme activity (CKBB) in predicting non
awakening from coma due to cardiac arrest. Background: Accurate predictors
of neurologic outcome in patients comatose after cardiac arrest are needed
to improve medical decision making. Methods: A total of 72 comatose patient
s had bilateral median SEPs, and of these, 52 had CSF and CKBB. Awakening w
as defined as following commands or having comprehensible speech. Both shor
t (N1) and long (N3) latency SEP peaks were analyzed. Nonparametric analyse
s were used. Results: For patients who had both tests, CKBB greater than or
equal to 205 U/L predicted nonawakening with a sensitivity of 49% and a sp
ecificity of 100%. Bilateral absence of the N1 peak predicted nonawakening
with a sensitivity of 53% and a specificity of 100%. Using CKBB greater tha
n or equal to 205 Un, bilaterally absent SEP N1 peaks, or both predicted no
nawakening with a sensitivity of 69% and a specificity of 100%. Using CKBB
greater than or equal to 205 U/L, bilaterally absent N1 peaks, bilateral N3
greater than or equal to 176 msec or absent, or some combination predicted
nonawakening with a sensitivity of 78% and a specificity of 100%. Conclusi
on: The combination of an absent N1 peak and elevated CKBB performs better
than either alone in predicting nonawakening after cardiac arrest. Prolonge
d or absent N3 latency may increase sensitivity. These results should be in
terpreted with caution given the small number of patients and the possibili
ty of a self-fulfilling prophecy.