Continuous monitoring of short-latency somatosensory evoked potentials during cardiac and aortic surgery.

The effectiveness of monitoring somatosensory evoked potentials (SEPs) intraoperatively to detect brain damage early remains controversial. To assess the diagnostic accuracy of this modality, a study was conducted between 1991 and 1994, recording SEPs in 287 consecutive patients undergoing cardiac and aortic surgery using cardiopulmonary bypass (CPB) with moderate hypothermia or deep hypothermic circulatory arrest. From P1 to N2 of the SEPs occurring within 50 ms latency in response to electrical stimulation of the median nerve were recorded over the contralateral postcentral cortex at 5-min intervals using a Neuropack-2 (Nihon Koden, Tokyo, Japan). Normal SEPs were recovered in 247 patients postoperatively; however, 2 of these patients had suffered a cerebral infarction and 1, a transient stroke intraoperatively, demonstrating a false-negative incidence of 1.2%. On the other hand, three different types of abnormal SEPs were recorded postoperatively. P1 and N1 absence, probably caused by a subcortical lesion, was observed in 4 patients; P2 and N2 absence, probably caused by a cortical lesion, was observed in 8 patients; and a flat SEP, representing diffuse damage, was observed in 2 patients. Among these 14 patients with abnormal SEPs, 7 showed no neurologic disturbance at all, demonstrating a false-positive incidence of 50%. Thus, we concluded that when normal SEPs are recovered during weaning from CPB, the incidence of brain damage could be predicted at below 5%. Conversely, when abnormal SEPs are demonstrated, the incidence of brain dysfunction impeding a return to active life is estimated to be about 70%.