The purpose of this study was to compare median somatosensory evoked potent
ials (SEP) in patients undergoing carotid endarterectomy (CEA) with routine
shunting and nonshunting (excluding the option of selective shunting) and
to evaluate the significance of a decrease in the amplitude of the cortical
ly generated waveforms of the SEP and/or an increase in the central conduct
ion time (CCT) on the one hand, and that of a loss of the cortical SEP, on
the other. Somatosensory evoked potentials were recorded in 32 patients bef
ore, during, and after CEA with routine shunting or nonshunting. The N13 an
d N20 latency, the CCT, and the N20/P25 amplitude were evaluated. In additi
on, a meta-analysis of 15 previous studies was performed comprising a total
of 3,136 patients. The intraoperative cortical SEP showed no differences b
etween shunted and nonshunted patients, apart from the preclamping value of
the N20/P25 amplitude which was lower in the nonshunted subjects. The numb
er of patients with decreased and/or delayed cortical SEP (findings frequen
tly used as criterion for selective shunting) was similar in the two study
groups. A loss of the cortical SEP occurred in one patient operated on with
out an indwelling shunt. None of these patients had a new neurologic defici
t after surgery. In the meta-analysis, the positive predictive value of dec
reased and/or delayed cortical SEP was extremely poor, that of absent corti
cal SEP was poor to moderate and the prevalence of new neurologic deficits
was similar in patients undergoing CEA with routine shunting-nonshunting an
d those with selective shunting-nonshunting. Our study suggests that decrea
sed and/or delayed cortical SEP are unreliable predictors of the neurologic
al outcome of CEA patients and consequently an unsuitable criterion for sel
ective shunting. The meta-analysis confirms this finding and shows that the
neurologic outcome is not improved by using an indwelling shunt selectivel
y based on SEP monitoring.