Parameters and grading of evoked potentials: prediction of unfavorable outcome in patients with severe stroke.

To explore the prognostic accuracy for unfavorable outcome with short-latency somatosensory evoked potential (SLSEP) and brainstem auditory evoked potential (BAEP) in patients with severe stroke, 100 acute severe supratentorial stroke patients [Glasgow Coma Scale, (GCS) <or=12] were monitored with SLSEP, BAEP, and GCS within 1 week after onset. All patients were evaluated with modified Rankin scale 6 months after onset-0 to 4 being favorable outcome, whereas 5 to 6 designated unfavorable outcome. Correlations between SLSEP, BAEP, or GCS and prognosis were analyzed respectively. Of all 100 patients, 92 suffered unfavorable outcome, only 8 had favorable outcome. GCS 3 to 8 was significantly correlated with unfavorable outcome (P < 0.01). Absence of cortical SLSEP (N20 response, at least one side), abnormal bilateral N20-P25 amplitude ratio, Cant grading 3, Judson grading 3, and Haupt grading 3 to 4 were all significantly correlated with unfavorable outcome (P < 0.01). Poor differentiation or absence of BAEP wave V, Cant grading 3, Hall grading 3 to 4, and Haupt grading 3 to 4 were all significantly correlated with unfavorable outcome (P < 0.01). Absence of SLSEP N20, poor differentiation or absence of BAEP wave V, and Hall and Haupt grading of BAEP were acceptably consistent with prognosis (kappa >0.4). Prognostic accuracy of SLSEP, BAEP, and GCS for unfavorable outcome was very high (94.3%-98.7%), among which BAEP is the highest (97.5%-98.7%). The overall prognostic accuracy of SLSEP (90.5%-93.7%) was higher than BAEP (83.0%-89.4%) and GCS (82%). The prognostic accuracy of SLSEP and BAEP for unfavorable outcome in patients with severe supratentorial stroke was high, whereas for favorable outcome, it was low. The overall prognostic accuracy was higher than GCS.