Objective. To define normal and abnormal patterns, test interobserver variability, and the prognostic accuracy of amplitude-integrated electroencephalography (aEEG) soon after the onset of neonatal encephalopathy. Methods. Consecutive cases of neonatal encephalopathy (n = 56; gestation median, 40; range, 35-42 weeks) and healthy infants (n = 14; gestation median, 40; range, 39-40 weeks) were studied. aEEG was recorded using a cerebral function monitor, at median, 0, range, 0-21 days of age. Of the infants, 24 of the 56 with encephalopathy and all of the normal infants were studied within 12 hours of birth (median, 5; range, 3-12 hours). Forty infants were suspected of having suffered birth asphyxia. Criteria for normal and abnormal patterns were defined and the interobserver variability of these classifications determined. Results were compared with neurodevelopmental outcome assessed at 18 to 24 months of age, aEEG also was compared with a standard EEG and with magnetic resonance imaging. Results. The median upper margin of the widest band of aEEG activity in the control infants was 37.5 mu V (range, 30-48 mu V), and median lower margin was 8 mu V (range, 6.5-11 mu V). We classified the aEEG background activity as normal amplitude, the upper margin of band of aEEG activity >10 mu V and the lower margin >5 mu V; moderately abnormal amplitude, the upper margin of band of aEEG activity >10 mu V and the lower margin less than or equal to 5 mu V; and suppressed amplitude, the upper margin of the band of aEEG activity <10 mu V and lower margin <5 mu V. Recordings were analyzed further for the presence of seizures, defined as periods of sudden increase in voltage accompanied by a narrowing of the band of aEEG activity. Tests of interobserver variability showed excellent agreement both for assessment of amplitude (kappa statistic = 0.85) and for identification of seizures (kappa statistic = 0.76) There was a close relationship between the aEEG and subsequent outcome: 19 of 21 infants with a normal aEEG finding were normal on follow-up at 18 to 24 months of age, whereas 27 of 35 infants with a moderately abnormal or suppressed aEEG and/or seizures died or developed neurologic abnormalities. Thus, aEEG predicted outcome with a sensitivity of 0.93, a specificity of 0.70, positive predictive value of 0.77, negative predictive value of 0.90, and the likelihood ratio of a positive result of 3.1 and a negative result of 0.06. For the 24 infants studied within 12 hours of birth, the corresponding results were sensitivity, 1.0; specificity, 0.82; positive predictive value, 0.85; negative predictive value, 1; likelihood ratio of a positive result, 5.5; and likelihood ratio of a negative result, 0.18. Conclusion. The aEEG is a simple but accurate and reproducible clinical tool that could be useful in the assessment of infants with encephalopathy.
