To examine whether using an amplitude-integrated electroencephalography (aEEG) severity pattern as an entry criterion for therapeutic hypothermia better selects infants with hypoxic-ischemic encephalopathy and to assess the time-to-normal trace for aEEG and magnetic resonance imaging (MRI) lesion load as 24-month outcome predictors.

Forty-seven infants meeting Norwegian therapeutic hypothermia guidelines were enrolled prospectively. Eight-channel EEG/aEEG was recorded from 6 hours until after rewarming, and read after discharge. Neonatal MRI brain scans were scored for summated (range 0-11) regional lesion load. A poor outcome at 2 years was defined as death or a Bayley Scales of Infant-Toddler Development cognitive or motor composite score of <85 or severe hearing or visual loss.

Three severity groups were defined from the initial aEEG; continuous normal voltage (CNV; n = 15), discontinuous normal voltage (DNV; n = 18), and a severe aEEG voltage pattern (SEVP; n = 14). Any seizure occurrence was 7% CNV, 50% DNV, and 100% SEVP. Infants with SEVP with poor vs good outcome had a significantly longer median (IQR) time-to-normal trace: 58 hours (9-79) vs 18 hours (12-19) and higher MRI lesion load: 10 (3-10) vs 2 (1-5). A poor outcome was noted in 3 of 15 infants with CNV, 4 of 18 infants with DNV, and 8 of 14 infants with SEVP. Using multiple stepwise linear regression analyses including only infants with abnormal aEEG (DNV and SEVP), MRI lesion load significantly predicted cognitive and motor scores. For the SEVP group alone, time-to-normal trace was a stronger outcome predictor than MRI score. No variable predicted outcome in infants with CNV.

Selection of infants with encephalopathy for therapeutic hypothermia after perinatal asphyxia may be improved by including only infants with an early moderate or severely depressed background aEEG trace.