Abstract
AIM: To characterize early amplitude-integrated electroencephalogram (aEEG) and single-channel EEG (aEEG/EEG) in very preterm (VPT) infants for prediction of long-term outcome. PATIENTS: Forty-nine infants with median (range) gestational age of 25 (22-30) weeks. METHODS: Amplitude-integrated electroencephalogram/EEG recorded during the first 72 h and analysed over 0-12, 12-24, 24-48 and 48-72 h, for background pattern, sleep-wake cycling, seizures, interburst intervals (IBI) and interburst percentage (IB%). In total, 2614 h of single-channel EEG examined for seizures. Survivors were assessed at 2 years corrected age with a neurological examination and Bayley Scales of Infant Development-II. Poor outcome was defined as death or survival with neurodevelopmental impairment. Good outcome was defined as survival without impairment. RESULTS: Thirty infants had good outcome. Poor outcome (n = 19) was associated with depressed aEEG/EEG already during the first 12 h (p = 0.023), and with prolonged IBI and higher IB% at 24 h. Seizures were present in 43% of the infants and associated with intraventricular haemorrhages but not with outcome. Best predictors of poor outcome were burst-suppression pattern [76% correctly predicted; positive predictive value (PPV) 63%, negative predictive value (NPV) 91%], IBI > 6 sec (74% correctly predicted; PPV 67%, NPV 79%) and IB% > 55% at 24 h age (79% correctly predicted; PPV 72%, NPV 80%). In 35 infants with normal cerebral ultrasound during the first 3 days, outcome was correctly predicted in 82% by IB% (PPV 82%, NPV 83%). CONCLUSION: Long-term outcome can be predicted by aEEG/EEG with 75-80% accuracy already at 24 postnatal hours in VPT infants, also in infants with no early indication of brain injury.