Cerebral function monitoring in paediatric intensive care: useful features for predicting outcome.

Neurological integrity in sick children is difficult to assess clinically. The aim of this study was to determine the predictive value of EEG activity recorded with a bedside EEG analysing monitor in an intensive care unit. EEG activity was monitored in 108 children (age range 2 weeks to 16 years, median 1.7 years) considered at risk for cerebral abnormalities with a cerebral function analysing monitor (CFAM). Recordings were evaluated for features of background EEG activity including mean amplitude, frequencies, and symmetry. Electrical seizure activity was quantified if present. Predictive value of the EEG features was evaluated relative to the clinical neurological outcome after one year. Asymmetrical recordings were not seen in any child with a normal outcome. Suppression of background activity was seen in 75% of the children who died. Seizures were present in 68% of children with a poor outcome. Seventeen of the 32 children (65%) who died had prolonged seizures. Absence of seizures and the presence of superimposed fast EEG activity in response to benzodiazepine infusions correlated with good outcome. A combination of two or more predictive EEG features demonstrated >90% specificity and positive predictive likelihood of poor outcome. EEG features provide information about the functional cerebral integrity of sick children. Changes in cerebral activity detected by the CFAM aid decision making by providing such information readily at the bedside in intensive care.