Predictors of nonconvulsive seizures among critically ill children

Abstract

Continuous electroencephalography (EEG) monitoring is a valuable tool for the detection of seizures among critically ill children, in particular when these seizures occur without clinical signs: termed nonconvulsive seizures. Continuous EEG monitoring is a limited resource in many centers. We sought to identify which critically ill children most frequently experience nonconvulsive seizures, and thus may particularly benefit from continuous EEG monitoring. Single-center review was undertaken of consecutive diagnostic continuous EEG (cEEG) recordings performed in our pediatric and neonatal intensive care units (ICUs). We examined the indications for monitoring, the clinical characteristics of monitored patients, the occurrence and timing of seizures, and clinical and EEG characteristics associated with nonconvulsive seizures. One hundred twenty-one patients underwent diagnostic continuous EEG monitoring, for a mean duration of 26 h. Seizures were detected in 32% of these patients, of which 90% experienced some nonconvulsive seizures, and 72% experienced exclusively nonconvulsive seizures. Patients with nonconvulsive seizures had significantly greater odds of having acute epilepsy, acute structural brain injury, prior in-hospital convulsive seizures, and the presence of interictal epileptiform abnormalities on EEG. Seizures are common among critically ill children undergoing diagnostic cEEG monitoring. The great majority of these seizures are nonconvulsive, requiring EEG for their detection. Predictors of nonconvulsive seizures include acute epilepsy, acute structural brain injury, prior in-hospital convulsive seizures, and interictal epileptiform abnormalities on EEG. These findings can help inform future allocation of limited cEEG monitoring resources to those patients at greatest risk for nonconvulsive seizures.