Investigation of the predictive value of routine electroencephalogram compared to continuous electroencephalogram in...

Early identification of epileptiform activity in patients with traumatic brain injury (TBI) remains a clinical challenge, with important implications for prognosis and treatment. The primary objective of this prospective single-center study was to determine the incremental diagnostic yield of 1 h EEG (1hEEG) and continuous EEG (cEEG; 24-h long EEG segment) compared with routine EEG (rEEG; first 30 min recording) for detecting interictal epileptiform discharges (IEDs) and electrographic seizures in adults with acute TBI and acute intracranial hemorrhage and/or brain contusion(s). A secondary exploratory objective was to assess whether clinical, neuroimaging, and EEG characteristics were associated with similarity between shorter recordings and the corresponding prolonged recording. Thirty-five adults underwent cEEG during ICU admission. Monitoring began a median of 2 days after TBI (range, 0-7 days) and lasted a median of 3 days (range, 1-9 days; mean, 3.8 days), yielding 134 analyzable segments (mean 3.8 segments per patient). For each segment, the first 30 min and first 60 min were compared with the full cEEG segment. IEDs were identified in 14/134 segments (10.4%) during the first 30 min and first hour and detection was increased to 19/134 segments (14.2%) during prolonged monitoring; thus, 5/19 IED-positive segments were detected only after the first hour. Electrographic seizures were identified during cEEG in 3/35 patients (8.6%); two had seizures within the initial 30 min, whereas one had a seizure only during prolonged monitoring. All seizures were electrographic/nonconvulsive. Prolonged EEG increased the detection of IEDs in a subset of recordings and identified an electrographic seizure that was missed by shorter recordings. Reduced level of consciousness, greater TBI severity, sedative-drug exposure, and additional trauma to other body parts were associated with a greater similarity (shorter mathematical distance) between rEEG, 1hEEG and cEEG findings (all p < 0.05). Clinical variables may help prioritize patients in whom shorter EEG recordings are more representative of prolonged EEG; however, these exploratory findings should not be used to exclude high-risk patients from cEEG.