Abstracts

Rationale: Continuous electroencephalographic monitoring (cEEG) is frequently used in critically ill children admitted in the pediatric intensive care unit (PICU) for seizure detection. cEEG is resource intense and limited in many hospitals. Early EEG biomarkers for seizures are needed to allocate this resource towards patents most likely to have electrographic seizures. The aim of this study is to assess the utility of early features of interictal epileptiform discharges (IED) and EEG background as predictors of ES in the PICU. Methods: Prospective study of pediatric patients from 44 weeks gestational age to 21 years who underwent a clinically indicated cEEG in the PICU from May 2016 to October 2017. cEEG was defined as greater than three hours of uninterrupted EEG. Patients were excluded if they had a prior diagnosis of epilepsy. ES were defined as any seizure detected on cEEG, whether electro-clinical or electrographic-only. The EEG background was categorized as normal, slow disorganized, attenuated and featureless, discontinuous, and burst suppression. Details of the EEG background included asymmetry, presence or absence of IEDs and time of first IED from the start of cEEG. The frequency of IEDs and presence of rhythmic or periodic patterns were described using the American Clinical Neurophysiology Society Standardized Critical Care EEG terminology. Results: In 139 patients, 52% (72/139) were male with a median age of 2.8 years (IQR 0.5 – 8.8).  The most common admitting diagnosis was new onset seizures in 29% (40/139). The EEG background was slow/disorganized in 76% (105/139), and showed an asymmetry in 28% (39/139) within the first hour of recording.  IEDs were identified in 34% (47/139), with 74% (35/47) having sporadic epileptiform discharges(SED). The remaining 26% (12/47) of IEDs consisted of periodic epileptiform discharges (PED) at 67% (8/12) or a rhythmic pattern 33% (4/12). The majority of PEDs were lateralized periodic discharges (LPD).  Of the SED, 68% (25/37) were occasional or rare with the remainder being frequent or abundant.  Nineteen percent (27/139) of patients had ES, of which 81% (22/27) had IEDs with 52% (14/27) having a background asymmetry. Patients with frequent or abundant SEDs had a higher likelihood of ES (83%. vs. 20%, p<0.001). Of note, the two patients with frequent discharges without an ES on EEG were  treated prior to cEEG for clinical seizures. Patients with periodic patterns were more likely to have an ES (75% vs. 17%, p<0.001). Of note, three patients with ES had both SEDs and PEDs. In patients with a normal background, or a slow disorganized background without IED or asymmetry, 3% (2/63) had ES.  The median time from EEG start to the first IED was 0.78 hours (IQR 0.13 – 1.73) with the median time of the first IED to ES being 1.75 hours (IQR 0.075 – 7.17). Conclusions: In this study we have found that more frequent SEDs suggests a higher likelihood of seizures. In addition, a slow/disorganized background without asymmetry or SEDs predicts the absence of seizures. Finally, we have identified the duration necessary to exclude SED in our population, suggesting that the decision to transition to cEEG for seizure detection may be possible with a prolonged study. Funding: None