Abstracts

Rationale: Seizures and spells are a common symptom of abusive head trauma (AHT) in infants. However, AHT is rarely considered as an etiology in the differential diagnosis of new-onset spells. The Pittsburgh Infant Brain Injury Score (PIBIS) is a validated clinical decision-making tool used in pediatric emergency rooms to help determine which infants should undergo urgent neuroimaging to screen for intracranial hemorrhage (ICH), but its use in the outpatient setting is unknown. This study combines the visual analysis of EEG, quantitative EEG (qEEG) analysis, and PIBIS to better determine which infants presenting with spells and seizure-like activity might benefit from urgent neuroimaging to screen for AHT. Methods: A retrospective analysis of infants age 30-364 days old who underwent EEG for evaluation of spells or seizure-like activity in calendar year 2017 and who had neuroimaging within 180 days of EEG was performed. Infants with a prior history of ICH and/or clinically significant neuroimaging abnormalities except for ICH were excluded. Cases or controls were classified based on the presence of ICH. EEGs were reviewed by a board-certified epileptologist for traditional visual analysis who was blinded to clinical information except for age. The Persyst ICU monitoring panel was applied, and the maximum Relative EEG Asymmetry Index (REASI) percent was measured. REASI of >=20% was used as a cut-off for significant asymmetry. The 5-point Pittsburgh Infant Brain Injury Score (PIBIS), which consists of head circumference, serum hemoglobin, age, and abnormalities on dermatologic examination, was retrospectively calculated. A PIBIS >=2 was used as a cut-off for increased ICH risk.  Results: Figures 1 and 2: The final cohort was comprised of 39 subjects with six cases, including three cases most likely consistent with AHT. EEG was abnormal by visual inspection in 2/6 cases (33%) and 2/33 controls (6.1%), p=0.10. EEG was abnormal either by visual inspection or by qEEG analysis in 3/6 cases (50%) and 5/33 controls (15%), p =0.088. Epileptiform activity was only seen by visual analysis in 2/6 cases (33%) and 1/33 controls (3.0%), p=0.056. Asymmetry was identified with visual analysis in 2/6 (33%) cases and 1/33 controls (3.3%), p=0.056 and with qEEG analysis in 3/6 cases (50%) and 5/33 controls (15%), p=0.088. A PIBIS score of >=2 was found in 5/6 (83%) of cases and 16/33 (49%) of controls, p=0.19. When visual analysis, qEEG analysis, and PIBIS were concurrently applied, no cases of ICH were missed.  Conclusions: AHT is rare, yet it is one of a few situations requiring urgent neuroimaging in infants with spells or seizures. By combining traditional visual analysis, qEEG, and PIBIS, the detection rate of ICH improved. While the study was underpowered to detect significant differences due to limited size, the approach to combine these three modalities showed promise. Subtle asymmetry detected on qEEG may be clinically relevant. In infants with spells or seizures, applying PIBIS criteria in addition to qEEG detection of asymmetry may be helpful to decide whether to pursue urgent neuroimaging. Funding: No funding