Indications for continuous electroencephalography and frequency of electrographic seizure detection in a pediatric and neonatal cardiovascular intensive care unit.

Objective: Seizures are a recognized complication of critical cardiovascular illness in infants and children. We assessed the diagnostic yield of continuous video-electroencephalography (cEEG) in a pediatric and neonatal cardiovascular intensive care unit (CVICU) by the symptoms and risk factors prompting cEEG evaluation.

Methods: This retrospective case series included all consecutive cEEGs in patients ≤21 years old performed in one CVICU over 38 months. cEEG indications were categorized as (1) index symptoms of concern and/or (2) clinical risk factors. Index symptoms were divided into (1) vital sign symptoms (i.e., heart rate, blood pressure, oxygen, respiration, or temperature) and (2) non-vital sign symptoms (i.e., mental status, abnormal movements, eye findings, weakness, or failed extubation). Indications for cEEG were extracted by manual chart review. The presence of seizures was established electrographically from neurophysiologist reports.

Results: There were 605 cEEGs from 411 patients. The median study was 26 h (25%-75%, interquartile range = 20-41 h). Seizures were detected in 57 of 605 (9%) cEEGs overall; in 34 of 356 (10%) cEEGs obtained for risk factors alone (odds ratio [OR] = 1.03, 95% confidence interval [CI] = .60-1.82, p = .90), 0 of 104 (0%) for isolated vital sign changes (p < .001), 10 of 101 (10%) for symptoms not involving vital signs (OR = 1.06, 95% CI = .52-2.09, p = .88), and in 13 of 44 (30%) for both vital sign and non-vital sign symptoms (OR = 4.93, 95% CI = 2.45-9.77, p < .001). On univariate analysis, symptoms involving gaze deviation, abnormal limb movements, or intermittent oxygen desaturation, and the risk factors of preexisting epilepsy, recent neurosurgery, acute stroke, and cardiac air embolism were associated with seizures (p < .05).

Significance: There were zero electrographic seizures in cEEGs obtained for isolated vital sign changes, whereas cEEGs obtained for the combination of vital sign changes and other non-vital sign symptoms were five times more likely to detect electrographic seizures than cEEGs obtained based on risk factors alone.