(798) Feasibility of a Novel EEG Scoring System to Assess Sleep in Critically Ill Children

Introduction: Sleep disturbance may be a modifiable risk factor for delirium and cognitive decline in critically ill children. Because polysomnography, the gold standard for sleep analysis, is time and cost intensive, we sought to devise a novel electroencephalogram (EEG) based scoring system to analyze sleep and assessed its feasibility.

Methods: In this retrospective observational cohort study across two academic pediatric intensive care units, available EEG data in patients ages 6 months to 18 years, were analyzed for sleep. Patients with pre-existing epileptic encephalopathy or a final diagnosis of brain death were excluded. EEGs were scored for sleep by pediatric neurologists using a four-category system: category 1- invariable background, no state cycling or reactivity, category 2- reactivity and variability present, no discernable sleep architecture, category 3- discernable sleep architecture, abnormal pattern of sleep cycle, and category 4- normal sleep including distribution; they also assessed for presence of N2, N3, rapid eye movement (REM), and total time spent in sleep. Other variables of interest included indication for EEG, presence of encephalopathy, sedative exposure, and severity of illness. Variables were compared with non-parametric tests (all data median [IQR]).

Results: Fifty-one subjects met criteria for inclusion, 25 from site one and 26 from site two. Thirty-four (67%) of the patients had no pre-existing neurological comorbidity, 40 (78%) had a diagnosis of acute encephalopathy, and subjects had a similar severity of illness assessed by Pediatric Logistic Organ Dysfunction (PELOD) score (site 1: 7 [4, 8.5], site 2: 5.5 [2, 8.25], p=0.139). Category 3 sleep was observed in 24 (47%) and category 4 in 9 (18%). There were 33 (65%) patients with presence of N2, 24 (47%) with presence of N3, and 13 (25%) with presence of REM.

Conclusions: In this small cohort, pediatric neurologists were able to quantify sleep in critically ill children using a novel EEG-based scoring system. A larger cohort study is warranted to compare this sleep categorization to polysomnography. If clinically validated, this scoring system could facilitate elucidation of modifiable risk factors leading to sleep interruption in the PICU.