This study aims to determine whether EEG can serve as an objective biomarker of delirium.

This single-center retrospective study included critically ill children (>28 days and <18 years) without prior neurologic injury with a positive (CAPD ³9, CAPD⁺) or negative (CAPD<9, CAPD^-) delirium screen at the time of EEG monitoring. EEG features and outcome metrics were assessed for associations with CAPD screen. Descriptive statistics were used.

Ninety-three patients met inclusion criteria. Median age was 4.36 years (interquartile range (IQR) 0.81-9.66). Sixty-eight patients were CAPD⁺ and 25 were CAPD^-. Normal background activity was seen in 21% of CAPD⁺ vs 56% of CAPD^- children (p=0.002). Posterior dominant rhythm was present in 22% of CAPD⁺ vs 64% of CAPD^- children (p<10^-3). Normal sleep features were observed in 37% of CAPD⁺ vs 76% of CAPD^- children (p=0.001). Generalized slowing and absence of sleep features occurred more frequently in CAPD⁺ than in CAPD^- children (59% vs 20%, p=0.001; 40% vs 12%, p=0.01, respectively). Attenuated featureless activity, sporadic epileptiform discharges and focal slowing occurred only in CAPD⁺ children (15%, p=0.06; 18%, p=0.03; 75% p=0.04, respectively).

CAPD⁺ children had worse neurologic outcomes (discharge PCPC score >2) than CAPD^- children (37% vs 0%, p<10^-4). PICU LOS was longer in CAPD⁺ than in CAPD^- children (5.5 days (3-23) vs 2 days (1-3); p<10^-4). More CAPD⁺ than CAPD^- children were discharged to a rehabilitation facility or another hospital (32% vs 4%, p=0.006).

Specific EEG features in critically ill children are associated with a positive or negative delirium screen. Additionally, a positive delirium screen is associated with worse outcomes. This data suggests EEG features may help identify delirium and perhaps facilitate its early prevention in critically ill children.