Typical and Atypical Frequency-Based Subtypes of EEG Reactivity in Hospitalized Patients and Association With Comorbidities and Outcomes.

Abstract

EEG reactivity (EEG-R) is currently classified as present, absent, or stimulus-induced rhythmic, periodic, or ictal discharges, per the American Clinical Neurophysiology Society's 2021 Critical Care EEG Terminology. However, our previous study suggested that there may be other forms of EEG-R that warrant distinction, potentially associated with different comorbidities and outcomes. We, therefore, aimed to define "typical" vs "atypical" EEG-R, referring to increased power of arrhythmic alpha/beta frequencies vs delta frequencies after stimulation, respectively, and to identify their associated comorbidities and outcomes. This was a retrospective single-center study, examining adult patients admitted between 2015 and 2022, who underwent continuous EEG (cEEG) as standard-of-care work-up for their illnesses. Univariate testing and logistic regression were performed. The primary patient outcome at hospital discharge was categorized by the modified Rankin Scale (0-2: good; 3-6: poor). The primary independent variable was the presence of typical vs atypical EEG-R. Additional covariates included age, presence of critical illness, and acute intracranial and nonintracranial pathologies. A total of 238 patients with typical EEG-R (median age 54 years, female 50%) and 322 with atypical EEG-R (median age 60 years, female 50%) were identified. Patients with typical EEG-R more commonly underwent cEEG for spell capture of non-neurological episodes (20% vs 0%, Δ20%, 95% CI 15-25, <i>p</i> < 0.0001) or primary seizures (13% vs 2%, Δ11%, 95% CI 7-16, <i>p</i> < 0.0001). Patients with atypical EEG-R were more critically ill (88% vs 32%, Δ56%, 95% CI 49-63, <i>p</i> < 0.0001) with more intracranial (69% vs 45%, Δ24%, 95% CI 15-32, <i>p</i> < 0.0001) and nonintracranial pathologies (92% vs 38%, Δ54%, 95% CI 48-61, <i>p</i> < 0.0001) and concomitant altered mental status (47% vs 30%, Δ17%, 95% CI 9-25, <i>p</i> < 0.0001). These patients were hospitalized longer (median 24.0 vs 7.0 days, Hodges-Lehmann estimator 14 days, 95% CI 12-17, <i>p</i> < 0.0001) with higher rates of poor outcomes (74% vs 22%, Δ52%, 95% CI 45-59, <i>p</i> < 0.0001). In the logistic regression model, the inverse association between atypical EEG-R and good outcome remained significant after adjusting for other covariates (odds ratio 0.20, 95% CI 0.11-0.34, <i>p</i> < 0.0001). Granular, frequency-based assessments of EEG-R, such as the differentiation between typical vs atypical EEG-R, may provide further insight into the potential outcomes of patients with altered states of consciousness. Limitations of this study include the single-center retrospective nature.