Abstracts

Rationale: Continuous EEG (cEEG) has been increasingly used as part of neuromonitoring in intensive care units (ICU). Retrospective and small prospective studies showed that cEEG characteristics may be helpful in predicting outcomes in critically ill patients. The goal of this study was to prospectively assess the value of cEEG as a predictor of outcomes in patients admitted to ICU. Methods: Data were prospectively collected from 112 ICU patients who received any cEEG. Patients admitted because of seizures/epilepsy were excluded. Explanatory variables included age, gender, medical comorbidities, cEEG duration, admission Glasgow Coma Scale (GCSA), and duration of hospital and ICU stay. EEG parameters investigated included background rhythm (BR), focal slowing (FS), and reactivity in the first and last 24 hours (24Hr) of monitoring. BR was further divided into A1 (normal alpha), A2 (intermittent alpha), and A3 and A4 (no alpha with moderate and severe generalized slowing respectively). Primary outcomes were discharge GCS (GCSD) and disposition (from better to worse: home, inpatient rehab, nursing home, death). Generalized linear modeling was used to identify characteristics associated with the primary outcomes. Results: 43/112 patients had epileptiform activity including seizures detected by cEEG, however this finding did not predict outcomes. After controlling for age, GCSA, comorbidities and ICU stay, both BR (p=0.0168) and reactivity in the last 24Hr (p=0.0015) predicted GCSD. Mean GCSD was 3.7 points higher in patients with reactivity in the last 24Hr than those without. Mean GCSD was significantly different between patients with A1 and A4 (p=0.0059), A2 and A4 (p=0.0075), and A3 and A4 (p=0.0247), with scores declining from A1 to A4 (Table 1). Majority of patients with A1 or A3 in the first 24 hours retained the same BR in the last 24Hr (9/9 and 48/66, respectively). Patients with reactivity in the first 24Hr maintained it in the last 24Hr (85/87). Both reactivity (p=0.0134) and BR (p=0.0056) in the last 24Hr had significant impact on disposition after adjusting for age, GCSA, and comorbidities. Given any disposition, the odds of being in an equal or better disposition for patients with reactivity in the last 24Hr equal 3.6 (95% CI=1.305-9.977) times the odds for those without reactivity. For BR in the last 24Hr, given any disposition, the odds of being in an equal or better disposition for patients with A1 were 14.3 (p=0.0005; 95% CI 3.2-64.1) times the odds for those with A4, 4.8 (p=0.034; 95% CI 1.1-20.8) times than for A2, and 4.5 (p=0.025; 95% CI 1.2-16.4) times than for A3. For patients with A3, the odds were 3.2 (p=0.017; 95% CI 1.2-8.3) times than for A4 (Table 2). Conclusions: Absence of background rhythm / presence of severe generalized slowing (A4), and lack of reactivity in the last 24 hours of cEEG are significant predictors of lower GCSD scores and relatively worse disposition. EEG background characteristics (BR and reactivity in the last 24Hr) consistently predict GCSD. Reactivity and BR in the first 24Hr, and any focal slowing did not impact GCSD and disposition.