Clinical sedation assessment versus EEG-based monitoring during deep sedation in VV-ECMO patients: A prospective blinded observational study

BACKGROUND: Severe ARDS patients on VV-ECMO often require deep sedation, where behavioural assessments may fail to quantify cortical suppression. We evaluated this suppression burden and its relation to routine sedation assessment using blinded EEG. METHODS: In this prospective observational study, adult VV-ECMO patients with severe ARDS underwent 24-h blinded Conox EEG monitoring. Burst suppression ratio (BSR) and Quantium Consciousness Index (qCON) were paired with hourly Richmond Agitation-Sedation Scale (RASS) assessments (target -4). Linear mixed-effects models with patient as a random effect tested whether hourly RASS predicted BSR and qCON. RESULTS: Twenty patients yielded 467 paired RASS-EEG observations. Burst suppression occurred in all patients. Median time-weighted average BSR was 7.3% (IQR 2.1-30.8), maximum BSR was 27.9% (IQR 6.2-71.2), and median time with BSR ≥ 5% was 37.7% (IQR 12.5-96.9). A 1-point RASS increase was associated with a 3.31% decrease in BSR (95% CI - 5.60 to -1.01; p = 0.005). At RASS -4, BSR was ≥5% in 64.5% and ≥10% in 51.7% of observations. Lower qCON values were associated with greater burst suppression burden, with BSR approaching zero above qCON ≈40. CONCLUSIONS: Processed EEG revealed a substantial burden of burst suppression during deep VV-ECMO sedation despite uniform behavioural targets. Within this deep sedation range, RASS had limited resolution for quantifying cortical suppression.