The effect of neuromuscular blockade on EEG-based measures of awareness

Background: Both basic and clinical consciousness research aims to find objective measures that re- liably distinguish conscious from unconscious brain states. Electroencephalogram (EEG) measures are widely used, although they may be contaminated by electrical signals from muscles. Methods: To assess this source of error, we investigated the impact of neuromuscular blockade (NMB) on proposed measures of consciousness (spectral slope, Lempel-Ziv complexity (LZc), connectivity, alpha peak frequency, power in canonical EEG frequency bands) computed from spontaneous high-density EEG recorded from six healthy volunteers in three different conditions: (1) awake-unparalysed, (2) awake-paralysed caused by neuromuscular blocking agent (NMBA), and (3) sedated-paralysed (sedated with propofol, paralysed by NMBA, (un)consciousness non-confirmable). Results: The markers we investigated distinguished awake-unparalysed states from sedated-paralysed with close to perfect accuracy. Our analysis revealed a serious failure of all measures, except alpha power, to recognise awake-paralysed, without sedation, as an aware state. Errors ranged from 19% of awake-paralysed time segments predicted as unaware (using spectral slope) to 100% (using LZc). Using alpha power, only 1% of all awake-paralysed segments were misclassified. Critically, the awake-paralysed is the state that is important to detect in sedated-paralysed patients, to prevent the experience of accidental awareness during general anaesthesia (AAGA). Conclusions: This study clearly demonstrates that many EEG-based measures fail to recognise awareness in awake-paralysed subjects, by using a unique high-density EEG data set. Alpha power was determined to be the most robust measure to detect AAGA, but this may not generalise to all types of general anaesthetic agents.