A systematic review of electroencephalography in acute cerebral hypoxia: clinical and diving implications

Introduction: Hypoxia can cause central nervous system dysfunction and injury. Hypoxia is a particular risk during rebreather diving. Given its subtle symptom profile and its catastrophic consequences there is a need for reliable hypoxia monitoring. Electroencephalography (EEG) is being investigated as a real time monitor for multiple diving problems related to inspired gas, including hypoxia. Methods: A systematic literature search identified articles investigating the relationship between EEG changes and acute cerebral hypoxia in healthy adults. Quality of clinical evidence was assessed using the Newcastle-Ottawa scale. Results: Eighty-one studies were included for analysis. Only one study investigated divers. Twelve studies described quantitative EEG spectral power differences. Moderate hypoxia tended to result in increased alpha activity. With severe hypoxia, alpha activity decreased whilst delta and theta activities increased. However, since studies that utilised cognitive testing during the hypoxic exposure more frequently reported opposite results it appears cognitive processing might mask hypoxic EEG changes. Other analysis techniques (evoked potentials and electrical equivalents of dipole signals), demonstrated sustained regulation of autonomic responses despite worsening hypoxia. Other studies utilised quantitative EEG analysis techniques, (Bispectral index [BIS (TM)], approximate entropy and Lempel-Ziv complexity). No change was reported in BIS (TM) value, whilst an increase in approximate entropy and Lempel-Ziv complexity occurred with worsening hypoxia. Conclusions: Electroencephalographic frequency patterns change in response to acute cerebral hypoxia. There is paucity of literature on the relationship between quantitative EEG analysis techniques and cerebral hypoxia. Because of the conflicting results in EEG power frequency analysis, future research needs to quantitatively define a hypoxia-EEG response curve, and how it is altered by concurrent cognitive task loading.