Association of Altered Anteriorization of Electroencephalographic Alpha Oscillation During Transition to Unconsciousness With Postoperative Delirium: A Prospective Observational Study.

BACKGROUND

During sevoflurane-induced anesthesia, electroencephalography analysis demonstrates alpha oscillations shifting from the occipital to the frontal regions, with further characteristic changes during maintenance and emergence. We aimed to exploratorily investigate whether alterations in alpha anteriorization and functional network efficiency are associated with postoperative delirium.

METHODS

This prospective, exploratory observational study enrolled 100 patients from a single tertiary university hospital. Patients were classified into nondelirium (n = 81) and delirium (n = 19) groups based on delirium assessments. After excluding 17 patients due to artifacts on electroencephalography, 19-channel electroencephalography data were collected from 83 elderly patients under sevoflurane-induced anesthesia. Artifact-free epochs were extracted during the induction, maintenance, emergence, and postanesthesia care unit phases. Spectral, topographical, and network analyses were conducted. Postoperative delirium was assessed using confusion assessment method for intensive care unit and a chart review.

RESULTS

Significant group differences emerged just before the loss of responsiveness, with the nondelirium group exhibiting robust alpha power amplification across all brain regions (median [interquartile range {IQR}] in dB; frontal: 4.24 [1.86-6.57] vs -0.88 [-1.59-3.38], P = .007; parietal: 3.71 [0.51-5.55] vs -0.95 [-2.33-1.11], P = .046; occipital: 2.81 [0.49-5.71] vs -0.04 [-2.06-1.27], P = .048), whereas the delirium group showed minimal changes. During the maintenance phase, spectral analysis revealed higher alpha power in the nondelirium group across all 19 channels (frontal: 5.42 [2.90-8.16] vs 2.89 [-0.03-5.09], P = .007; 18 of 19 channels significant by false-discovery-rate (FDR)-corrected permutation test). Network analysis demonstrated longer averaged characteristic path lengths in the delirium group (mean ± standard deviation [SD], 4.49 ± 0.62 vs 3.57 ± 0.16, P = .026, FDR-corrected), consistently in the frontal, parietal, and occipital regions, suggesting reduced network efficiency.

CONCLUSIONS

Delirium was associated with early disruption of alpha anteriorization during induction and globally reduced functional network efficiency during maintenance. Perioperative electroencephalogram (EEG) transition features may serve as early markers of delirium susceptibility.