Peak alpha frequency and spectral exponent in postoperative delirium: a high-density EEG cohort study

Present concepts around the neurophysiology of delirium suggest impairments in excitatory/inhibitory balance, leading to EEG slowing, altered synaptic activity and integration of information. We aim to clarify the mechanisms involved, hypothesising that, in postoperative delirium, the EEG peak alpha frequency (PAF; decreased peak frequency indicating slowing in alpha frequency) would slow and the spectral exponent (SE; a higher slope indicating an increased neural inhibitory tone) would be steeper compared with a baseline collected before surgery. The primary aim was to determine the association of PAF and SE with postoperative delirium. The secondary aim was to determine the association of perioperative changes in inflammation and executive function with PAF and SE.

High-density 256-channel EEG data were collected perioperatively (i.e. obtained pre-, post-, after 1-yr post-surgery) in a cohort study of 202 patients aged >65 yr undergoing major surgery conducted from 2015 to 2025 at a major quaternary care hospital. Thirty-three participants experienced postoperative delirium, 169 participants did not, with 113 participants followed up at the 1-yr interval. Linear mixed-effects modelling was conducted for PAF and SE estimates. Further analyses tested associations of PAF and SE with the Delirium Rating Scale, Trail Making Test-B (TMT-B) and plasma cytokine levels.

After surgery, mean PAF was significantly slower and mean SE was significantly steeper compared with preoperative and 1-yr postoperative values (P<0.0001). These changes were larger in the delirium compared with the non-delirium group (P<0.05 to P<0.001). After surgery, PAF and SE estimates were significantly associated with delirium severity scores. Decreases in PAF or SE estimates (pre to post) were significantly associated with increases in TMT-B time and cytokine levels (reflecting worsening cognitive performance and increases in inflammation levels, respectively).

Lower mean PAF and steeper SE are consistent with slower neuronal repolarisation and increased inhibition in delirium. PAF and SE estimates provide new information about delirium pathophysiology and could be further explored as noninvasive EEG biomarkers of detection and monitoring of delirium state.