Intraoperative alpha-band power in frontal electrodes may provide helpful information about the balance of hypnosis and analgesia and has been associated with reduced occurrence of delirium in the post-anesthesia care unit (PACU). Recent studies suggest that narrow-band power computations from neural power spectra can benefit from separating periodic and aperiodic components of the EEG. This study investigates if such techniques are more useful in separating patients with and without delirium in the PACU at the group level as opposed to conventional power spectra. Intraoperative EEG recordings of 32 patients who developed perioperative neurocognitive disorders (PACU-D) and 137 patients who did not (noPACU-D) were considered in this post-hoc secondary analysis. We calculated power spectra using conventional methods and applied the "fitting oscillations & one over f" (FOOOF) algorithm to separate aperiodic and periodic components to see if the EEG signature is different between groups. At the group level, noPACU-D and PACU-D patients presented with significantly higher alpha-band power and a broadband increase in power, allowing a "fair" separation based on conventional power spectra. Within the first third of emergence, the difference in median absolute alpha-band power amounted to 8.53dB (AUC:0.74[0.65;0.82]), reaching its highest value. In relative terms, the best separation was achieved in the second third of emergence with a difference in medians of 7.71% (AUC:0.70[0.61;0.79]). AUC values were generally lower towards the end of emergence with increasing arousal. Increased alpha-band power during emergence in noPACU-D patients can be traced back to an increase in oscillatory alpha activity and an overall increase in aperiodic broadband power. Although differences between PACU-D and noPACU-D patients can be detected relying on traditional methods, the separation of the signal allows a more detailed analysis. Together this may enable clinicians to detect patients at risk for PACU-D early in the emergence phase.
Read full abstract