Intraoperative electroencephalogram-derived measures and their association with postoperative delirium: a systematic review and meta-analysis.

Abstract

Postoperative delirium (POD) is a common and serious complication in older surgical populations. Intraoperative EEG may help identify those at risk. Proprietary processed EEG (pEEG) devices were not developed to identify POD risk and studies utilising such devices may confound attempts to identify non-proprietary EEG-derived measures associated with POD. We performed a systematic review of non-proprietary intraoperative EEG-derived measures and their association with POD including a secondary analysis excluding studies that utilised pEEG-guided anaesthesia. This review followed PRISMA guidelines and the Population, Index test, Target outcome (PIT) framework. The protocol was prospectively registered with PROSPERO (ID: CRD42024564983). Seven databases were searched up to May 30, 2025. Eligible studies reported intraoperative EEG-derived measures and POD incidence. Meta-analysis was performed for measures reported in three or more studies. Risk of bias was assessed using the Quality in Prognostic Studies tool. Forty-two studies (13115 patients) were included, and the mean incidence of POD was 18.8%. Forty-nine EEG-derived measures were identified including time domain, frequency domain, and functional connectivity measures. Meta-analyses identified that lower intraoperative alpha and delta mean power, lower alpha peak frequency, and lower spectral edge frequency were significantly associated with POD. Delta mean power was not significantly associated with POD in non-pEEG-guided studies. Both the presence and duration of burst suppression were associated with POD. Alpha oscillation and burst suppression measures appear to be robust indicators of POD vulnerability, even in the presence of pEEG-guided anaesthesia. The use of pEEG-guided anaesthesia did not eliminate the associations between these measures and POD. PROSPERO (CRD42024564983).