Abstract
BACKGROUND: Guidelines currently suggest considering EEG guidance during general anesthesia in elderly patients to avoid prolonged burst suppression (BS), with the aim of mitigating postoperative delirium (POD). Our study aimed to investigate the association between POD and intraoperative BS duration dependent on the general anesthetic agent used (propofol vs. sevoflurane). METHODS: In this prospective study (2019-2022), EEGs from 265 patients over 70 years undergoing general anesthesia were analyzed for intraoperative BS duration both visually and using one new automated algorithm to evaluate its accuracy. Associations between BS duration, anesthetic agent, and postoperative delirium (POD) were evaluated using multivariable logistic regression, adjusting for confounders. RESULTS: BS duration was markedly shorter than in prior cohorts but did not reduce overall postoperative delirium (POD) incidence. POD occurred more frequently with sevoflurane than propofol (44% vs. 30%, p = 0.017), despite shorter median BS [0 s (IQR 0-4.9) vs. 20.6 s (IQR 0-151.7); p = 0.012]. A significant interaction between anesthetic agent and BS (p = 0.033) showed that BS under sevoflurane conferred 3.8-fold greater POD risk than under propofol. Sevoflurane plus BS increased POD odds 9.3-fold compared to propofol without BS. Our new automated BS detection algorithm demonstrated high precision (median error <2.17 s). CONCLUSION: Sevoflurane markedly increased POD risk versus propofol, independent of BS duration. Sevoflurane and BS interaction amplified delirium odds. BS appears a vulnerability marker rather than a causal factor. The validated machine-learning BS detector offers a reliable tool for future EEG-based delirium risk research.