摘要BACKGROUND:Although the Confusion Assessment Methods for the Intensive Care Unit(CAM-ICU)is a recommended tool for diagnosing sepsis-associated encephalopathy(SAE),it has several limitations.Mismatch-negativity(MMN)and P3a are components of event-related potentials(ERPs)used with electroencephalography(EEG)and are associated with cerebral function changes in critically ill patients.This study aimed to provide a quantitative,non-invasive method to guide SAE diagnosis in non-sedated patients.METHODS:From January 2022 to March 2023,sepsis patients without sedation were enrolled and assessed via the CAM-ICU,Glasgow Coma Scale(GCS),and ERP under standard procedures.Both MMN and P3a data were collected.The diagnostic value of MMN and P3a was assessed with processed ERP data.RESULTS:Thirty-six patients were included in this study,comprising 19 patients with SAE and 17 patients without SAE(NSAE).MMN and P3a amplitudes decreased,and only FzMMN amplitude significantly decreased in SAE patients(2.03[1.08,2.93]mV vs.3.21[1.92,4.34]mV,P=0.040).After median dichotomization,low F3P3a and FzP3a amplitudes were associated with higher CAM-ICU positivity rates and APACHE II scores.Both amplitude in F3P3a(AUC=0.710,95%CI:0.527-0.893,P=0.034)and FzP3a(AUC=0.700,95%CI:0.519-0.881,P=0.041)exhibited moderate diagnostic efficacy for SAE,while FzMMN amplitude lacks effective diagnostic value.CONCLUSION:In this pilot study,ERP components F3P3a and FzP3a amplitudes demonstrated moderate diagnostic value for SAE.These exploratory findings require confirmation in larger and powered cohorts.