Electroencephalogram Monitoring in Critical Care: Multicenter Analysis of Timing, Duration, and Readmissions.

importanceGaining insights into acute and critical care electroencephalogram (EEG) practice patterns can enable more efficient resource utilization without compromising care quality.

objectivesWe aimed to identify factors associated with the timing of EEG initiation during acute care hospitalizations, the duration of EEG monitoring, and hospital readmissions with EEG monitoring. DESIGN, SETTING, AND

participantsThis is a retrospective cohort study of inpatient admissions to three academic medical centers between 2009 and 2024. Patients were included if older than 18 years and underwent EEG monitoring (routine or long-term) during the hospitalization. Demographic and clinical variables were extracted, along with admission information, primary diagnosis defined by the International Classification of Diseases (ICD), 9th Edition and 10th Edition codes, drug administration, and characteristics of EEG use: timing of EEG initiation during hospitalization, duration of monitoring, presence of seizures, and rhythmic and periodic patterns (RPPs). Descriptive statistics and regression analysis were performed. MAIN OUTCOMES AND MEASURES: Our outcome measures were: 1) time to EEG monitoring relative to the day of admission (hr), 2) duration of monitoring (hr), and 3) readmissions with EEG monitoring within 12 months of the first admission.

resultsA total of 34,773 patients met the inclusion criteria. The most frequent primary neurologic admission diagnosis based on ICD codes were seizures/status epilepticus (n = 3219, 9.26%), traumatic brain injury (n = 1825, 5.25%), and ischemic stroke (n = 1787, 5.14%). The most frequent nonneurologic primary diagnostic category was toxic-metabolic disease and altered mental status (n = 6798, 19.55%). Key covariates associated with earlier EEG monitoring during the index admission were primary diagnostic categories of cardiac arrest/anoxic brain injury and seizures/status epilepticus. A diagnosis of aneurysmal subarachnoid hemorrhage, electrographic seizures, and lateralized periodic discharges were associated with longer durations of monitoring. Patients with a diagnosis of sepsis had later and shorter duration of monitoring. Factors associated with readmissions with EEG monitoring included a primary index admission diagnosis of seizures/status epilepticus and brain tumors. Presence of electrographic seizures, RPPs was associated with longer monitoring duration (29.03 hr; interquartile range [IQR], 3.43-73.80 hr). However, even among patients without seizures or RPPs, the median duration of monitoring was 16.78 hours (IQR, 0.82-37.34 hr). CONCLUSIONS AND RELEVANCE: Characterizing EEG utilization patterns in critically ill patients allows identification of potential areas for practice optimization. Patients admitted with primary nonneurologic diagnoses underwent EEG monitoring later and for shorter durations compared with those admitted with primary neurologic conditions. These findings suggest opportunities to refine EEG triage and resource allocation, including earlier initiation of monitoring in nonneurologic patients at elevated seizure risk (e.g., sepsis) and timely discontinuation of EEG in patients without seizures or RPPs.