Rationale:
Electroencephalography (EEG) is a critical tool for diagnosing epilepsy and detecting acute symptomatic seizures in critically ill patients. However, its overuse in low-yield clinical scenarios can lead to unnecessary resource utilization and increased patient burden. This project evaluates EEG ordering patterns in the cardiovascular intensive care unit (CVICU), where high-risk patients – such as those on extracorporeal membrane oxygenation (ECMO) or post-cardiac surgery – require complex care. By analyzing utilization trends, we aim to develop strategies that optimize EEG use while maintaining high-quality patient care.
Methods:
We conducted a retrospective analysis of EEGs performed at our institution in 2023 and 2024, examining seizure detection rates, clinical management changes, time to seizure onset, and EEG duration. Primary indications for ordering EEGs were categorized into three groups: suspected clinical seizures, protocol-led monitoring based on American Clinical Neurophysiology Society consensus guidelines (patients on ECMO, post-cardiac arrest, or with intracranial pathology), and unexplained hemodynamic or metabolic abnormalities.
Results:
A total of 112 long-term EEGs were analyzed. The patient population ranged from 1 day to 17 years, with a mean age of 2.43 years. Seizure detection rates were highest when EEGs were ordered for clinical seizure concern (33.3%) or per protocol (34.8%), both of which led to more frequent changes in management (40.5% and 37.0%, respectively). In contrast, EEGs ordered for unexplained hemodynamic or metabolic abnormalities had a much lower seizure detection rate (6.9%) and rarely influenced clinical management (6.9%), suggesting limited utility for these indications.
Analysis of EEG duration and time to seizure onset showed that seizures were typically detected early. Among patients with seizures, the majority (73%) were detected within the first five hours, and nearly all (97%) occurred within 24 hours, supporting a 24-hour monitoring standard in most cases. The median time to seizure onset was 2.14 hours, with an average of 5.28 hours. The overall average EEG duration was 36.22 hours. Patients with seizures had longer EEG durations, averaging 63.40 hours, reflecting the need for continued monitoring. In contrast, patients without seizure activity had an average EEG duration of 23.97 hours, aligning with the goal of obtaining at least 24 hours of data.
Conclusions:
EEGs ordered for clinical seizure suspicion and protocol-driven indications demonstrated high diagnostic yield and frequently led to management changes, supporting current ordering practices. In contrast, EEGs obtained for hemodynamic or metabolic abnormalities had limited utility, suggesting a need for more selective use in these scenarios. The early detection of seizures within the initial hours further supports the appropriateness of a 24-hour monitoring standard in most cases. These findings are significant because they identify actionable targets to refine EEG utilization in ICU settings, reducing unnecessary monitoring while preserving diagnostic effectiveness.Funding: None.