Seizure Assessment and Forecasting With Efficient Rapid-EEG: A Retrospective Multicenter Comparative Effectiveness...

Area of Science:

• Neuroscience
• Critical Care Medicine
• Medical Technology

Background:

• Critically ill patients frequently experience seizures, often without clear clinical signs, necessitating electroencephalography (EEG) for accurate diagnosis.
• Conventional EEG (cEEG) is resource-intensive and cannot meet the growing demand for continuous seizure detection.
• Brief EEG screening methods are needed to triage cEEG resources effectively.

Purpose of the Study:

• To assess the noninferiority of the 2HELPS2B seizure risk prediction score calculated on a 1-hour rapid response EEG (rrEEG) compared to cEEG.
• To evaluate rrEEG as a potential solution for seizure risk stratification in resource-constrained critical care settings.

Main Methods:

• A multicenter retrospective diagnostic accuracy study involving 240 patients (≥18 years old) from January 2018 to June 2022.
• The 2HELPS2B score was calculated using 1-hour rrEEG recordings and compared to cEEG.
• Noninferiority was assessed using the area under the receiver operator characteristic curve (AUC) with a predefined noninferiority margin.

Main Results:

• The 2HELPS2B score on 1-hour rrEEG demonstrated noninferiority to cEEG for seizure prediction (AUC 0.85).
• No significant differences were observed in AUC between rrEEG and matched cEEG (AUC 0.89).
• False negative rates and survival analyses showed comparable results between rrEEG and cEEG.

Conclusions:

• A 1-hour rrEEG using the 2HELPS2B algorithm is a noninferior alternative to cEEG for seizure risk prediction.
• Patients with a low-risk score (2HELPS2B = 0) may potentially avoid prolonged cEEG monitoring.
• This approach could enable more efficient allocation of cEEG resources to higher-risk patients.