Seizure identification in the intensive care unit by non-neurophysiologists using quantitative electroencephalogram:...

Continuous electroencephalogram (cEEG) is a critical neuromonitoring tool for intensive care unit (ICU) patients and is the gold standard for detecting seizures, which are associated with poor neurological outcomes. However, because cEEG review is often intermittent, delays in identifying ictal events are common. Quantitative EEG (qEEG) provides a rapid screening alternative and demonstrates good sensitivity when interpreted by expert neurophysiologists. Enabling ICU caregivers to screen qEEG trends may facilitate earlier seizure recognition and more timely clinical interventions. We aimed to summarize the existing evidence and evaluate the diagnostic accuracy of qEEG for seizure detection when interpreted by non-neurophysiologists. We searched Embase, Medline, Scopus, and LILACS to identify studies estimating the accuracy of qEEG interpretation by non-expert reviewers in ICU patients. Risk of bias assessment was performed using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool. The systematic review included 12 studies; however, a meta-analysis could not be performed due to heterogeneity among studies in qEEG trends, seizure identification methods, and outcomes reported. Most studies had low risks of bias; however, there was a concern of bias in patient selection in some studies. Sensitivities ranged from 64 to 100%, specificities ranged from 0 to 95%, and there were variable false positive rates. Studies using multiple qEEG trends in combinations tended to have higher sensitivities. Additionally, performance improved when non-experts used a tailored approach, such as referencing a confirmed pattern of the patient's initial seizure. Across studies, qEEG trends whether individual or combined tended to have low specificities and PPVs. Nonetheless, qEEG remains a potentially valuable rapid screening tool that can prompt timely expert review of corresponding raw EEG and potentially expedite seizure detection and treatment in the ICU. In particular, providing non-experts with patient-specific templates of confirmed seizures could improve seizure detection using qEEG.