To determine test characteristics of Neuro ICU nurse interpretation of real-time bedside qEEG compared to cEEG

Non-convulsive status epilepticus (NCSE) and non-convulsive seizures (NCS) are common in the Neuro ICU. Patients who suffer from NCSE/NCS have been found to have increased morbidity and mortality. NCSE/NCS can only be diagnosed via gold standard continuous EEG (cEEG) and is interpreted intermittently, causing delayed seizure recognition.

Quantitative EEG (qEEG) distills cEEG data into its salient features, allowing more facile interpretation by non-neurophysiologists, such as ICU nurses. This is the first prospective study of real-time nurse qEEG interpretation in patients with unknown NCSE/NCS presence. If shown to be clinically useful, it could decrease the time to recognition of seizures in the Neuro ICU.

Neuro ICU nurses evaluate a qEEG display at the bedside on an hourly basis after undergoing a brief, standardized qEEG training session and report the number of seizures they see each hour. Post-hoc review of cEEG data by two neurophysiologists will be performed. The raw cEEG data will be scored for the number of seizures present per hour, seizure duration, and seizure spatial extent (focal, hemispheric or generalized).

Thus far, 82 patients with 539 1-hour blocks of nurse interpretations have been studied. Six patients experienced seizures while on study. Our preliminary data show a sensitivity of 100% (82%, 100%), specificity of 96% (94%, 97%), positive likelihood ratio 25 (16, 41), negative likelihood ratio 0 (0.0, 0.15), PPV 48% (37%, 58%), NPV 100%, and false alarm rate 0.04/hour.

Real-time nurse interpretation of qEEG is feasible in the Duke Neuro ICU along with other demands of patient care. Nurse interpretation yields a clinically useful specificity and low false alarm rate, lending credence to nurse qEEG interpretation as a screening tool. Further analyses are pending end of enrollment period in February 2020.