Objective: Although several studies have shown the potential of amplitude integrated electroencephalography (aEEG) in detecting neonatal seizures, no publications have evaluated the diagnostic use of aEEG for the detection of seizures in adult patients.
Methods: In this prospective blinded observational study, bifrontal single-channel electroencephalography (EEG) recordings were performed with a portable EEG monitor (CSM M3 ICU, Danmeter-Goalwick Holdings Limited, Odense, Denmark) during the out-of-hospital care of emergency cases. Four intensive care unit (ICU) physicians received training in the interpretation of aEEG recordings. After the training they evaluated the stored aEEG traces for the presence of epileptic seizure activity during the recording time. The physicians were blinded to the clinical data of the patients. The results obtained were compared with the clinical diagnosis and the evaluation of the raw EEG signal. The level of interrater agreement was quantified using Fleiss' ĸ.
Results: The aEEG traces from 10 patients with generalized epileptic seizures and 46 patients without seizures were analysed. Overall, the nonexpert ICU physicians failed to identify recordings obtained from patients with seizures reliably, when compared with clinical diagnosis and the single-channel EEG results (mean sensitivity 40%, range 40-60%; mean specificity 89%, range 87-93%). Agreement between observers was high for the cases with seizures ( ĸ = 0.80 ± 0.13). Patients who suffered status epilepticus during the recordings were difficult to identify by most raters.
Conclusion: Recording of aEEG without access to the raw EEG data is not a reliable diagnostic tool for the identification of epileptic seizures in the hands of nonexpert ICU physicians.