Abstracts

Rationale: Conventional electroencephalography (EEG) using a full set of 18+ electrodes is the gold standard for detecting seizures, particularly nonconvulsive seizures. Rapid detection and treatment of nonconvulsive seizures – which are increasingly recognized as a cause of altered mental status that are associated with increased morbidity, mortality, and risk of neuronal injury – have critical implications for patient outcome. EEG with a reduced number of electrodes has the potential to expedite management of possible seizures. While several studies have demonstrated its utility, reduced EEG is not widely used. To our knowledge, there has been no systematic review of studies on the diagnostic utility of reduced EEG. Methods: We performed a systematic review of PubMed, EMBASE, Web of Science, Cochrane, and AES Annual Meeting databases to identify studies (published and unpublished) that reported the sensitivity of reduced EEG compared to full EEG for detection of seizures in adult or pediatric/neonatal patients based on visual inspection. We excluded studies that solely utilized automated detection algorithms, did not describe the reduced EEG array or reference standard, or presented raters with fewer than 10 EEG samples. For each included study, we extracted information on: sensitivity for seizures or status epilepticus (reported as median [IQR], range), patient population (adult/pediatric), study design (retrospective/prospective), reduced EEG description (electrode number, spatial coverage), financial disclosures, and study conclusion (favorable to reduced EEG versus unfavorable). Results: We reviewed 5244 records and identified 22 studies (20 journal articles, 2 abstracts) that met our criteria. These studies describe a total of 29 reduced EEG montages comprised of 2-14 electrodes offering coverage of temporal (n=16), frontal (n=17), central (n=8), parietal (n=9), and occipital (n=11) regions. The median sensitivity of reduced EEG for ictal activity was 85% [IQR 70-93%], range: 50-100%. The majority of these studies were done with adult patients (n=16) using a retrospective study design (n=17). Although the majority of studies presented raters with both the reduced and full EEG (n=13), only 3 studies actually reported an estimate of the sensitivity of full EEG for ictal activity (88% [IQR 86.5-93%], range: 85-98%). Financial conflict of interest was disclosed in 5 studies that reported a lower median sensitivity (68% [IQR 54-77%], range: 50-93%) compared to those that did not (86% [IQR 76-93%], range: 64-100%). Unexpectedly, studies that recommended further investigation or use of reduced EEG (n=17) did not report markedly different median sensitivity for seizure activity (89% [IQR 77-93%], range: 50-100%) compared to studies that did not endorse further use (70% [IQR 68-70%], range: 64-81%). Conclusions: Our findings question the seemingly axiomatic notion that reduced EEG results in suboptimal diagnostic sensitivity for seizure activity and highlight the need for more prospective studies that can address the lingering question of the impact these tools may have on clinical management. Funding: Stanford University Medical Scholars Research Program