Optimizing seizure detection by quantitative EEG in paediatric refractory status epilepticus

ABSTRACT

@#Background & Objective: Continuous electroencephalography (cEEG) is valuable in the diagnosis and management of refractory status epilepticus (RSE) but requires intensive skilled interpretation. Density spectral array (DSA) is a quantitative analytic tool used to screen cEEG recordings for seizures. This study aims to determine the optimal amplitude setting and to compare the use of single-averagetrendgraph display and eight-trendgraph display in seizure detection with DSA. Methods: Five excerpts from pediatric cEEG recordings with RSE were identified. In Phase 1 of the study, each of 4 readers determined the DSA amplitude setting one most preferred for each excerpt, and marked all seizures in one excerpt using one’s preferred setting. Inter-rater agreement in seizure detection was measured. In Phase 2, readers marked all seizures in all excerpts, first using single-average-trendgraph display, and then using eight-trendgraph display after a wash-out period. Intra-rater agreement in seizure detection between the two display methods was calculated. Results: In Phase 1, DSA readers’ choice of preferred amplitude settings varied widely but inter-rater agreement in seizure detection was high. In Phase 2, seizure detection using single- and eight-trendgraph displays showed high agreement with each other and, where they disagreed, single-average-trendgraph was more sensitive. Additionally, low seizure-to-background amplitude ratio in EEG recordings was associated with worse detection sensitivity/specificity. Conclusions: DSA amplitude settings do not affect seizure detection. Single-trendgraph display is comparable to eight-trendgraph display in screening cEEG for seizures. Seizure detection with DSA performs better in cEEG recordings with high seizure-to-background amplitude ratio.