Abstracts

Rationale: Inpatient video-EEG monitoring is the gold standard for characterizing clinical episodes concerning for seizures. However, the utility of this test is often compromised by poor capture of the event of interest. To improve the yield of video-EEG studies, we need to define process and outcome quality metrics with which we can measure baseline performance and identify targets for intervention. In this study, we propose potential quality indicators for pushbutton events and describe how frequently such measures were met. Methods: This was a retrospective cohort study of all pushbutton events on long-term video-EEG monitoring at the University of Michigan during October 2018. We assembled multidisciplinary stakeholders and developed a list of metrics relevant to electro-clinical event correlation. Process metrics included audio (i.e., whether the reason for push was stated) and video (i.e., whether the patient was visible on camera) parameters. The outcome metric was the presence of behaviors that impaired the electroencephalographer’s understanding of the clinical semiology necessary for ictal determination and/or localization. Each pushbutton event was reviewed starting 30 seconds prior through 60 seconds after nursing first response. Results: Long-term video-EEG records for 95 patients were archived during the month, of which 30 patients had at least one pushbutton event for a clinical episode concerning for seizure. A total 84 unique pushbutton events were analyzed, of which 15 (18%) corresponded to an electrographic seizure, 16 (19%) were psychogenic nonepileptic spells, and the remaining events were scalp-negative non-psychogenic clinical events. There were 49 events (58%) that occurred on the general neurology floor and 50 events (60%) occurred during daytime business hours. The most common button pushers were family members (N=30, 36%), patients (N=17, 20%), and nurses (N=8, 10%). Regarding process metrics, the reason for push was not stated for the camera in 68 events (81%) and people were blocking the camera during some portion of the event in 37 events (44%). Regarding the outcome metric, 29 (35%) had at least one barrier to interpretation - these included most commonly incomplete narration (15), people blocking the camera (12), objects blocking the camera (4), and camera on the wrong side of the patient obscuring the relevant body part (3). Conclusions: Barriers to interpretation were common. The most frequent occurrences that impacted diagnosis included incomplete narration and individuals blocking the patient from camera view. Additionally, we found that family members are frequent observers and therefore interventions to improve video EEG quality should engage family. The significance of this work is that it identifies modifiable targets for future interventions aimed at improving quality of inpatient video-EEG monitoring. Funding: No funding