Abstracts

Rationale: The ability to remotely monitor critically ill patients undergoing continuous electroencephalography (EEG) has expanded access and improved patient care. However, timely interpretation is still challenging in hospitals where remote connectivity is not available or fails. Smartphones have the ability to provide rapid transmission of bedside EEG data to a remote expert reviewer, but the reliability of this modality for EEG interpretation has not been systematically analyzed. The objective of this study is to determine whether EEG interpretation using a video recording of EEG data recorded with a smartphone is as accurate as traditional review of EEG data via a remote connection to an EEG server, especially for detection of clinically relevant abnormalities.Methods: Thirteen de-identified intensive care unit EEG samples representing various patterns were selected by consensus of two board-certified clinical neurophysiologists. Three minute clips containing each EEG pattern were displayed using Natus NeuroWorks software (Natus Medical Incorporated, San Carlos, CA) and recorded with an iPhone 4 (Apple Incorporated, Cupertino, CA). Five blinded, fellowship-trained clinical neurophysiologists reviewed the iPhone video recordings of EEG display on a standardized 15-inch laptop/desktop screen. Reviewers were asked to identify the predominant pattern out of 15 possible choices and to classify samples as ictal versus non-ictal. After a period >1 month, reviewers interpreted the same EEG recordings in a randomized order using Natus NeuroWorks software.Results: Using either method, the total responses of all five reviewers resulted in the correct identification of an ictal versus non-ictal pattern in 61 out of 65 cases (94%). All four incorrect responses using iPhone videos involved mistakenly calling a “generalized seizure” pattern non-ictal generalized periodic discharges (Figure 1). Four incorrect responses using Natus NeuroWorks software involved calling “lateralized intermittent rhythmic delta” and “polymorphic focal slowing” patterns ictal, while two reviewers called a “generalized seizure” pattern non-ictal. Using either method, the cumulative responses of the reviewers resulted in the correct identification of individual EEG patterns in 50 out of 65 cases (77%). The most commonly missed patterns for both groups were “generalized seizure” and “ventilator artifact.” The accuracy of ictal pattern recognition (Figure 1) and individual EEG pattern recognition (Figure 2) was similar using iPhone video compared to NeuroWorks clips for all 13 patterns. Interrater agreement was assessed using κ statistics, which showed substantial agreement (0.61-0.80) between reviewers using iPhone videos versus NeuroWorks clips, 0.68 and 0.62 respectively.Conclusions: Smartphone technology represents a novel and reliable method of connecting any healthcare provider to timely and accurate expert review of EEG data when traditional remote server connections do not exist or are limited.