A randomized controlled educational pilot trial of interictal epileptiform discharge identification for neurology residents

Abstract

This study evaluated a technology-enhanced, retrieval-based program for teaching neurology residents to identify interictal epileptiform discharges (IEDs). Junior residents were randomized to rate 500 candidate IEDs with instant feedback on either a web browser (intervention 1) or an iOS app (intervention 2), or to an inactive control group. All participants completed pre- and post-training assessments.

Twenty-one residents completed the study (control n=8; intervention 1 n=6; intervention 2 n=7), most without prior EEG experience. Intervention 1 produced significant improvements in AUC (.74→.85), sensitivity (.53→.75), and confidence (1.33→2.33; all p<.05). Intervention 2 improved AUC (.81→.86) and confidence in identifying IEDs and spike–wave discharges (2.00→3.14; p<.05). Controls showed no meaningful change.

This program significantly improved IED identification, with web-based training producing the greatest objective gains. It may serve as an effective supplement to resident EEG education.

Background

Neurology residents are expected to competently interpret EEGs by graduation, yet evidence shows that many—including senior trainees—do not achieve EEG proficiency, especially in correctly identifying interictal epileptiform discharges (IEDs). This skill is critical for accurate diagnosis and management of epilepsy and is a core expectation of the ACGME neurology milestones.

To address this gap, we developed an educational program focused on teaching IED identification. The program combined foundational instruction through three online video tutorials with practical training in rating candidate IEDs, both without and with instant feedback. Its design was grounded in technology-enhanced learning and retrieval practice, integrating these learning theories into a single framework.

We evaluated whether this program improved residents’ IED identification skills compared with an inactive control group.

Methods

We conducted a prospective randomized controlled trial evaluating a new EEG teaching program for identifying interictal epileptiform discharges (IEDs). Junior neurology residents at MGB and Yale voluntarily completed a pre-survey and a 500-item pre-spike test in which they classified candidate IEDs without feedback. Test items were drawn from 13,262 expert-labeled waveforms, evenly sampled across ten bins reflecting the level of expert agreement.

Residents were then randomized to an inactive control group or to one of two intervention groups. Both intervention groups viewed three online tutorials and completed a 500-item spike test with instant feedback; intervention 1 used a web browser with adjustable montages, and intervention 2 used an iOS app displaying static bipolar images. Feedback differed only in visual presentation.

Thirty days later, participants completed a post-survey and a 500-item post-spike test identical in structure to the pre-assessments. Performance metrics included accuracy, sensitivity, false-positive rate, calibration, AUC, noise, and threshold/bias.

Within-group changes were assessed with Wilcoxon tests; between-group differences with Kruskal–Wallis and Dunn tests (p<.05, Bonferroni-corrected). 

Data Description

For both the pre- and post-training assessments, each participant completed a 500-item spike test composed of ten-second EEG segments containing a marked 0.5-second candidate IED window. Items were randomly and evenly sampled from all ten bins to ensure balanced representation of waveform difficulty and ambiguity. Participants also completed a post-intervention version of the same 500-item test. In the intervention groups, an additional 500-item spike test with instant feedback was used as part of the training. Educational content included three prerecorded online video tutorials covering navigation of the spike test, the six IFCN criteria for IED identification, and application of these criteria to example waveforms with differing levels of expert consensus. Participants were junior neurology residents from Massachusetts General Brigham and Yale.

Usage Notes

Data and code to reproduce all results and figures are available
through a public data sharing repository at: https://github.com/bdsp-core/spike-test-pilot-trial.git.

Ethics

This study of human subjects was approved by the Mass General Brigham Institutional Review Board (IRB approval # 2012P001929), including review of EEG and other clinical data. The Partners Healthcare Human Research Committee provided a waiver of written consent for this study. All data is deidentified.

Acknowledgements

This work was supported by grants from the NIH (R01NS102190, R01NS102574, R01NS107291, RF1AG064312, RF1NS120947, R01AG073410, R01HL161253, R01NS126282, R01AG073598) and the NSF (2014431).

Conflicts of Interest

F. Nascimento is an Associate Editor for Epileptic Disorders. Centaur Labs developed and holds financial interest in the DiagnosUs app. C. Traner is a paid consultant for Ceribell, Inc. R. Strowd serves as a consultant for Monteris Medical and Novocure, receives a stipend from the American Academy of Neurology, and has received research support from the American Academy of Neurology, the American Society for Clinical Oncology, Jazz Pharmaceuticals, and the International Association for Medical Science Educators. He also receives royalties from Elsevier, Lecturio, and Kaplan Medical. J. Moeller receives royalties from Wolters Kluwer (UpToDate). M. B. Westover is a co-founder and equity holder in Beacon Biosignals and receives royalties from Pocket Neurology (Wolters Kluwer) and Atlas of Intensive Care Quantitative EEG (Demos Medical). F. Nascimento, J. Jing, W. Y. Kong, and M. Olandoski report no disclosures relevant to this manuscript.