Interobserver agreement in neonatal seizure identification
Related Experiment Video
Updated: Jun 22, 2026
Continuous Video Electroencephalogram during Hypoxia-Ischemia in Neonatal Mice
Published on: June 11, 2020
Aileen Malone1, C Anthony Ryan, Anthony Fitzgerald
1Department of Paediatrics and Child Health, University College Cork, Ireland.
Objectives:
Accurate diagnosis of neonatal seizures is critically important and is often made clinically, without EEG (electroencephalography) monitoring. This observational study aimed to determine the accuracy and interobserver reliability of healthcare professionals in distinguishing clinically manifested seizures from other neonatal movements, when presented with clinical histories and digital video recordings only.
Methods:
Twenty digital video recordings of paroxysmal movements in term and preterm infants were selected from a video-EEG database. The movements were categorized as seizure and nonseizure using EEG. Health care professionals (n = 137) from eight neonatal intensive care units (NICUs) were shown the video recordings with additional relevant clinical data, excluding EEG findings. The observers were asked to indicate which movements they considered to be seizure or nonseizure. A multirater Kappa statistic was used to assess agreement between observers and with the true diagnosis.
Results:
Twenty video clips (11 seizure, 9 nonseizure) were evaluated by 91 doctors and 46 other professionals. The average number of correctly identified events was 10/20. Clonic seizures were correctly identified most frequently (range 36.5-95.6% of observers). Subtle seizures were poorly identified (range 20.4-49.6% of observers). The interobserver agreement (Kappa) for doctors and other health care professionals was poor at 0.21 and 0.29, respectively. Agreement with the correct diagnosis was also poor at 0.09 for doctors and -0.02 for other healthcare professionals.
Discussion:
It is often impossible to accurately differentiate between seizure-related and nonseizure movements in infants using clinical evaluation alone. In addition, doctors do not have a higher capacity for discriminating between neonatal paroxysmal events than other health care professionals. Until reliable continuous neurologic monitoring of newborn babies is available, it is likely that some babies with seizures will remain undetected and others with nonseizure movements will continue to be treated with potentially harmful anticonvulsants.
