Reliability of Motor Evoked Potential Identification in Pediatric Populations

Motor evoked potentials (MEPs) elicited through transcranial magnetic stimulation (TMS) may provide valuable insights into neuromotor development in pediatric populations. However, the reliability of identifying MEPs using rater-based approaches, particularly in infants, remains poorly defined. This study addresses this gap by evaluating interrater reliability (IRR) and consensus rates for MEP identification in adolescents and infants.

Three experienced TMS researchers independently scored 5,738 electromyography (EMG) trials from 18 adolescents and 5,086 trials from 20 infants using a 5-point confidence scale. Scores reflected the raters' confidence in MEP presence, guided by predefined electrophysiological criteria. IRR, measured by intraclass correlation coefficients (ICC), and scoring consensus were compared across age groups, and signal characteristics were analyzed to identify predictors of disagreement.

IRR for adolescent MEP scores was excellent, ICC(C,3) = 0.973, 95% confidence interval (CI) [0.972, 0.974], while infant MEPs showed good IRR, ICC(C,3) = 0.784, 95% CI [0.773, 0.794]. Consensus rates were significantly lower for infant data (76.8%) compared to adolescent data (89.1%) (p < 0.001). Greater baseline EMG variability and lower post-stimulation peak-to-peak amplitudes were strongly associated with reduced consensus (p < 0.001 for both).

MEP identification using a rater-based scoring protocol is feasible in pediatric cohorts but presents greater challenges in infants due to distinct neurophysiological and behavioral factors. Despite these difficulties, good reliability can still be achieved when multiple trained raters independently score EMG trials. These findings support the need for consensus-driven best practices to guide MEP identification in early developmental research.