Unveiling the Immediate Impact of Prechtl’s General Movement Assessment Training on Inter-Rater Reliability and Cerebral Palsy Prediction

Abstract

Prechtl's General Movement Assessment (GMA) is a qualitative video assessment that is an early predictor of cerebral palsy (CP) in infants. GMA raters undergo two levels of training: basic and advanced. Using a retrospective study design, we evaluated the impact of the GMA training level on three key measures: inter-rater reliability (IRR), predictability for a later CP diagnosis, and generalizability for both medically trained and non-medically trained raters. As part of a larger study on former level-IV neonatal intensive care unit (NICU) infants, participants had a GMA video at 3-4 months of corrected age with parental consent. Four GMA raters with basic training interpreted the videos. Subsequently, two raters underwent advanced training and reinterpreted the videos. We evaluated IRR with Gwet's AC1 coefficient, CP prediction with logistic regression, and generalizability with Gwet's Linearization method and McNemar's test. Following basic GMA training, all four raters had moderate to almost perfect agreement. The CP predictability of the raters' individual GMA scores ranged from 73% to 81%. Advanced GMA training significantly improved IRR and enhanced CP predictability. GMA rater proficiency was similar regardless of medical training. Advanced GMA training improved readers' capabilities to correctly identify abnormal movement patterns and aid in early CP diagnosis.