664 - General Movements Assessment for cerebral palsy in high-risk preterms: Accuracy following implementation of a new program

Publication Number: 2647.664

Melina Papadimitriou, McGill University Faculty of Medicine and Health Sciences, Beaconsfield, PQ, Canada; Christine Labelle, CUSM, Montreal, PQ, Canada; Stephanie Arpin, McGill University Faculty of Medicine and Health Sciences, Montreal, PQ, Canada; Jasmine Canse, McGill University Faculty of Medicine and Health Sciences, Montreal, PQ, Canada; Andreea Gorgos, McGill University Faculty of Medicine and Health Sciences, Montreal, PQ, Canada; Jarred Garfinkle, McGill University Faculty of Medicine and Health Sciences, Montreal, PQ, Canada

Background: A small but important subset of preterm infants ( < 37 weeks) develop cerebral palsy (CP). Early identification of CP allows timely referral and intervention. Prechtl’s General Movements Assessment (GMA) has shown excellent predictive accuracy for CP but its accuracy immediately after implementation in the “real-world” merits further evaluation.

Objective: We aimed to evaluate the accuracy of GMA for predicting CP over time following its local implementation.

Design/Methods: Single-centre retrospective cohort study of infants born < 37 weeks’ gestation and admitted to the Montreal Children’s Hospital NICU between September 2020 and December 2023 who completed GMA and 18-month neurodevelopmental follow-up. The GMA was initially only performed in infants with additional risk factors for CP such as severe intraventricular hemorrhage. Writhing movement patterns were coded as normal, poor-repertoire, or cramped-synchronized (CS) and fidgety movements as present or absent. The primary outcome was CP diagnosis by 18 months’ corrected age. The cohort was divided in 2 by year of birth.

Results: Seventy-four infants met inclusion criteria (median GA 27.9 weeks [IQR 26.1, 29.4]; 45% male; 35% grade 3/4 IVH; and 67.6% with BPD). Twenty infants (27%) were diagnosed with CP at a median corrected age of 12.1 months (9.4, 20.6), most commonly spastic quadriplegia (55%) with a median GMFCS of 2 (1, 4). Of the 20 infants with CP evaluated in the writhing period, 13 had CS (sensitivity 65%) while 35 of 52 infants without CP did not have CS (specificity 67%). All 16 infants with CP evaluated in the fidgety period had absent fidgety (sensitivity 100%) while 32 of 46 infants without CP had fidgety movements (specificity 70%). When divided by year of birth (2020-2022 vs 2023), the sensitivity (63% vs. 75%) and specificity (73% vs. 63%) of CS and the sensitivity (100% vs. 100%) and specificity (74% vs. 67%) of absent fidgety did not change over time (p>0.05 for all comparisons).

Conclusion(s): In this Canadian NICU cohort, abnormal GMA in both cramped synchronized and absent fidgety movements was strongly predictive of later CP albeit with a specificity lower than that historically reported. Diagnostic accuracy remained similar over successive years of program implementation, underscoring the value of standardized training and integration of GMA into routine neonatal follow-up. Broader adoption of GMA into Canadian clinical contexts may enable earlier CP diagnosis and timely rehabilitation referral for preterm infants at risk for CP.