PREDICTIVE VALIDITY OF NEONATAL SCREENING ASSESSMENTS FOR NEURODEVELOPMENTAL OUTCOMES IN PRETERM INFANTS: A PROSPECTIVE COHORT STUDY FROM EASTERN INDIA

Abstract

Background: Preterm birth is a leading cause of long-term neurodevelopmental impairment, and India carries the largest absolute burden of preterm births worldwide. Early neonatal and early-infancy screening could direct scarce early-intervention resources, but the predictive validity of these tools has not been examined in Eastern Indian populations. Objective: To evaluate the predictive validity of the General Movements Assessment (GMA) and the Hammersmith Neonatal Neurological Examination (HNNE) for cognitive and motor outcomes in preterm infants, and to estimate the prevalence of neurodevelopmental delay in this population. Methods: In this prospective observational cohort study, 80 preterm infants (gestational age 25–36 weeks; mean 30.8 ± 3.5 weeks) were recruited from NICUs and follow-up clinics in Eastern India. GMA and HNNE were performed as index tests, and the Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III) served as the reference standard at 18 months corrected age. Neurodevelopmental delay was defined as a cognitive or motor composite score < 85. Sensitivity, specificity, predictive values, likelihood ratios, and area under the receiver operating characteristic curve (AUC) were calculated; Spearman correlation and logistic regression examined associations. Results: Neurodevelopmental delay was present in 40 infants (50.0%); 25 (31.3%) had moderate-to-severe delay (composite < 70). Abnormal GMA (poor repertoire, cramped-synchronized, or absent fidgety movements) predicted delay with a sensitivity of 95.0% and specificity of 95.0% (AUC = 0.950). An HNNE global score ≤ 68 showed a sensitivity of 92.5% and specificity of 97.5% (AUC for the continuous score = 0.979). HNNE scores correlated strongly with Bayley-III cognitive (ρ = 0.92) and motor (ρ = 0.93) composites (both p < .001). Cramped-synchronized or absent fidgety movements identified moderate-to-severe delay with 96.0% sensitivity and 100% specificity. Neonatal sepsis was the only independent clinical predictor of delay (adjusted OR = 10.61, 95% CI 2.54–44.35). Conclusion: GMA and HNNE showed high predictive validity for 18-month neurodevelopmental outcomes in preterm infants from Eastern India. Both tools are low-cost and suitable for triaging infants to early intervention in resource-constrained settings, although external validation in larger multicentre cohorts is required.