Recent neurodevelopmental outcomes of extremely and very preterm infants: A multicenter retrospective cohort study

Advances in neonatal intensive care have improved survival of extremely and very preterm infants, yet neurodevelopmental outcomes remain a major concern. Contemporary data on developmental trajectories and risk factors are essential for optimizing follow-up strategies.

We conducted a retrospective cohort study of 193 preterm infants born at <32 weeks' gestation at two tertiary NICUs in Korea (2013-2024). Neurodevelopmental outcomes were assessed using the Korean Developmental Screening Test (K-DST) and Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III) at 18-24 months corrected age. The primary objective was to identify perinatal and neonatal risk factors associated with developmental delay (Bayley-III composite score < 70). Secondary objectives included evaluating the diagnostic accuracy of K-DST against Bayley-III, with temporal trends analyzed as an exploratory subanalysis. Risk factors were identified using Firth's penalized likelihood method, with overall and EP/VP-stratified analyses.

Of 432 infants meeting the inclusion criterion of <32 weeks' gestation, 62 died before discharge, leaving 370 survivors. Of these, 193 (52.2%) completed neurodevelopmental follow-up at 18-24 months corrected age, including 175 (47.3%) with K-DST and 108 (29.2%) with Bayley-III assessments. K-DST screening (n = 175) identified 62.3% as below normal range, with language most frequently affected (38.9%). Bayley-III (n = 108) mean composite scores were 90.7 ± 16.8 (cognitive), 89.3 ± 19.4 (language), and 91.8 ± 17.0 (motor), with delay prevalence of 12.0%, 13.9%, and 12.0%, respectively. EP infants scored significantly lower than VP infants in cognitive (84.3 vs. 93.5, p = 0.009) and motor (85.6 vs. 94.5, p = 0.020) domains. K-DST showed moderate sensitivity (58-85%) and specificity (75-85%) against Bayley-III thresholds, with high negative predictive value (>80%) supporting its utility as a screening instrument. On multivariate analysis adjusting for gestational age, sepsis (adjusted OR 4.75-6.16) and ROP requiring surgery (adjusted OR 1.85-59.71) were independently associated with developmental delay, although wide confidence intervals reflect model instability with limited events. In an exploratory subanalysis, the 2018-2024 birth cohort showed lower delay prevalence than the 2013-2017 cohort, but substantial case-mix differences (the earlier cohort had more EP infants and higher sepsis rates) preclude attribution to genuine improvements in care.

ROP requiring surgery and sepsis were the most robust independent risk factors for developmental delay in extremely and very preterm infants. K-DST demonstrated acceptable diagnostic accuracy as a screening tool against Bayley-III. Apparent temporal improvements in outcomes are likely confounded by case-mix differences between birth era cohorts and should be considered hypothesis-generating. Comprehensive developmental surveillance using complementary instruments is essential for this population.