508 - Latent Profile Modeling of Childhood Neurocognitive Outcomes: Association with Antenatal Steroid Exposure and Sex

Poster Number: 508
Publication Number: 508.1108

Background: While most studies on perinatal exposures and neurodevelopment focus on impairment in single domains, more recent studies classified patients into outcome groups derived from assessments across multiple domains, such as cognition, motor function, and behavioral competencies.

Objective: Latent profile analysis (LPA), a patient-centered global approach, was used to identify cognitive, social and behavioral subgroups of children born to mothers at risk for late preterm delivery (34-36 weeks).

Design/Methods: Secondary analysis of a prospective follow-up study of children aged 6-11 years old whose mothers participated in an antenatal steroids (betamethasone) trial due to their risk of late preterm delivery. Children were included if they completed neurocognitive testing at ≥ 6 years old. The Differential Abilities Scales, 2nd Edition (DAS-II) measured cognitive function, the Social Responsiveness Scale (SRS) social abilities, and the Child Behavior Checklist (CBCL) identified problem behaviors. Children with similar patterns of scores on the DAS-II, SRS, and CBCL were identified using LPA and standard approaches to obtain the optimal number of profiles. Distributions of scores were also examined by treatment assignment and sex.

Results: Of 1,026 children enrolled, 944 (92.0%) completed neurocognitive testing. Median gestational age at delivery was 36 weeks and median age at follow-up 7 years. Three profiles were identified: 490 (51.9%) in profile 1, 352 (37.3%) in profile 2, and 102 (10.8%) in profile 3. Children in Profile 1 had normal average scores for cognition, social responsiveness, and behavioral problems. Children in Profile 3 had the most severe outcomes, with a mean SRS T-score of 72.5 indicative of moderate social impairment and mean CBCL T-score of 65.6 (> 90th percentile). There were four profiles by treatment assignment, and children in profiles 2-4 had mild, moderate, and severe SRS T-scores, with higher scores reflecting increasing levels of social responsiveness impairment. The distribution of profiles differed by treatment assignment, but not sex. Specifically, children in the betamethasone group were more likely to be in profile 1 compared with the placebo group (54.0% vs. 43.6%, p=0.009).

Conclusion(s): We identified distinct subgroups with varying levels of cognitive, social and behavioral problem severity among children born predominantly late preterm and demonstrated how LPA, as compared to a single neurodevelopmental assessment, might provide a more comprehensive profile of neurodevelopmental outcomes.