(35) Early Prediction of Cognitive Impairment in Preterm Infants: A Systematic Review

Background: Preterm infants are at a higher risk than their full-term counterparts for developmental disabilities later in life. One such disability is in cognitive development, impacting learning, school performance, and academic success during critical years. Numerous studies have attempted to identify early NICU predictors of adverse cognitive outcomes to enable aggressive early interventions and minimize long-term consequences.

Objectives: We performed a systematic review to identify and critically appraise all prognostic studies in preterm infants that have published the development, validation, or updating of a model that uses multivariable modeling techniques for making individualized predictions of adverse cognitive impairment/adverse development.

Design/Methods: Articles were extracted from various databases such as Embase, Scopus, PubMed, CINAHL, and PsycINFO using specific search terms (Figure 1) to ensure comprehensive coverage of relevant literature. Two independent reviewers then conducted a rigorous screening process based on predetermined criteria, followed by a third adjunct reviewer resolving any conflicting decisions. Specifically, we only selected studies that made predictions prior to six months of age for cognitive outcomes measured after at least eighteen months of age and had a minimum sample size of 50 infants of gestational age ≤36 weeks.

Results: Ultimately, 33 studies met the inclusion criteria and were included in the review (Tables 1 and 2). We critically appraised each study using a standardized measure of risk of bias (Figure 2). We found considerable heterogeneity in the approach taken by and respective performance of each model presented. Based on reported measures, conventional brain MRI appeared to be the most accurate and the most studied modality. However, as seen in both MRI-based and other models, overlap of the specific predictors between studies with similar areas of focus was minimal. This lack of replication makes the assessment of specific predictors’ relative strength quite difficult, given the potential influence of study-specific bias and limitations.

Conclusions: Large, well-designed, internally validated studies are necessary to develop robust and accurate models that can enable early, individualized prediction of cognitive impairment in preterm infants.