Impact of Erythropoiesis-Stimulating Agents on Behavioral Measures in Children Born Preterm

Abstract

Objective: To evaluate the impact of erythropoiesis-stimulating agents (ESAs) administered during initial hospitalization and family demographic factors on behavior at 3.5-4 years of age.

Study design: Children were enrolled who had previously participated in a randomized study of ESAs (n = 35) or placebo (n = 14) in infants born preterm with birth weights of 500-1250 g. A term healthy control group (n = 22) also was recruited. Behavior was evaluated by parent report with the Behavioral Assessment System of Children-2. Principal component analyses identified 2 demographic factors, a Socioeconomic Composite (SEC) and a Family Stress Composite. A multivariate general linear model evaluated the impact of study group and sex on the 4 composite scales of the Behavioral Assessment System of Children-2. Demographic factors were treated as covariates and interactions with study group (ESA, placebo, and term) were examined.

Results: The ESA group had significantly better scores than the placebo group on behavioral symptoms (P = .04) and externalizing scales (P = .04). An interaction was observed between study group and SEC (P = .001). A beneficial effect of ESAs was maximal in the children with lower SEC scores.

Conclusions: The beneficial effects of ESAs on childhood behavior were maximal in children with lower SEC scores. ESAs seemed to ameliorate the adverse impact of lower SEC on behavioral domains seen in the placebo group. This effect was independent of the beneficial effect of ESAs on global cognition we reported previously.

Trial registration: ClinicalTrials.gov: NCT01207778 and NCT00334737.

Keywords: cognitive development; darbepoetin; erythropoietin; neurodevelopmental outcome; very low birth weight.

Figure 1.

Numbers of infants that were screened for eligibility for the BRITE study, that were eligible for the study, and that were evaluated at 3.5–4 years with completed BASC-2 questionnaires. RCT, randomized controlled trial.

Figure 2.

The interaction of study group (ESA vs placebo) and the SEC factor on BASC-2 scales. SEC was split at the median to create high and low groups. Bars represent SDs. Data from children in the placebo group are represented by white columns and from children in the ESA group by gray columns. Hatched columns represent high SEC data. *P < .05, low SEC placebo group vs the other 3 columns (high SEC placebo group, low SEC ESA group, and high SEC ESA group).