(LB02-05-24) The Association of Prenatal Nutrient Intake With Children’s Autism Outcomes Using a Mixtures Approach: Results From the ECHO Program

Objectives: When examined separately, prenatal intake of folate, vitamin D, omega 3 and omega 6 fatty acids, and iron have been associated (usually inversely) with children’s autism outcomes; however, nutrients are not consumed individually. We used Bayesian kernel machine regression (BKMR) to explore how these nutrients may associate with autism outcomes in combination while addressing concerns of multicollinearity and multiple testing.

Methods: Participants were drawn from the large, diverse US Environmental influences on Child Health Outcomes (ECHO) cohort consortium. Maternal prenatal intake was reported via validated food frequency questionnaires. Children’s (age 2.5-18y) autism-related traits were measured using the Social Responsiveness Scale, and autism diagnoses were based on parent report and/or medical records. BKMR was used to examine combined, individual, and interactive effects of the 5 nutrients on autism traits score and, separately, diagnosis. Models adjusted for potential confounders including cohort characteristics, maternal sociodemographic factors, and child sex and birth year. BKMR models provide graphical illustrations of relationships, with less focus on numerical measures of magnitude or significance, so traditional linear and logistic regression models were run for comparison.

Results: Analyses included 2,614 maternal-child dyads across 7 cohorts. In BKMR models, higher concentrations of the 5 nutrient mixture were associated with reduced autism traits scores. U-shaped associations and bivariate interactions of folate and omega 3 fatty acids with autism traits were observed in BKMR but not traditional regression models. Clear associations with autism diagnosis were not observed.

Conclusions: Using innovative statistical methods and a large sample, we found complex associations between prenatal intake of multiple nutrients and children’s autism-related traits that were not previously identified using traditional models. This underscores the need for continued exploration of complex outcome relationships involving multiple nutrients.

Funding Sources: The ECHO Program is funded by the NIH.