Prenatal Nutritional Supplementation and ASD: Causal Inference Analyses

Background:   Studies have suggested that early prenatal supplementation with folic acid, iron, or multivitamins may be protective against ASD, although other studies have not found evidence of this

Objectives:   Using a large population-based study with high case ascertainment, we investigated relationships between early maternal nutritional supplementation (folic acid, iron, and/or multivitamins) and ASD, stratified by intellectual disability (ID).

Methods:   The Stockholm Youth Cohort, a prospective cohort study, was examined using Swedish national and regional health registers.  The exposure of interest was nutritional supplementation at the first antenatal visit, recorded in the Medical Birth Register.  Children born between 1996 and 2005 were followed up for a diagnosis of ASD until December 31, 2011.

Results: Of 206,771 participants in this sample, 4317 cases (2.1%) of ASD were identified, consisting of 984 (0.5%) low-functioning autism cases (LFA; defined as the presence of comorbid ID) and 3333 (1.6%) high-functioning autism cases (HFA; defined as the absence of comorbid ID).  Supplement users were different from non-users across multiple medical and social characteristics.

Our primary analysis indicated that overall, taking multivitamins or multiple vitamins (MV) by the first antenatal visit (median 10.1 weeks gestation) was associated with a lower likelihood of LFA relative to mothers who did not take nutritional supplements (adjusted odds ratio: 0.70 [95% confidence interval: 0.56-0.87]), whereas taking folic acid or iron alone was not associated with reduced risks of LFA (adjusted odds ratios: 1.24 [95% confidence interval: 0.91-1.70] and 1.04 [95% confidence interval: 0.89, 1.20], respectively).  Supplement use was not associated with overall risk of HFA.

To inform interpretation of our LFA results, we will present results from causal inference analyses including a matched sibling analysis and propensity score (PS) matching.  Since siblings share the same mother, a matched sibling analysis reduces the potential for residual confounding by maternal health status and behaviors.  PS matching is a technique that matches supplement users and non-users based on the estimated probability of taking specific nutritional supplements prenatally.  Matching on the PS can also reduce the potential for residual confounding since there is less reliance on model-based assumptions.

Conclusions:   Early maternal vitamin supplementation, especially multivitamins or iron and folic acid used in combination, may be protective against low functioning autism.  However, there was no evidence that either iron or folic acid supplements used individually was associated with either high or low functioning ASD.  Results from sibling analyses and propensity score matching will aid in interpretation of potential causality.