Maternal Intake of Supplemental Iron and Risk of Autism Spectrum Disorder

Rebecca J. Schmidt; Daniel J. Tancredi; Paula Krakowiak; Robin L. Hansen; Sally Ozonoff

Disclosures

Am J Epidemiol. 2014;180(9):890-900. 

In This Article

Results

Characteristics of participants with ASD and those with typical development (TD) are shown in Table 1. Children with ASD were more likely to have been born earlier than children with TD, and their mothers were significantly more likely to have some college education but no bachelor's or higher degree, to have smoked cigarettes, and to have taken a multivitamin during the index period and less likely to have eaten cereal during the index period. Their mothers also tended to be older, were more likely to have been born outside of the United States, and were less likely to have private medical insurance. Parents of children with ASD were significantly less likely to own their homes.

Of the 520 ASD and 346 TD CHARGE participants eligible for these analyses, 510 (98%) ASD participants and 341 (99%) TD participants had information on maternal use of iron supplements during the index period, and 454 (87%) ASD participants and 307 (89%) TD participants had information on total average maternal iron intake from all sources.

Twenty-five percent of mothers of children with ASD and 31% of mothers of TD children reported taking an iron-specific supplement at any time during the index period (Table 2), producing a significant association between reported iron supplement use and ASD (odds ratio (OR) = 0.63, 95% confidence interval (CI): 0.44, 0.91) after adjustment for maternal folic acid intake, home ownership, and the child's birth year. More mothers reported taking an iron supplement in the latter half of pregnancy (Figure 1). Adjusted odds ratios for associations between ASD and reported use of iron supplements were consistently below the null across the index period but were nonsignificant except during breastfeeding (among those who breastfed) (Figure 2; also see Web Table 1 http://aje.oxfordjournals.org/content/180/9/890/suppl/DC1, available at http://aje.oxfordjournals.org/). Adjusted odds ratios also differed by the child's birth year (Web Table 2, Web Figure 1 http://aje.oxfordjournals.org/content/180/9/890/suppl/DC1).

Figure 1.

Percentages of mothers of children with autism spectrum disorder and mothers of children with typical development who took iron supplements during the index period (from 3 months before pregnancy through the end of pregnancy and during breastfeeding), Childhood Autism Risks from Genetics and the Environment (CHARGE) Study, California, 2003–2009. A trend towards a significant difference between groups was found during the breastfeeding (BF) period (P = 0.06).

Figure 2.

Adjusted odds ratios for associations between reported maternal iron supplement intake during the index period (from 3 months before pregnancy through the end of pregnancy and during breastfeeding) and autism spectrum disorder, Childhood Autism Risks from Genetics and the Environment (CHARGE) Study, California, 2003–2009. Odds ratios were adjusted for maternal folic acid intake during the first month of pregnancy, child's year of birth, and home ownership. Vertical bars represent 95% confidence intervals. Frequencies and P values are presented in WebTable 2. There was a significant association with autism spectrum disorder during the breastfeeding (BF) period (P < 0.05) and a borderline-significant association during months 4, 5, 6, and 9 of pregnancy (P < 0.10).

Prenatal vitamins were the greatest source of iron for mothers in both diagnostic groups during the index period (Table 3). Mean daily maternal iron intake from cereal and total iron intake from all collected sources during the index period were significantly lower for children with ASD than for children with TD, while maternal intake of iron from multivitamins was higher (Table 3).

Mean total iron intake was highest in the second half of pregnancy for both groups (Figure 3). Mean intake reported by case mothers was lower than that reported by control mothers, especially for the months before pregnancy and during early pregnancy (Figure 3). During this time (from 3 months before pregnancy through the second month of pregnancy), nonsignificantly more case mothers (74%) than control mothers (69%) (P = 0.13) had iron intakes below the Dietary Reference Intake for iron established by the Institute of Medicine for nonpregnant (18 mg/day) and pregnant (27 mg/day) women aged 19–50 years.[39]

Figure 3.

Mean iron intakes of mothers of children with autism spectrum disorder and mothers of children with typical development during the index period (from 3 months before pregnancy through the end of pregnancy and during breastfeeding), Childhood Autism Risks from Genetics and the Environment (CHARGE) Study, California, 2003–2009. Vertical bars represent 95% confidence intervals. Dietary Reference Intakes for iron in females aged 19–50 years are 18 mg/day for all women, 27 mg/day during pregnancy, and 9 mg/day during lactation (39). Significant differences between mothers of children with autism spectrum disorder and mothers of children with typical development were found for the 3 months before and the first month of pregnancy (P < 0.01) and for the second month of pregnancy (P < 0.05). A borderline-significant association was observed for months 3, 4, and 6 of pregnancy and during the breastfeeding (BF) period (P < 0.10).

The highest category of maternal iron intake (≥86 mg/day) during the index period was associated with significantly reduced risk of ASD in the child, before and after adjustment for supplemental periconceptional folic acid intake, child's birth year, and home ownership (OR = 0.49, 95% CI: 0.29, 0.82) (Table 4). ASD risk decreased as mean maternal iron intake increased (P trend = 0.01) (Table 4). These findings remained after imputation of missing values and when using survey weights (Table 4).

Adjusted odds ratios for the association between ASD and each quintile of maternal iron intake differed across the index period: Odds ratios were near the null during the months before pregnancy; above the null for the first 2 months of pregnancy (after adjustment for the highly correlated folic acid intake during this time); consistently below the null for the highest quintile from the third month of pregnancy onward; and below the null during breastfeeding (Figure 4, WebTable 3 http://aje.oxfordjournals.org/content/180/9/890/suppl/DC1). Unadjusted odds ratios were more consistent across time (Web Figure 2 http://aje.oxfordjournals.org/content/180/9/890/suppl/DC1). Post-hoc analysis showed that estimates for iron-specific supplement intake and the highest quintiles of iron intake during breastfeeding were not meaningfully different after adjustment for folic acid intake during breastfeeding (Web Table 4 http://aje.oxfordjournals.org/content/180/9/890/suppl/DC1).

Figure 4.

Adjusted odds ratios for associations between mean maternal iron intake during the index period (from 3 months before pregnancy through the end of pregnancy and during breastfeeding) and autism spectrum disorder, Childhood Autism Risks from Genetics and the Environment (CHARGE) Study, California, 2003–2009. Odds ratios were adjusted for maternal folic acid intake during the first month of pregnancy, child's year of birth, and home ownership. Vertical bars represent 95% confidence intervals. Frequencies and P values are presented in WebTable 3. There was a significant inverse association with autism spectrum disorder for the highest iron quintile (P = 0.02) during the breastfeeding (BF) period; a significant positive association for the fourth quintile (P = 0.04) and a borderline-significant association for the third quintile (P = 0.10) during the second month of pregnancy; and a borderline-significant association for the fourth quintile (P = 0.06) during breastfeeding.

There were no significant differences in mean maternal iron intake on the basis of ASD-related characteristics such as regression, delayed or atypical development, seizure, or verbal status (Web Table 5 http://aje.oxfordjournals.org/content/180/9/890/suppl/DC1); however, children with ASD who had experienced early-onset symptoms and delayed or atypical development had larger differences from TD children in mean maternal iron intake than children with ASD who had experienced regression and who did not have delayed or atypical development.

No significant interactions were observed between maternal iron intake and child sex, child or maternal race/ethnicity, or interpregnancy interval. A significant multiplicative interaction (P = 0.002) was found between low maternal iron intake and older maternal age at delivery, with more than a 5-fold increased risk of ASD for mothers aged 35 years or older with iron intake in the lowest quintile as compared with younger mothers with iron intake in the highest quintile (OR = 5.01, 95% CI: 1.98, 12.69) (Table 5). The estimate for the combination of older maternal age and low iron intake was more than 5 times that expected from adding (OR = 0.98) or multiplying (OR = 0.80) their independent associations. There was also a significant multiplicative interaction (P = 0.01), with higher ASD risk being associated with the combination of mothers having metabolic conditions and iron intake in the lowest quintile (OR = 4.72, 95% CI: 1.69, 13.15), which was over twice that expected from adding (OR = 2.07) or multiplying (OR = 2.36) their independent associations (Table 6).

In analysis conducted to assess recall bias, the associations between reduced ASD risk and both use of an iron-specific supplement during the index period (OR = 0.52, 95% CI: 0.30, 0.90) and highest quintile of maternal iron intake versus the lowest (OR = 0.38, 95% CI: 0.18, 0.83; P trend = 0.03) were stronger when data were limited to children who were under 3.6 years of age (the median age of controls) at the time of interview start than in children who were older (for taking an iron-specific supplement, OR = 0.74, 95% CI: 0.45, 1.20; for highest quintile of maternal iron vs. lowest, OR = 0.65, 95% CI: 0.31, 1.37 (P trend = 0.11)) (Web Table 6 http://aje.oxfordjournals.org/content/180/9/890/suppl/DC1). There were no significant differences in associations for taking an iron-specific supplement during the index period or higher maternal iron intake when data were stratified by maternal folic acid intake during the first month of pregnancy (<600 µg/day, ≥600 µg/day) (Table 7). Finally, adjusted odds ratios for the association between quintile of iron intake and ASD differed somewhat by the child's birth year, but not for the highest quintile of iron intake compared with the lowest, for which odds ratios were consistently below the null (Web Table 7, Web Figure 3 http://aje.oxfordjournals.org/content/180/9/890/suppl/DC1).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....