Breastfeeding, prenatal depression and children's IQ and behaviour: a test of a moderation model

This study investigates whether breastfeeding is associated with children’s neurodevelopmental outcomes—specifically IQ and behavioural/emotional problems—through mid-childhood, and whether breastfeeding moderates the known adverse associations of prenatal maternal depression and anxiety with child outcomes. Prior work links breastfeeding with improved cognitive outcomes but evidence for emotional/behavioural benefits is mixed. Prenatal maternal distress (depression and anxiety) is robustly associated with poorer cognitive and behavioural outcomes in offspring. The authors aim to clarify these relationships using a large, prospective cohort with extensive confounder control, testing the novel hypothesis that breastfeeding may buffer (moderate) the effects of prenatal distress on child neurodevelopment.

Prior research often reports positive associations between breastfeeding and child cognition, potentially via nutritional components of breast milk such as DHA, AA, sialic acid, zinc, choline, and vitamin B12 that support neurodevelopmental processes including myelination and membrane biosynthesis. Breastfeeding may also enhance maternal sensitivity. Evidence for effects on emotional and behavioural outcomes is less consistent. Conversely, prenatal maternal depression and anxiety predict a range of adverse child outcomes, including lower IQ and increased behavioural/emotional problems. Interest has grown in identifying moderators of prenatal distress effects; breastfeeding is a plausible candidate. However, past studies vary in design quality and adjustment for confounding, leaving causal inference uncertain.

Design and cohort: Prospective longitudinal analysis using the Avon Longitudinal Study of Parents and Children (ALSPAC), a UK birth cohort of pregnancies with expected delivery dates between April 1, 1991 and December 31, 1992. From 14,541 initial participants, 13,988 children were alive at 1 year and 11,096 had data on infant feeding at 1 month, IQ, and SDQ outcomes. Premature and/or low-birthweight infants were excluded. Measures: Breastfeeding was assessed by maternal questionnaires at 1 month (classified as exclusive breastfeeding, mixed feeding, or exclusive formula) and at 6 months (any breastfeeding). Maternal depressive symptoms were assessed with the EPDS at 32 weeks’ gestation and 8 weeks postpartum. Maternal anxiety was assessed with the Crown-Crisp Experiential Index anxiety subscale at the same time points. Child IQ was measured at age 4 years using the Wechsler Preschool and Primary Scale of Intelligence (WPPSI) in the Children in Focus subsample (N=728), and at age 8 years using an abbreviated WISC-III across the full cohort. Behavioural/emotional problems were parent-reported using the Strengths and Difficulties Questionnaire (SDQ) at 57 months (4 years 8 months) and 9 years 7 months; subscales used were hyperactivity/inattention, emotional symptoms, conduct problems, and total difficulties. Covariates: Maternal smoking during pregnancy; partner’s daily cigarette consumption (8 months postpartum); maternal smoking after birth (8 weeks postpartum); maternal and partner education (five-category scales); HOME score at 18 months (1–12); maternal age at delivery; primiparity; gestational age; household crowding index; maternal return to work by 8 months; child sex; birthweight. Race/ethnicity could not be examined due to homogeneity (~97% white/British). Procedures and ethics: Standard ALSPAC procedures with ethics approvals; informed consent obtained. Statistical analysis: Conducted in SPSS v24. Pearson correlations examined bivariate associations. Hierarchical linear regressions tested main effects and interactions. Model 1 included covariates plus prenatal depression or anxiety; Model 2 added breastfeeding; Model 3 added interaction terms and postnatal depression/anxiety. Moderation tests assessed breastfeeding × prenatal (and postnatal) distress interactions. Exploratory mediation analyses tested whether prenatal distress influenced child outcomes via breastfeeding using Sobel tests. Missing data were handled via multiple imputation (5 imputations) for SDQ outcomes, covariates, and prenatal/postnatal distress; IQ at 4 years was not imputed beyond the CIF subsample; IQ at 8 years had imputation as needed. Diagnostics included checks for multicollinearity and residual normality.

- Breastfeeding prevalence: At 1 month, 43% of mothers exclusively breastfed; an additional 16.8% used mixed feeding. - IQ outcomes: After adjustment for covariates and parental distress, exclusive breastfeeding at 1 month predicted higher full-scale IQ at 8 years (B ≈ 2.2 points; SE/SD ≈ 0.36; p < .01), and mixed feeding predicted a smaller IQ increase (B ≈ 1.6 points; SE/SD ≈ 0.53; p < .01). Effects were not significant for IQ at 4 years after full adjustment. Covariates positively associated with IQ included HOME score, maternal and paternal education, maternal age, birthweight, and primiparity; crowding was negatively associated. - Behavioural outcomes: Exclusive breastfeeding at 1 month was associated with fewer hyperactivity/inattention symptoms at 4 years (B ≈ −0.31; SE ≈ 0.05; p < .01). Mixed feeding at 1 month was associated with higher hyperactivity/inattention at 9 years (B ≈ 0.20; SE ≈ 0.09; p < .01). Mixed feeding at 6 months was also associated with higher hyperactivity/inattention at 9 years (B ≈ 0.21; SE ≈ 0.08; p = .01). No significant associations were found between breastfeeding (exclusive or mixed) and emotional symptoms, conduct problems, or total difficulties at either age. - Prenatal and postnatal distress: Prenatal maternal depression and anxiety were associated with higher SDQ problem scores (including hyperactivity/inattention) at 4 and 9 years; prenatal depression showed a very small negative association with IQ at 8 years in adjusted models (Table 2). Postnatal distress also related to behavioural problems. - Moderation and mediation: No evidence that breastfeeding moderated the associations of prenatal (or postnatal) depression or anxiety with child IQ or behavioural/emotional outcomes at either age (all interaction terms non-significant). Sobel tests found no evidence that breastfeeding mediated associations between prenatal distress and child outcomes. - Robustness: Results were consistent across imputed and non-imputed datasets; residuals were approximately normal; multicollinearity indices were acceptable (VIF < 10).

The findings indicate specific benefits of breastfeeding for cognitive development and selected behavioural domains, independent of prenatal maternal distress. The positive associations with full-scale IQ at 8 years for both exclusive and mixed breastfeeding, alongside the lack of consistent associations with emotional or conduct problems, support a more specific (likely nutritional) effect on neurodevelopment rather than broad psychosocial benefits. Contrary to the moderation hypothesis, breastfeeding did not buffer the impact of prenatal depression or anxiety on child outcomes; instead, breastfeeding and prenatal distress exhibited largely independent main effects. This suggests distinct mechanisms—potentially nutritional constituents of breast milk versus intrauterine exposure to stress-related hormones—that differentially influence neurodevelopment. The results align with prior meta-analytic evidence of modest IQ advantages associated with breastfeeding and reinforce the need to understand which breast milk components influence cognition and how these effects evolve developmentally.

In a large, well-characterized cohort with extensive confounder control, exclusive and mixed breastfeeding were associated with higher IQ in mid-childhood and exclusive breastfeeding with fewer hyperactivity/inattention symptoms at age 4. Breastfeeding neither moderated nor mediated the associations between prenatal (or postnatal) depression/anxiety and child cognitive or behavioural outcomes, indicating largely independent pathways. Future research should examine these relationships in more ethnically diverse populations, identify specific breast milk components driving cognitive benefits and hyperactivity/inattention associations, and evaluate implications for infant formula development when breastfeeding is not feasible.

- Potential shared method variance for parent-reported behavioural outcomes, though lack of associations across all SDQ domains argues against a global bias. - No direct data on breast milk nutritional composition; inability to link specific nutrients to outcomes. - Data collected in the 1990s; infant formula composition and breastfeeding prevalence may have changed since, potentially affecting generalizability. - Residual confounding possible due to unavailable factors (e.g., maternal IQ, school quality, child medical history). - No information on changes in feeding status over time beyond measured time points. - Lack of racial/ethnic diversity in ALSPAC limits generalizability to minority groups. - The absence of moderation for prenatal distress may not generalize to other outcomes (e.g., immune health).