Several studies have evaluated breastfeeding and its association with childhood neurodevelopmental outcomes. Some of these studies, including a recent meta-analysis, have demonstrated a relationship between breastfeeding, particularly exclusive breastfeeding, and better child neurodevelopmental outcomes.1–7 A few studies have demonstrated associations with breastfeeding and continued intellectual and achievement benefit through adolescence and adulthood.8–11 In addition, in a population-based study of an urban cohort in Brazil, short duration of breastfeeding (less than 1 month) was reported as an independent predictor of low IQ scores at age 6 years, after adjusting for multiple demographic and family conditions.12 Similarly, lack of breastfeeding was associated with both low IQ score and attention deficit–hyperactivity disorder (ADHD) in children aged 8–11 years from schools in five Korean cities, after adjustment for social and demographic factors including maternal IQ score.13 A recent meta-analysis of 11 studies also found that lack of breastfeeding and short breastfeeding duration were associated with the development of ADHD.14
The composition of breast milk, particularly early in infancy, is rich in long-chain polyunsaturated fatty acids, compared with infant formulas.15,16 Infant brain development depends, in part, on myelination and the corresponding expansion of neural networks throughout infancy and early childhood.17–19 Myelination contributes to the development of neural pathways that then provide the foundation for emerging cognitive and behavioral developments.20 The myelination process requires the delivery of long-chain polyunsaturated fatty acids, among other nutrients.21 Theoretically, through enhanced myelination, breastfeeding could be associated with early child neurodevelopment and potentially long lasting benefit.15
We hypothesized that breastfeeding and its duration are associated with a reduced risk of low IQ scores at age 5 years and, potentially, other neurodevelopmental problems. The primary objective of this analysis was to evaluate whether breastfeeding and its duration are associated with a lower risk of low IQ scores at age 5 years as compared with no breastfeeding. The secondary objective was to evaluate whether breastfeeding is associated with a lower risk of other neurodevelopmental problems.
We conducted a secondary analysis of two Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network trials, titled, “Treatment of Subclinical Hypothyroidism or Hypothyroxinemia in Pregnancy.”22 These trials were parallel-design randomized controlled trials of thyroxine for either subclinical hypothyroidism or hypothyroxinemia and took place across 33 hospitals in the United States. People with singleton pregnancies were screened before 20 weeks of gestation for subclinical hypothyroidism (thyrotropin 3.0 milliunits/L or greater and free thyroxine 0.86–1.9 ng/dL) or hypothyroxinemia (thyrotropin 0.08–3.99 milliunits/L and free thyroxine less than 0.86 ng/dL) and randomized to receive levothyroxine or placebo. Patients with overt hypothyroidism or hyperthyroidism were excluded. For participants, thyroid function was assessed monthly and (depending on trial) adjusted to achieve normal free thyroxine levels. Sham adjustments were performed for participants in the placebo group. Children underwent annual neurodevelopmental testing for 5 years. The primary outcome of the parent trial was IQ score at age 5 years, or at age 3 years if the 5-year data were missing or death occurred after age 3 years. The study’s conclusions were that neither treatment for subclinical hypothyroidism nor hypothyroxinemia beginning between 8 and 20 weeks of gestation resulted in significantly better cognitive outcomes in children through age 5 years, as compared with no treatment for these conditions. Institutional review board approval was obtained at each clinical site for the primary trial.
In this secondary analysis, we included people for whom breastfeeding data, collected by self-report 2 years after delivery, were available and 3- or 5-year child neurodevelopmental data were available. We excluded children with chromosomal or structural anomalies and those who required neonatal intensive care admission. The primary outcome for this secondary analysis was a low IQ score at age 5 years, defined as score less than 85 on the WPPSI-III (Wechsler Preschool and Primary Scale of Intelligence III). Secondary outcomes included the following: an IQ score at age 5 years; low DAS-II (Differential Ability Scales-II) General Conceptual Score at age 3 years, defined as score less than 85; DAS-II General Conceptual Score at age 3 years; DAS-II–specific scores on subtests at age 4 years (recall digits forward and picture recognition, with a low score defined as less than the 25th percentile); Bayley III Cognitive, Motor, and Language scores (ages 1 and 2 years), evaluated by total score and a score of less than 85 for each domain; a CBCL (Child Behavioral Checklist) T-score higher than 60 (ages 3 and 5 years); and Conners’ Rating Scales-Revised for assessment of an ADHD T-score higher than 60 (age 4 years). T-scores higher than 60 for both the CBCL and Conner’s Rating Scales-Revised were considered to be of possible clinical concern.22 Breastfeeding occurrence was self-reported 2 years after delivery and categorized as any compared with none. Duration was categorized as zero for no breastfeeding, and the median of the following categories: less than 4 months, 4–6 months, 7–9 months, 10–12 months, and more than 12 months. Demographic and clinical data were obtained by interview, and chart abstraction by trained research personnel. Race was determined by participant self-identification as either White, Black, Asian, Native Hawaiian or Pacific Islander, other, not reported. Ethnicity was self-reported as Hispanic or non-Hispanic. Owing to small numbers, Asian, Native Hawaiian or Pacific Islander, other and not reported were collapsed into other. Small for gestational age was defined as a birth weight less than tenth percentile for gestational age based on a neonatal sex and race and ethnicity–specific nomogram.23 Trained and certified examiners assessed all neurodevelopmental outcomes. Research staff, examiners and patients were unaware of trial or treatment group.
Univariable analyses were performed using χ2 or Fisher exact tests for categorical variables and Wilcoxon rank sum test for continuous variables to compare maternal characteristics and child outcomes based on breastfeeding occurrence (any vs none). Linear regression analysis examining the association between breastfeeding and child outcomes was reported on a continuous scale; logistic regression analyses examining the association between breastfeeding and binomial outcomes were also performed. To develop the final adjusted models, stepwise backward proceeding multivariable regression models were used. Variables eligible for multivariable analysis included maternal age, race and ethnicity, prepregnancy body mass index (BMI, calculated as weight in kilograms divided by height in meters squared), parity, education level, insurance type, smoking, alcohol use, thyroid status, treatment group, gestational age at delivery, birth weight less than the 10th percentile, neonatal sex, and age at neurodevelopmental examination. Variables with a P<.1 in the stepwise models were included in final regression models, and those that did not achieve P<0.1 were excluded from the final regression models. For outcomes in which there was a significant association with breastfeeding in the multivariate analysis, we also examined the duration of breastfeeding. To evaluate the association between duration of breastfeeding and neurodevelopmental outcomes, we ran logistic regression models using 0 for no breastfeeding and the midpoint of each of the prespecified breastfeeding duration categories2,6,8,12,18 in the models to provide the adjusted odds per month of breastfeeding and a 95% CI. We further explored whether there was a curvilinear relationship with duration of breastfeeding by generating models that included a quadratic term for duration of breastfeeding. For the primary outcome and all secondary outcomes, we performed sensitivity analyses to evaluate whether inclusion of treatment (thyroxine vs placebo) or group assignment (subclinical hypothyroidism vs hypothyroxinemia) altered the conclusions. Similarly, we performed a propensity score matching analysis as an alternative analytic strategy to corroborate our findings.
Imputation for missing data was not performed. Nominal P values less than .05 were considered statistically significant. No adjustments were made for multiple comparisons. All analyses were performed using SAS 9.4. Approval was obtained from the institutional review board at each participating institution.
Of the 1,203 participants in the trial, 772 were included in this analysis (Fig. 1) of whom 614 (80%) reported breastfeeding: 188 (31%) for less than 4 months; 118 (19%) for 4–6 months; 65 (11%) for 7–9 months; 93 (15%) for 10–12 months; and 141 (23%) for more than 12 months. Of the 141 who reported breastfeeding for more than 12 months, 34 (24.1%) reported that they were still breastfeeding at the time of the 2-year exam ination. Nine participants breastfed for an unknown duration. Maternal demographic and clinical characteristics according to breastfeeding occurrence are displayed in Table 1. Participants who breastfed were older and had lower BMIs, on average, and were more likely to be non-Hispanic White, privately insured, nonsmokers, have a college degree, have subclinical hypothyroidism, and were less likely to have subclinical hypothyroxinemia.
WPPSI-III IQ scores were available for 756 children whose average age at examination was 60.1±1.7 months. The Bayley 1-year examination scores were available for 758 children whose average age at examination was 12.0±1.2 months. The Bayley 2-year examination scores were available for 764 children whose average age at examination was 24.0±1.1 months. The DAS-II 3-year examination scores were available for 757 children whose average age at examination was 36.0±1.3 months. The Conners’ Rating Scales-Revised ADHD was available for 747 children whose average age was 48.0±1.1 months. The CBCL 5-year evaluation score was available for 760 children whose average age at examination was 60.1±1.7 months.
In unadjusted analysis, low IQ scores at age 5 years occurred less frequently after any breastfeeding (21.5%) than without breastfeeding (36.2%, odds ratio [OR] 0.48, 95% CI 0.33–0.71) (Table 2). Similarly, mean IQ scores were higher at age 5 years with any breastfeeding (96.7±15.1) than without (91.2±15.0, mean difference 5.5, 95% CI 2.8–8.2). The DAS II outcomes assessed at age 3 years were similar to the age 5 years IQ scores. Low DAS II scores occurred less frequently after any breastfeeding (32.2%) as compared with no breastfeeding (43.5%, OR 0.62, 95% CI 0.43–0.88), and the mean score was higher among breastfed infants (92.0±15.8) as compared with no breastfeeding (88.2±15.4, mean difference 3.8, 95% CI 0.99 to 6.6). Low scores for the DAS II subtest on picture recognition at age 4 years occurred less frequently with breastfeeding (17.6%) as compared with no breastfeeding (30.0%, OR 0.50, 95% CI 0.33–0.75). Similarly, low CBCL scores at age 5 years were less common among breastfed children (9.2%) as compared with those who were not breastfed (16.3%, OR 0.52, 95% CI 0.31–0.87). Mean Bayley III motor scores and Bayley III language scores at age 2 years were also significantly higher with any breastfeeding compared with no breastfeeding. In unadjusted analyses, all other neurodevelopmental assessments were similar between those children who were breastfed and those who were not.
In multivariable logistic regression analysis, breastfeeding was associated with reduced odds of low IQ score at age 5 years (adjusted odds ratio [aOR] 0.62, 95% CI 0.41–0.93, P=.02) (Table 2; see Table 3 for variables included in the final regression models). When breastfeeding duration was examined, we observed 3.3% reduced odds of an IQ score less than 85 at age 5 years with each month of breastfeeding (aOR 0.97, 95% CI 0.939–0.996, P=.03).
Models that examined a curvilinear relationship between breastfeeding duration and low IQ score at age 5 years were not significant for duration of breastfeeding. No other neurodevelopmental outcome, including mean IQ scores at age 5 years, was found to be significantly associated with breastfeeding in multivariate analyses (Table 2; see Table 3 for variables included in the final regression models).
Sensitivity analyses were performed in which treatment and group assignment were included in each of the models. The results and conclusions were not altered in any of the analyses. The results for multivariable analyses for the association between breastfeeding and its duration are provided in Tables 4 and 5, respectively. The results for the other sensitivity analyses are not shown. In the propensity score analysis, 158 no breastfeeding and 614 breastfeeding participants resulted in 122 matched pairs with a total sample size of 244. Although not significant with this reduced sample size, breastfeeding still suggested a protective effect with respect to the primary outcome (OR 0.72, 95% CI 0.413–1.270).
We found breastfeeding to be associated with a lower odds of low IQ scores in children at age 5 years and that each month of breastfeeding is associated with a modest but significant reduction in the odds of a low IQ score at age 5 years. We did not find significant associations between breastfeeding and other neurodevelopmental outcomes including mean IQ scores at age 5 years. These conclusions remained unchanged when group assignment (subclinical hypothyroidism vs hypothyroxinemia) and treatment (thyroxine vs placebo) were included in our models.
Our analysis focused on low IQ score (less than 85) as the primary outcome. No breastfeeding has been reported as an independent marker of low cognitive function among children born prematurely. In a French population-based cohort study of 1,503 children born in 1997 before 33 weeks of gestation, parental socioeconomic status and lack of breastfeeding, in addition to medical factors, were independently associated with an increased risk of mild cognitive impairment.24 This same cohort was subsequently analyzed along with a second cohort of children born prematurely before 33 weeks of gestation in France between 2003 and 2008. When these additional 1,733 children were added to the analysis, breastfeeding was similarly independently associated with a reduced risk of suboptimal neurodevelopment assessment at ages 2 and 5 years.25 Short breastfeeding duration (less than 1 month), as compared with breastfeeding for one month or more, has also been associated with more frequently low IQ score in a 2004 birth cohort from Pelotas, southern Brazil, a middle-income country.12 It is possible that breastfeeding has a differential association for children who may be at higher risk for poor neurodevelopmental outcomes, compared with those who are not at high risk. Further study of the association between neurodevelopmental outcomes and breastfeeding by risk stratification may be warranted.
Short breastfeeding duration also has been reported to be significantly associated with the subsequent development of attention deficit hyperactivity disorders in children.13,14 A recently published meta-analyses analyzed 11 previous studies and found a significant association between not breastfeeding or short duration of breastfeeding and the subsequent diagnosis of attention deficit disorders.14 However, our analysis did not find an association between breastfeeding occurrence and abnormal results on the Conners' Rating Scales-Revised for assessment of attention in adjusted analyses. Our discordant results may be the result of small sample size with inadequate power in our study cohort.
In addition to the limitations in detecting small differences in outcome measures owing to the fixed sample size, our analysis is limited in that other factors that could influence child neurodevelopmental outcomes such as parental IQ score and the home and educational environment were not available to include in these analyses. Similarly, it must be noted that neonates who were admitted to the neonatal intensive care unit were excluded from our analyses; thus, our conclusions cannot be applied to this population. Although unlikely to be a major contributor, the use of donor breast milk was not specifically captured in the data set. Our study also was limited, as are all observational studies, in that it cannot address causality. Although the association between breastfeeding (and breastfeeding duration) and lower odds of low IQ score in children persisted despite adjusting for multiple covariates, we acknowledge the possibility of unmeasured residual confounding. Finally, no statistical adjustments were made for multiple comparisons. As a result, the conclusions should be interpreted accordingly.
The strengths of our study are numerous. The patient population is racially and ethnically diverse and drawn from 33 geographically varied hospitals across the United States, making the findings broadly generalizable. In addition, annual child assessments were performed from ages 1 to 5 years to provide a comprehensive assessment of neurocognitive outcomes. We were able to adjust for a number of key sociodemographic covariates such as maternal education level and insurance type which are associated with the outcome of interest, IQ scores in the offspring. Furthermore, trained and certified personnel, blinded to maternal data, performed each of these assessments to ensure valid measurement.
1. Horta BL, Loret de Mola C, Victora CG. Breastfeeding and intelligence: a systematic review and meta-analysis. Acta Paediatr 2015;104:14–9. Doi: 10.1111/apa.13139
2. Horwood LJ, Darlow BA, Mogridge N. Breast milk feeding and cognitive ability at 7–8 years. Arch Dis Child Fetal Neonatal Educ 2001;84:F23–5. Doi: 10.1136/fn.84.1.f23
3. Anderson JW, Johnstone BM, Remley DT. Breast-feeding and cognitive development: a meta-analysis. Am J Clin Nutr 1999;70:525–35. Doi: 10.1093/ajcn/70.4.525
4. Fonseca ALM, Albernaz EP, Kaufmann CC, Neves IH, De Figueiredo VLM. Impact of breastfeeding on the intelligence quotient of eight-year-old children. J Pediatr 2013;89:346–53. Doi: 10.1016/j.jped.2012.12.010
5. Belfort MB, Rifas-Shiman SL, Kleinman KP, Guthrie LB, Bellinger DC, Taveras EM, et al. Infant feeding and childhood cognition at ages 3 and 7 years: effects of breastfeeding duration and exclusivity. JAMA Pediatr 2013;167:836–44. Doi: 10.1001/jamapediatrics.2013.455
6. Guxens M, Mendez MA, Molto-Puigmarti C, Julvez J, Garcia-Esteban R, Forns J, et al. Breastfeeding, long-chain polyunsaturated fatty acids in colostrum, and infant mental development. Pediatrics 2011;128:e880–9. Doi: 10.1542/peds.2010-1633
7. Kim JI, Kim B-N, Kim J-W, Hong S-B, Shin M-S, Yoo HJ, et al. Breastfeeding is associated with enhanced learning abilities in school-aged children. Child Adolesc Psychiatry Ment Health 2017;11:36. Doi: 10.1186/s13034-017-0169-0
8. Mortensen EL, Michaelsen KF, Sanders SA, Reinisch JM. The association between duration of breastfeeding and adult intelligence. JAMA 2002;287:2365–71. Doi: 10.1001/jama.287.18.2365
9. Horwood LJ, Fergusson DM. Breastfeeding and later cognitive and academic outcomes. Pediatrics 1998;101:E9. Doi: 10.1542/peds.101.1.e9
10. Victora CG, Horta BL, de Mola CL, Quevedo L, Pinheiro RT, Gigante DP, et al. Association between breastfeeding and intelligence, educational attainment, and income at 30 years of age: a prospective birth cohort study from Brazil. Lancet Glob Heal 2015;3:e199–205. Doi: 10.1016/S2214-109X(15)70002-1
11. Huang J, Peters KE, Vaughn MG, Witko C. Breastfeeding and trajectories of children’s cognitive development. Dev Sci 2014;17:452–61. Doi: 10.1111/desc.12136
12. Camargo-Figuera FA, Barros AJ, Santos IS, Matijasevich A, Barros FC. Early life determinants of low IQ at age 6 in children from the 2004 Pelotas Birth Cohort: a predictive approach. BMC Pediatr 2014;14:308. Doi: 10.1186/s12887-014-0308-1
13. Park S, Kim B-N, Kim J-W, Shin M-S, Yoo HJ, Cho S-C. Protective effect of breastfeeding with regard to children’s behavioral and cognitive problems. Nutr J 2014;13:111. Doi: 10.1186/1475-2891-13-111
14. Tseng P-T, Yen C-F, Chen Y-W, Stubbs B, Carvalho AF, Whiteley P, et al. Maternal breastfeeding and attention-deficit/hyperactivity disorder in children: a meta-analysis. Eur Child Adolesc Psychiatry 2019;28:19–30. Doi: 10.1007/s00787-018-1182-4
15. Deoni SCL, Dean DC, Piryatinsky I, O’Muircheartaigh J, Waskiewicz N, Lehman K, et al. Breastfeeding and early white matter development: a cross-sectional study. Neuroimage 2013;82:77–86. Doi: 10.1016/j.neuroimage.2013.05.090
16. Ballard O, Morrow AL. Human milk composition. Pediatr Clin North Am 2013;60:49–74. Doi: 10.1016/j.pcl.2012.10.002
17. Dean DC, O’Muircheartaigh J, Dirks H, Travers BG, Adluru N, Alexander AL, et al. Mapping an index of the myelin g-ratio in infants using magnetic resonance imaging. Neuroimage 2016;132:225–37. Doi: 10.1016/j.neuroimage.2016.02.040
18. Barkovich AJ, Kjos BO, Jackson DE, Norman D. Normal maturation of the neonatal and infant brain: MR imaging at 1.5 T. Radiology 1988;166:173–80. Doi: 10.1148/radiology.166.1.3336675
19. Dai X, Hadjipantelis P, Wang J, Deoni SCL, Müller H. Longitudinal associations between white matter maturation and cognitive development across early childhood. Hum Brain Mapp 2019;40:4130–45. Doi: 10.1002/hbm.24690
20. Nagy Z, Westerberg H, Klingberg T. Maturation of white matter is associated with the development of cognitive functions during childhood. J Cogn Neurosci 2004;16:1227–33. Doi: 10.1162/0898929041920441
21. Soldán M, Pirko I. Biogenesis and significance of central nervous system myelin. Semin Neurol 2012;32:9–14. Doi: 10.1055/s-0032-1306381
22. Casey BM, Thom EA, Peaceman AM, Varner MW, Sorokin Y, Hirtz DG, et al. Treatment of subclinical hypothyroidism or hypothyroxinemia in pregnancy. N Engl J Med 2017;376:815–25. Doi: 0.1056/NEJMoa1606205
23. Alexander G, Himes J, Kaufman R, Mor J, Kogan M. A United States national reference for fetal growth. Obstet Gynecol 1996;87:163–8. Doi: 10.1016/j.ajog.2006.03.069
24. Beaino G, Khoshnood B, Kaminski M, Marret S, Pierrat V, Vieux R, et al. Predictors of the risk of cognitive deficiency in very preterm infants: the EPIPAGE prospective cohort. Acta Paediatr 2011;100:370–8. Doi: 10.1111/j.1651-2227.2010.02064.x
25. Rozé J-C, Darmaun D, Boquien C-Y, Flamant C, Picaud J-C, Savagner C, et al. The apparent breastfeeding paradox in very preterm infants: relationship between breast feeding, early weight gain and neurodevelopment based on results from two cohorts, EPIPAGE and LIFT. BMJ Open 2012;2:e000834. Doi: 10.1136/bmjopen-2012-000834