A cohort study on full breastfeeding and child neuropsychological development: the role of maternal social, psychological, and nutritional factors

Authors

  • Jordi Julvez,

    Corresponding author
    1. Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain
    2. Hospital del Mar Research Institute (IMIM), Barcelona, Spain
    3. CIBER Epidemiologia y Salud Pública (CIBERESP), Barcelona, Spain
    • Correspondence to Dr Jordi Julvez, Centre for Research in Environmental Epidemiology – PRBB, C. Doctor Aiguader 88, 08003 Barcelona, Spain. E-mail: jjulvez@creal.cat

    Search for more papers by this author
  • Monica Guxens,

    1. Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain
    2. Hospital del Mar Research Institute (IMIM), Barcelona, Spain
    3. CIBER Epidemiologia y Salud Pública (CIBERESP), Barcelona, Spain
    4. Division of Environmental Epidemiology, Institute for Risk Assessment Sciences, Utrecht University, Utrecht, the Netherlands
    Search for more papers by this author
  • Anne-Elie Carsin,

    1. Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain
    2. Hospital del Mar Research Institute (IMIM), Barcelona, Spain
    Search for more papers by this author
  • Joan Forns,

    1. Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain
    2. Hospital del Mar Research Institute (IMIM), Barcelona, Spain
    Search for more papers by this author
  • Michelle Mendez,

    1. Department of Nutrition and Carolina Population Center, University of North Carolina, Chapel Hill, NC, USA
    Search for more papers by this author
  • Michelle C Turner,

    1. Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain
    2. CIBER Epidemiologia y Salud Pública (CIBERESP), Barcelona, Spain
    3. McLaughlin Centre for Population Health Risk Assessment, Institute of Population Health, University of Ottawa, Ottawa, ON, Canada
    Search for more papers by this author
  • Jordi Sunyer

    1. Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain
    2. Hospital del Mar Research Institute (IMIM), Barcelona, Spain
    3. CIBER Epidemiologia y Salud Pública (CIBERESP), Barcelona, Spain
    4. Pompeu Fabra University, Barcelona, Spain
    Search for more papers by this author

Abstract

Aim

This study investigated whether duration of full breastfeeding is associated with child neuropsychological development and whether this association is explained by social, psychological, and nutritional factors within families.

Method

Participants in this study were a population-based birth cohort in the city of Sabadell (Catalonia, Spain). Females were recruited during the first trimester of pregnancy between July 2004 and July 2006. Information about parental characteristics and breastfeeding was obtained through questionnaires. Full breastfeeding was categorized as never, short term (≤4mo), long term (4–6mo), or very long term (>6mo). A trained psychologist assessed the neuropsychological development of children at 4 years of age (n=434) using the McCarthy Scales of Children's Abilities (MSCA).

Results

Full breastfeeding showed an independent association with child general MSCA scores after adjusting for a range of social, psychological, and nutritional factors (>6mo, coefficient=7.4 [95% confidence interval=2.8–12.0], p=0.011). Maternal social class, education level, and IQ were also associated with child neuropsychological scores, but did not explain breastfeeding associations. Omega-3 (n3) fatty acid levels were not associated with child neuropsychological scores.

Interpretation

Very long-term full breastfeeding was independently associated with neuropsychological functions of children at 4 years of age. Maternal indicators of intelligence, psychopathology, and colostrum n3 fatty acids did not explain this association.

Abbreviations
INMA

Infancia y Medio Ambiente (Environment and Childhood) project

MSCA

McCarthy Scales of Children's Abilities

n3

Omega-3

n6

Omega-6

SCL-90-R

Symptom Checklist, Revised

The human brain is highly sensitive to environmental exposures occurring during particular periods of vulnerability.[1, 2] In early life, the brain's developmental biological activity is so predominant that any factor enhancing or disturbing this process could result in permanent effects on cerebral function.[3]

A wide range of environmental determinants, including physical, biological, psychological, and social factors, modulate brain structure and function by interacting with genes and expression mechanisms (i.e. epigenetic determinants).[2, 4-6] Long-term breastfeeding, particularly full breastfeeding, is one of the most studied factors of neurodevelopment in recent years, with several studies reporting beneficial effects on child neuropsychological development.[7-17] However, the difficulty with breastfeeding research is being able to disentangle the underlying casual pathways of this association.

Research into breastfeeding and breastfeeding behaviour has thus far focused on mother-to-child transfer of physical and biological components, such as pollutants and essential nutrients;[18, 19] the psychological effects from long-term skin-to-skin contact;[20, 21] and the possibility that breastfeeding could be a surrogate for advantageous psychosocial factors, such as maternal IQ, level of education, or social class.[8, 17] Some studies have explored the influence of some of these factors on the breastfeeding–neurodevelopment association.[22] One previous cohort study that investigated the role of an extended range of cofactors was that of the Infancia y Medio Ambiente (Environment and Childhood, INMA) project.[18] In this study, Guxens et al.[18] concluded that duration of breastfeeding showed an independent association with 14-month mental development which was only partly explained by nutritional factors (omega-3 [n3] fatty acid in colostrum).

The aim of the present study was to further investigate whether full-breastfeeding duration was associated with child neurodevelopment after a wide range of social, psychological, and nutritional factors within families were controlled for in an extended follow-up of the INMA project. We investigated detailed neuropsychological outcomes in children at 4 years of age. Extended psychometric assessments of both parents were also compiled.

Method

Design and study participants

A population-based birth cohort was identified in the city of Sabadell, Catalonia, Spain as part of the INMA project.[23] A detailed description is provided elsewhere.[18, 23] In brief, between July 2004 and July 2006, a total of 657 pregnant women who visited the public health centre of Sabadell for an ultrasound scan in their first trimester were recruited, providing they fulfilled the inclusion criteria (aged >16y, intention to deliver at the reference hospital, no problems of communication, a singleton pregnancy, and no assisted conception). A total of 99.5% of Spaniards have public health insurance, and an estimated 70% to 90% of women use public health services during pregnancy.[18] The participation rate was 60% (n=434). Children were followed up at 4 years of age for neuropsychological testing. Informed consent was obtained from all participants and the study was approved by the clinical research ethics committee of the Municipal Institute of Health Care, Barcelona, Spain.

Neuropsychological testing

A Spanish standardized version of the McCarthy Scales of Children's Abilities (MSCA)[24] was used to evaluate neuropsychological development of children at 4 years of age. The MSCA includes six conventional scales: general cognitive, verbal, quantitative, memory, perceptive-performance, and motor functions. All testing was performed in the health care centre by one specially trained psychologist (MF). The psychologist was blinded to any exposure information of the child, including full breastfeeding. The psychologist excluded from the data analyses children with neuropsychological tests of uncertain quality (n=12). Other factors of the children provided by the mothers were taken into consideration in the analyses, such as the child feeling ill during examination (n=29), the child not sleeping well the night before (n=39), recent mood changes of the child (n=17), the child being a regular patient of a psychologist (n=48), or the child having a diagnosis of a neuropsychological disorder (n=27). However, results did not change upon their exclusion (data not shown) and they are included in the final regression models here as a binary covariate. Raw MSCA scores were normalized to a mean of 100 (SD 15). This was to obtain indexes in accordance with a local normative sample and to avoid the use of US norms provided in the manual. A new outcome scale for executive function was also created based on half of the MSCA subtests.[25] Cronbach's alpha coefficients for internal consistency for all MSCA scales were found to be acceptable (≥0.65), with the exception of the motor scale (0.45).

Duration of full-breastfeeding

Detailed information about each child's feeding was completed by interviewer-administered questionnaires with mothers when the children were 6 months and 14 months of age. Full breastfeeding was defined as receiving breast milk only, but allowing supplementations of non-milk liquids (e.g. water or water-based drinks such as sweetened and flavoured water, teas, infusions), fruit juice, oral rehydration salts solution, drops, and syrup forms of vitamins, minerals, and medicines.[18, 26] We assumed that some of these supplementations are included in the World Health Organization category of ‘predominant milk full breastfeeding’ (which includes both exclusive breastfeeding and predominant breastfeeding). The duration of full breastfeeding was categorized into four groups: children who were partially breastfed (reference group, a combination of breast milk, formula, and other milk-based feeds), short-term full breastfeeding (≤4mo, as Spanish law provides for maternity leave of up to 16wks), long-term full breastfeeding (4–6mo, according to the former World Health Organization recommendations), and very-long-term full breastfeeding (>6mo, according to the new World Health Organization recommendations).[18] Women who did not do any breastfeeding were excluded from the main analysis (n=33) as the aim of this study was to assess the association of full-breastfeeding duration with important factors such n3 fatty acids in colostrum.[18]

n3 fatty acids in colostrum

Colostrum was collected in the first 48 to 96 hours after birth at the hospital in a random subsample of participants (n=277) by an experienced nurse. The sample was collected in the morning, at the end of the feeding, in sterile polypropylene tubes by mechanical expression of one breast using a breast pump. Milk was transported to the laboratory in ice boxes less than 2 hours after collection, where samples were stored at −80°C until analysis. Thirty-eight fatty acids were identified and quantified, including the fatty acids of interest for the present study (total n3/omega-6 [n6] polyunsaturated fatty acid ratio). Detailed information on fatty acid measurements is published elsewhere.[18]

Other parental variables

Information on parental education (in years), social class (using the UK Registrar General's 1990 classification according to current International Standard Classification of Occupations parental occupation), and other clinical and socio-demographic characteristics (such as gestational age, birthweight, mother's occupation [unemployment, part-time job, parental leave, etc.], parents living together or not, etc.), were obtained through questionnaires during the first and third trimesters of pregnancy, when the child was 14 months of age, and when the child was 4 years of age. In relation to social class and level of education, we used the data from the third trimester of pregnancy. Detailed information is described elsewhere.[18] Parental psychometric data collected at 14 months of age of the child[18] included factor ‘G’ of the Cattell and Cattell test,[27] the 12-item General Health Questionnaire,[28] and the parental–child attachment Condon questionnaires.[29] At the 4-year follow-up, parental psychopathological symptoms were assessed using the Symptom Checklist, Revised (SCL-90-R).[30] This is a self-reported questionnaire widely used in both typical and distressed populations. Ninety items are classified into nine domains and a general score: somatization, obsessive–compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism. A general score SCL-90-R alpha coefficient of 0.97 was obtained for mothers and an alpha coefficient of 0.96 was obtained for fathers. Additionally, the similarities subtest of the Wechsler Adult Intelligence Scales, Third Edition,[31] was used as a proxy for maternal Verbal IQ.

Environmental contaminants

Levels of organochlorine compounds (dichlorodiphenyl dichloroethylene and polychlorinated biphenyls) were analysed in maternal serum extracted in the first trimester of pregnancy.[32] Total mercury levels were analysed in cord blood using thermal decomposition, amalgamation, and atomic absorption spectrometry.[33] Gas stove use at home during pregnancy was reported through questionnaires.[18]

Statistical analysis

Only participants for whom data on neuropsychological development and full-breastfeeding were complete were included in the analysis. Differences in some socio-demographic characteristics were found between the children included (n=434) and not included (n=223) in this study. Compared with children included in the study, children not included had lower maternal social class (IV and V, 66% vs 44%), lower maternal education (basic level, 40% vs 23%), higher maternal smoking use (21% vs 13%), and lower mean birthweight (3166 vs 3261g).[18] Among the included participants (n=434), multiple imputation of missing values for the remaining variables was performed using chained equations in which 100 completed datasets were generated and analysed using the standard combination rules for multiple imputations (Tables SI–SIII, online supporting information).[18]

Since duration of full breastfeeding did not show a linear association with child neuropsychological development in generalized additive models, all analyses presented here were conducted using the full-breastfeeding duration categories described above. Multivariable linear regression models were used to examine the relationship between full-breastfeeding duration and child neuropsychological development at 4 years of age. All models included child age (in years) during examination and the quality of the test performance flagged by the psychologist (good vs not-so-good) as mandatory covariables. Furthermore, different groups of covariables were sequentially included in regression models according to the following order: (1) basic socio-demographic factors (child sex, parental age, parity, maternal alcohol consumption and smoking during pregnancy, child day care attendance at 14mo of age, and maternal country of birth); (2) maternal social class and education; (3) maternal IQ; (4) maternal attachment to the child, mental health, and psychopathological symptoms; and (5) pollutant exposure during pregnancy.[18] The final fully adjusted model included all covariables simultaneously (1, 2, 3, 4, and 5). Collinearity between variables in the final fully adjusted model was assessed according to the variance inflation factor test.[34] Finally, to assess the relationship between colostrum n3 fatty acid levels (in tertiles) and child neuropsychological development, the same fully adjusted models were analysed including n3 fatty acid levels in colostrum (total n3/n6) as the exposure variable. Statistical significance was set at α<0.10 (two-tailed). Statistical analyses were conducted using stata 12.0 (Stata Corp, College Station, TX, USA).

Results

Twenty-three per cent of mothers had a bachelor's degree and were from social classes I and II. The average maternal age at delivery was 32 years, and 13% of mothers (n=58) smoked during pregnancy. Fifteen per cent of all mothers who breastfed their infants (n=434) did not fully breastfeed. The median duration of full breastfeeding was 17 weeks. Normalized median general and specific MSCA scores tended to increase with increasing category of full-breastfeeding duration, particularly in the very-long-term full-breastfeeding duration group (Table 1).

Table 1. Child neuropsychological outcomes (median MSCA scores) of children at 4 years of age by duration of full breastfeeding (n=434)
Child outcomesAll participants, median (range)Full breastfeeding p a
Never (n=64), medianShort term (≤4mo, n=136), medianLong term (4–6mo, n=183), medianVery long term (>6mo, n=51), median
  1. a

    p-value for trend using crude linear regression models. MSCA, McCarthy Scales of Children's Abilities.

General101 (53–147)981011011080.002
Verbal102 (54–138)971021021050.006
Perceptive-performance100 (54–164)961001001090.016
Memory101 (66–135)97102991060.023
Quantitative100 (64–141)951001001070.012
Motor102 (54–147)1001001001080.028
Executive function102 (56–144)991001021080.002

Maternal SCL-90-R scores tended to suggest fewer psychopathological symptoms among mothers in the longer duration full-breastfeeding group, particularly for the somatization, anxiety, and phobic anxiety subscales. These findings were not observed for paternal SCL-90-R scores. Maternal IQ (verbal and perceptive-performance functions) was also higher among those with longer full-breastfeeding duration. Parental attachment scores showed no difference across full-breastfeeding duration categories (Table 2).

Table 2. Parental psychometric characteristics by duration of full breastfeeding
Psychometric characteristic n All participants, median (range)Full breastfeeding p a
Never (n=64), medianShort term (≤4mo, n=136), medianLong term (4–6mo, n=183), medianVery long term (>6mo, n=51), median
  1. a

    p-value of trend test among ordered groups. SCL-90-R, Symptom Checklist, Revised; GHQ, General Health Questionnaire.

Mother
Psychopathological symptoms (SCL-90-R), 4y421      
General score 48 (36–100)494848450.111
Somatizations 48 (36–92)485047470.032
Obsessions 48 (37–93)485048450.417
Interpersonal sensitivity 48 (38–91)484848450.400
Depression 48 (36–91)484848460.230
Anxiety 47 (37–100)504947450.057
Phobic anxiety 45 (45–128)474545450.086
Hostility 47 (38–85)504747470.251
Paranoid ideation 48 (40–84)484848450.317
Psychoticism 46 (42–107)484646460.301
Verbal IQ, 4y42410 (3–21)9910100.028
Perceptive-performance IQ, 14mo419102 (54–136)961041021040.057
Mental health (12-GHQ), 14mo41410 (1–27)10101090.143
Attachment (Condon), 14mo41559 (31–70)575859590.663
Father
Psychopathological symptoms (SCL-90-R), general score, 4y39448 (35–91)494948480.857
Mental health (12-GHQ), 14mo3969 (0–25)99890.253
Attachment (Condon), 14mo40950 (33–60)505050490.187

Table 3 presents associations between full-breastfeeding duration and MCSA scores from multivariate linear regression models. Results from minimally adjusted models and fully adjusted models were similar, with children in the very long-term full-breastfeeding duration group demonstrating higher general and specific MCSA scores (ranging approximately from 5 to 9 points greater on average). The inclusion of any set of covariables had a negligible impact on regression coefficients. There was also a clear positive trend between categories of full-breastfeeding duration and executive function. The variance inflation factor test did not indicate evidence for high amounts of multicollinearity between variables in the final fully adjusted model, as none of the values were higher than 5.

Table 3. Adjusted models of duration of full breastfeeding and child neuropsychological development (MSCA scores) of children at 4 years of age
Child outcomes (n=434)Full breastfeedinga p
Short term (≤4mo, n=136), coefficient (95% CI)Long term (4–6mo, n=183), coefficient (95% CI)Very long term (>6mo, n=51), coefficient (95% CI)
  1. aReference group comprised children never receiving full breastfeeding (n=64). Adjusting for paternal characteristics shown in Table 2, social class and education level did not change the results (data not shown). bAdjusted for neuropsychological test conditions and child age. cAdditionally adjusted for child sex, parental age, parity, alcohol consumption and smoking during pregnancy (both as binary variables), child day care attendance at 14 months of age, and maternal country of birth. dAdditionally adjusted for maternal education level, social class, mental health, attachment, and perceptive-performance IQ at the age of 14 months; maternal psychopathological symptoms and Verbal IQ at the age of 4 years; and pollutant exposures during pregnancy (dichlorodiphenyl dichloroethylene, polychlorinated biphenyls, and mercury levels, as well as using cooking gas stove). *p<0.10; **p≤0.05. CI, confidence interval; SC, social class; ED, education level; MH mental health.

General (minimally adjusted)b3.6 (−0.2–7.5)*3.6 (0.0–7.3)**8.5 (3.7–13.3)**0.003
General (adjusted)c3.3 (−0.4–7.1)*4.0 (0.4–7.5)**9.0 (4.3–13.6)**0.001
General (adjustedc+SC+ED)2.8 (−0.8–6.5)2.8 (−0.7–6.3)7.6 (3.1–12.2)**0.005
General (adjustedc+SC+ED+IQs)2.9 (−0.6–6.6)2.7 (−0.7–6.2)7.3 (2.8–11.8)**0.008
General (adjustedc+SC+ED+IQs+MH)3.0 (−0.6–6.6)2.7 (−0.8–6.1)7.5 (2.9–12.0)**0.008
General (adjustedc+SC+ED+IQs+MH+pollutants: full model)d3.1 (−0.6–6.7)*2.7 (−0.8–6.2)7.5 (2.9–12.1)**0.009
Verbal (full model)d3.4 (−0.7–7.4)2.9 (−0.9–6.8)6.8 (1.8–11.9)**0.030
Perceptive-performance (full model)d1.6 (−2.3–5.5)1.9 (−1.8–5.6)4.9 (0.3–9.8)**0.075
Memory (full model)d1.4 (−2.8–5.6)0.7 (−3.3–4.8)5.8 (0.5–11.1)**0.125
Quantitative (full model)d2.4 (−1.6–6.4)1.4 (−2.5–5.3)6.9 (1.8–12.0)**0.059
Motor (full model)d1.2 (−2.8–5.3)1.8 (−2.1–5.7)5.3 (0.3–10.4)**0.058
Executive function (full model)d2.3 (−1.5–6.1)3.1 (−0.5–6.8)*6.9 (2.1–11.7)**0.007

Table 4 presents associations between selected maternal characteristics and child MSCA outcomes in multivariate regression models. Maternal social class, level of education, and IQ, particularly Verbal IQ, were strongly associated with child neuropsychological development, whereas mental health, psychopathological symptoms, and attachment scores showed weaker, or no, associations. However, coefficients for associations observed attenuated with adjustment for the full range of maternal social and psychological factors.

Table 4. Adjusted models of maternal characteristics and child neuropsychological development (MSCA scores) at 4 years of age
Determinants (n=434)General cognitiveExecutive function
Exposurea (simply adjusted), coefficient (95% CI)Exposureb (fully adjusted), coefficient (95% CI)Exposurea (simply adjusted), coefficient (95% CI)Exposureb (fully adjusted), coefficient (95% CI)
  1. aAdjusted for neuropsychological test conditions; child age and sex; parental age, parity, and alcohol consumption and smoking during pregnancy; child day care attendance at 14 months of age; and maternal country of birth. bAdditionally adjusted for maternal education level, social class, mental health, attachment, and perceptive-performance IQ at the age of 14 months; maternal psychopathological symptoms and Verbal IQ at the age of 4 years; and pollutant exposures during pregnancy (dichlorodiphenyl dichloroethylene, polychlorinated biphenyls, and mercury levels and using cooking gas stove). cMaternal education level was excluded from the fully adjusted models. dMaternal social class was excluded from the fully adjusted models. eIQ scores were not mutually adjusted in the fully adjusted models. fMental health, psychopathological, and attachment indicators were not mutually adjusted in the fully adjusted models. None of the pollutant exposure assessed during pregnancy was associated with MSCA, McCarthy Scales of Children's Abilities (MCSA) outcomes (data not shown). *p<0.10; **p≤0.05. CI, confidence interval; SCL-90-R, Symptom Checklist, Revised; GHQ, General Health Questionnaire.

Social class (reference=I+II, 24%)c
III (32%)−3.4 (−6.5 to −0.3)**−2.6 (−5.8–0.6)−3.0 (−6.3–0.2)*−2.3 (−5.6–1.0)
IV+V (44%)−7.4 (−10.5 to −4.2)**−4.6 (−8.0 to −1.1)**−7.2 (−10.4 to −3.9)**−4.7 (−8.4 to −1.1)**
Education level (reference=primary, 23%)d
Secondary (43%)5.8 (2.8–8.8)**3.8 (0.6–7.0)**5.0 (1.9–8.2)**3.1 (−0.2–6.5)*
University (34%)8.3 (4.9–11.7)**4.8 (0.8–8.7)**7.4 (3.9–10.9)**4.1 (−0.1–8.2)*
IQe
Perceptive-performance, 14mo (10-point increase)1.7 (0.8–2.6)**0.7 (−0.3–1.8)1.5 (0.6–2.5)**0.6 (−0.4–1.7)
Verbal, 4y1.5 (0.9–2.1)**1.0 (0.3–1.6)**1.3 (0.7–1.9)**0.9 (0.2–1.6)**
Mental health, attachment, and psychopathologyf
12-GHQ, 14mo−0.0 (−0.3–0.3)−0.0 (−0.3–0.2)−0.0 (−0.4–0.2)−0.1 (−0.4–0.2)
Attachment, 14mo0.0 (−0.2–0.2)0.0 (−0.2–0.2)0.0 (−0.2–0.3)0.1 (−0.2–0.3)
SCL-90-R, 4y (>65 points [8%])−2.4 (−6.8–2.1)−1.7 (−6.1–2.6)−3.4 (−8.0–1.1)−3.0 (−7.5–1.6)

Positive non-significant associations were observed between n3 fatty acid levels in colostrum and child general and executive function MSCA scores (Table 5). Associations between full-breastfeeding duration and child MSCA scores were unchanged after adjustment for n3 fatty acid levels. Similarly, associations between n3 fatty acid levels and child MSCA scores were unaffected with adjustment for full-breastfeeding duration.

Table 5. Adjusted modelsa of duration of full breastfeeding and n3 fatty acid levels in colostrum and child neuropsychological development (MSCA scores) at age 4 years
DeterminantsGeneral cognitiveExecutive function
Simple exposure,b coefficient (95% CI)Multiple exposure,c coefficient (95% CI)Simple exposure,b coefficient (95% CI)Multiple exposure,c coefficient (95% CI)
  1. aAdjusted for neuropsychological test conditions; child age and sex; parental age, parity, and alcohol consumption and smoking during pregnancy; child day care attendance at 14 months of age; maternal country of birth; maternal education level, social class, mental health, attachment, and perceptive-performance IQ at the age of 14 months; maternal psychopathological symptoms and Verbal IQ at the age of 4 years; and pollutant exposures during pregnancy (dichlorodiphenyl dichloroethylene, polychlorinated biphenyls, and mercury levels and using cooking gas stove). bFull breastfeeding and fatty acids were included in separated models. cFull breastfeeding and fatty acids were included in the same multivariate models. dThe reference group comprised children never received full breastfeeding (n=64). eThe reference group comprised children with low levels of fatty acids (tertile 1). *p<0.10; **p≤0.05. MSCA, McCarthy Scales of Children's Abilities.

Full breastfeedingd
Short term, <4mo (n=136)3.1 (−0.6–6.7)*3.1 (−0.5–6.8)*2.3 (−1.5–6.1)2.3 (−1.5–6.2)
Long term, 4–6mo (n=183)2.7 (−0.8–6.2)2.7 (−0.8–6.3)3.1 (−0.5–6.8)*3.1 (−0.5–6.8)*
Very long term, >6mo (n=51)7.5 (2.9–12.1)**7.4 (2.8–12.0)**6.9 (2.1–11.7)**6.8 (2.0–11.6)**
p 0.0090.0110.0070.009
n3 fatty acids (n=434)e
Tertile 20.5 (−2.9–3.9)0.4 (−3.0–3.8)−0.6 (−4.2–3.0)−0.6 (−4.2–3.0)
Tertile 31.8 (−1.7–5.4)1.5 (−2.0–5.1)1.4 (−2.2–5.1)1.1 (−2.5–4.8)
p 0.3190.3860.4380.521

No interaction between full-breastfeeding duration and n3 fatty acid levels for any neuropsychological outcome was observed (data not shown). Similarly, there was no interaction between full-breastfeeding duration and maternal smoking habits during pregnancy (data not shown). Results were similar in complete case analyses, in which we repeated all the models shown in Tables 3 and 5 (Tables SIV–SVIII, online supporting information). The additional inclusion of the mothers who never breastfed their children (n=33) into the full-breastfeeding reference group did not change the main results of Table 3 (Table SVI). Repeating the final models and additionally adjusting for type of delivery, child gestational age, and birthweight did not meaningfully change results (Table SIX, online supporting information).

We also examined the association between full-breastfeeding duration and maternal Verbal IQ to assess whether this association was explained by maternal education and social class. Full-breastfeeding duration was associated with maternal Verbal IQ only when maternal education and social class were excluded from the model (data not shown).

Discussion

In this population-based birth cohort study from a working-class suburb of Barcelona, neuropsychological functions of children at 4 years of age, particularly executive functioning, were positively associated with full-breastfeeding duration after adjustment for maternal psychological factors, socio-demographic indicators, and environmental pollution exposures during pregnancy. Mother–child pairs with very long-term full-breastfeeding duration (>6mo) presented the best psychometric profiles. Maternal social class, level of education, and IQ were also strongly associated with neurodevelopment, as expected. Maternal psychopathological symptoms and n3 fatty acid levels in colostrum were poorly related to neurodevelopment and did not mediate full-breastfeeding duration and outcome associations.

There are numerous publications describing whether breastfeeding is associated with child neurodevelopment, many of which reported a positive association.[7, 9, 35-37] However, several of these studies have been based on small sample sizes and specific populations. In addition, breastfeeding variables were sometimes retrospectively recorded, and a number of studies have not assessed breastfeeding duration, nor have they clearly defined the different modes of breastfeeding (i.e. full breastfeeding).[7-12, 15] Moreover, in a number of newer reports, which assessed the potential role of maternal IQ, parenting skills, and mental health, findings were less consistent.[8, 18, 35-37] Der et al.,[8] in a meta-analysis of longitudinal datasets of 5475 children in which breastfeeding was defined in both ways, as a dichotomous variable (yes, no) and as months of duration, concluded that the beneficial association of breastfeeding was confounded by maternal IQ. In contrast, the largest controlled trial (n=17 046 healthy breastfeeding infants) showed beneficial effects on neuropsychological development of children of 6 years of age in the prolonged and exclusive breastfeeding group.[14] A few prospective studies have assessed the potential beneficial role of n3 fatty acid levels in colostrum and breast milk jointly with breastfeeding duration, with conflicting results.[18, 38] There are also concerns regarding whether breast milk may transfer neurotoxic pollutants from mother to child; however, recent findings showed that the positive effects of breastfeeding counterbalanced any detrimental effect of pollutant toxicity,[18, 39] but further consideration is needed. The present cohort study sought to address the methodological gaps noted here and in several reviews.[9, 11, 12, 19] After accounting for a wide array of potentially important covariates, the findings suggested an independent association between full-breastfeeding duration and improved neuropsychological performance of the child.

Several studies have reported that mothers with advantageous socio-demographical and psychosocial characteristics, such as higher socio-economic status, higher level of education, and higher IQ, tend to breastfeed for longer periods, which is likely to be the result of the fact that these mothers are more aware of the potential beneficial effects of breastfeeding.[22] But some reports also noted confounding by such factors, particularly maternal IQ, in associations between breastfeeding and child neurodevelopment.[8, 40] Our study observed higher IQ scores among mothers who breastfed fully for longer periods of time. However, after controlling for maternal IQ, as well as for other socio-demographic factors, the strong positive associations observed with child neuropsychological development of children at 4 years of age remained. Nevertheless, these maternal characteristics were also independent determinants of neurodevelopment.

Mothers receive physical stimulation during breastfeeding, which promotes higher levels of oxytocin release in the blood and milk. If this process is repeated over a long period it can benefit the mother's mood, and may, indirectly, affect the child's neurodevelopment.[20, 21, 41] In the present study, mothers who fully breastfed for long periods, particularly for more than 6 months, tended to show better scores in anxiety and other neurotic traits 4 years later. However, this finding may also be due, in part, to reverse causation, as the mothers with higher neurotic traits are the ones who breastfed for shorter periods. Full-breastfeeding–outcome associations were unchanged after controlling for maternal psychological factors. In support of this psychological attachment hypothesis, there are several studies that link maternal postpartum depression to impaired child neurodevelopment, but further population-based studies are required.[42] In our study, higher scores of maternal psychopathological symptoms were associated with lower child performance in executive functions, but studies with larger sample sizes are needed to confirm this finding.

A number of experimental studies demonstrated that n3 fatty acids are structural lipids critical for retina and cortical brain development, neuroprotection, and neurotransmission.[19, 43] Several controlled trials using supplemented formulas showed positive associations between n3 fatty acids and visual acuity and general mental development. However, most of these studies were based on preterm infants and did not focus on the factors related to the choice between breastfeeding or feeding with formula.[38, 44] Two previous studies based on the present sample at 14-months’ follow-up found that colostrum n3 fatty acid levels were associated with both cognition and genetic variants of the FADS gene cluster and the ELOVL gene family, which interacted with n3 fatty acid levels for infant cognition.[18, 45] However, another study found no association between n3 fatty acid intake of children in the first 4 months of age and neuropsychological development.[38] The present work could not confirm the former finding in relation to n3 fatty acid levels and neuropsychological development, and we did not observe any interaction between fatty acid levels and full-breastfeeding duration on cognition of children at 4 years of age.[18] Additionally, coefficients in full-breastfeeding duration–outcome models did not change after including the colostrum n3 fatty acid as a confounder–mediator. The null association found in the present study at older ages could be partially explained by the influence of unmeasured socio-environmental factors that occurred during the period between the two outcomes. Moreover, in a recent study,[46] colostrum n3 fatty acid levels tended to be higher in full-term children than in preterm children and transitional and mature breast milk content was lower in n3 fatty acids.[46] The potential higher levels here could also explain the undetected beneficial effects seen in this study.[46]

This cohort study is one of the most comprehensive in terms of the availability of potentially relevant covariates for addressing this topic, with detailed assessments of factors related to the main breastfeeding hypotheses of underlying casual pathways. The external validity of the results may be important, since the data are based on a population cohort; however, we cannot rule out some selection bias effect owing to the participants lost at follow-up. Additionally, we analysed whether there was confounding by maternal socio-demographic characteristics (social class, level of education, and IQ) and whether there was mediation by psychological and nutritional mechanisms probably enhanced during long-term full breastfeeding, and we assessed the uncertain role of environmental pollutants.[17, 18] Both child and parental psychological assessments were performed by one trained psychologist who followed standardized and strict protocols and the psychological data showed good psychometric characteristics. However, maternal Verbal IQ was a proxy measure here. The similarities subtest of the Wechsler Adult Intelligence Scales, Third Edition, might be a weaker measure than the full verbal scale; nevertheless, it correlates well with the full verbal scale and has been used in short forms.[47] With all these data to hand, the findings did not provide clear insights into cofactors underlying the full breastfeeding–child cognition associations. The complexity of breastfeeding probably goes beyond the indicators normally assessed in neuroepidemiology[2] and there might be multiple, multilevel, unmeasured pathways interacting with each other. However, several weaknesses need to be pointed out. Potential misclassification could affect trends across the four classes of full breastfeeding, since children who were partially breastfed for a long period were included in the reference group. The use of colostrum n3 fatty acids should not be treated as a proxy of breast milk fatty acid, since fatty acid percentages have been reported to vary during the course of lactation, showing higher levels in colostrum and progressively decreasing levels throughout transitional and mature milk.[46] In addition, we did not measure n3 fatty acid levels in colostrum for all mother–child pairs. Nevertheless, multiple imputations of missing values allowed us to take into account the entire sample in the analyses. Another fact important to note is that the sample size of the very long-term full-breastfeeding group was relatively small and larger population-based birth cohort studies with detailed assessments are needed to confirm our findings, although such studies are costly to fund. Finally, the gut–brain axis hypothesis (the potential effect of microbiota provided by breast milk) is another biological mechanism not assessed here which could have contributed to the benefits of breastfeeding.[48]

Conclusion

This cohort study assessed whether there was an association between child neuropsychological development and prolonged full breastfeeding, as well as the extent to which this association was influenced by factors such as maternal level of education, IQ, mental health and psychopathology, colostrum n3 fatty acid levels, and pollutant biomarkers. Although several of these maternal characteristics were closely related to both full-breastfeeding duration and the outcome being studied, prolonged full breastfeeding demonstrated strong and independent associations with improved neuropsychological performance of children of 4 years of age, particularly executive functioning.

Acknowledgements

We acknowledge Silvia Fochs, Núria Pey, Muriel Ferrer, and all the study participants for their generous collaboration. We are also grateful to Carolina Moltó-Puigmartí for her contribution to the n3 fatty acid level measurements in colostrum. A full list of the INMA Sabadell study investigators can be found at www.proyectoinma.org/cohorts/sabadell/en_membres-sabadell.html.

This work was supported by grants from the Spanish Ministry of Health (FIS-PI041436); the Instituto de Salud Carlos III (Red INMA G03/176 and CB06/02/0041); the Spanish Ministry of Education and Science and the Spanish Ministry of Science and Innovation (AGL2008–04124/ALI and AGL2009–09730/ALI); JCI-2011–09771 – MICINN; the Generalitat de Catalunya-CIRIT (1999SGR 00241); the ‘Comissionat per a Universitats i Recerca del Departament d'Innovació, Universitats i Empresa de la Generalitat de Catalunya’; The EU sixth framework project NEWGENERIS (FP6–2003-Food-3-A-016320); and the Fundación Roger Torner. There is no potential conflict of interest, real or perceived.

Ancillary