Late weaning and maternal closeness, associated with advanced motor and visual maturation, reinforce autonomy in healthy, 2-year-old children

We studied neurodevelopmental outcomes and behaviours in healthy 2-year old children (N = 1306) from Brazil, India, Italy, Kenya and the UK participating in the INTERGROWTH-21st Project. There was a positive independent relationship of duration of exclusive breastfeeding (EBF) and age at weaning with gross motor development, vision and autonomic physical activities, most evident if children were exclusively breastfed for ≥7 months or weaned at ≥7 months. There was no association with cognition, language or behaviour. Children exclusively breastfed from birth to <5 months or weaned at >6 months had, in a dose-effect pattern, adjusting for confounding factors, higher scores for “emotional reactivity”. The positive effect of EBF and age at weaning on gross motor, running and climbing scores was strongest among children with the highest scores in maternal closeness proxy indicators. EBF, late weaning and maternal closeness, associated with advanced motor and vision maturation, independently influence autonomous behaviours in healthy children. INTERGROWTH-21st was a large, multicentre, population-based, research project conducted between 2009 and 2016, in eight delimited urban areas across five continents. The primary aim was to study growth, health, nutrition and development from early pregnancy to 2 years of age in mothers and children with adequate health, nutritional, environmental and socio-economic conditions at both individual and population levels10. FGLS, one of the main components of the INTERGROWTH-21st Project, included pregnant women from these eight populations, who met the individual entry criteria13,55–58. Children from the cohort of mothers enrolled during pregnancy in FGLS were followed up to 2 years of age (the Infant Follow-up Study (IFS)), and evaluated for growth, nutrition, health and developmental outcomes (the WHO gross motor milestones)11,59. All participants contributed data towards the construction of the international INTERGROWTH-21st Fetal Growth and Preterm Postnatal Growth Standards13,58. At the 2 year IFS visit, children completed a comprehensive neurodevelopment assessment in five of the eight original sites: the cities of Pelotas (Brazil); Turin (Italy); Oxford (UK); the central area of Nagpur (India) and the Parklands suburb of Nairobi (Kenya), using a set of tools specifically developed or selected for this purpose60. The sites in China, Oman and the USA did not participate in this phase because of local logistical and administrative reasons, all unrelated to the nature of the questions explored in the follow-up study. Across all study sites, we implemented standardised clinical care and infant feeding practices based on protocols developed by the INTERGROWTH-21st Neonatal Group (www.intergrowth21.org.uk). Exclusive breastfeeding up to 6 months was promoted61. During pregnancy, at birth and at 1 and 2 years of age, standardised information was obtained on health, anthropometric measures, severe morbidities, duration of breastfeeding, timing of the introduction of solid and semi-solid foods, feeding practices and food intake (www.intergrowth21.org.uk). We excluded children with severe morbidities up to 2 years of age, such as tuberculosis, hepatitis, HIV/AIDS, malaria, hearing problems, neurometabolic conditions, epilepsy, meningitis, seizures, cerebral palsy, cardiovascular problems, cystic fibrosis, blindness, haemolytic disorders and any malignancy. The baseline characteristics of the full cohort and follow-up methodology have recently been published11, as well those of the developmental sample12. The INTERGROWTH-21st Neurodevelopment Assessment (INTER-NDA) is a brief, objective, psychometric tool, measuring multiple dimensions of early child development, targeted at children aged 22–30 months60. We designed the tool specifically for implementation by non-specialists across international settings62. It consists of 37 items measuring cognition, expressive and receptive language, fine and gross motor skills, and positive and negative behaviour using a combination of directly administered, concurrently observed and caregiver reported items60. In addition, we also administered the attentional problems and emotional reactivity sub-scales of the CBCL to our study population24. The INTER-NDA shows good to moderate agreement with the Bayley Scales of Infant Development III edition62, which is considered the gold-standard child development assessment at individual level for screening and monitoring purposes63. It also has good levels of inter-rater (k  =  0.70; 95% CI: 0.47–0.88) and test/re-test reliability (k = 0.79; 95%CI: 0.48–0.96)60. INTER-NDA data were collected using a tablet-based system developed specifically for IFS, with an incorporated operation manual as well as visual cues, examples and fully integrated quality checks60. Staff administering the assessments were aware of the study’s general principles but not the specific hypotheses being tested. Data were uploaded to an encrypted, cloud-based server as soon as each assessment ended. The INTER-NDA’s negative behaviour domain describes negative aspects of a child’s behaviour, observed during the assessment, beyond what is expected for the child’s age. It includes ratings on distractibility items (poor attention to tasks, easily distractible, leaves tasks incomplete) and negative affect items (excessive tantrums, fussing, pouting, whining, crying and aggressive outbursts). For these analyses, we reversed negative behaviour scores to allow them to be expressed in a positive format, i.e. a higher ‘low negative’ behaviour score is more favourable. Individual children’s performance on the CBCL’s attentional “problems” and emotional reactivity subscales21 were estimated as the mean value for individual scores from the raw data and as CBCL centiles to identify children with scores above the clinical range on the CBCL, i.e. greater than 97th centile of their transformed distribution. The age of achievement of the gross motor development milestones “standing alone” and “walking alone”, as defined by WHO59,64, were evaluated for their association with feeding practices. There were some discrepancies between the reports at 1- and 2-year follow-up visits for 19 children out of 1292 for “standing alone” and 18 children out of 1296 for “walking alone”; in these cases, the 1-year information was preferred. The two gross motor development milestones were modified to correspond to an “earlier age for standing” and an “earlier age for walking”. The new variables represent the difference between 26 months and the age at which the child stood or walked alone (i.e. the larger the value, the earlier the child stood or walked on their own, in exact months). Vision was assessed using the Cardiff Visual Acuity and Contrast Sensitivity tests for binocular vision65. We identified a priori, three items for collection by direct observation at the time of the assessment, as indicators of the child’s autonomy i.e. independence from the mother: runs alone, climbs upstairs and drinks spontaneously from a cup deposited on the nursery table in front of the child, without any inducement from the examiner or mother. These were analysed as separate items. INTER-NDA items were scored on a four-point scale (range 1 to 4) and behavioural items were scored on a three-point scale (range 1 to 3 for observed behaviour and 0 to 2 for caregiver reported questionnaire items). The mean domain scores were multiplied by a factor of 10 to make it easier to interpret the results. Items related to the child’s independence from the mother and the INTER-NDA’s “executive-function like” item were analysed using their original scale. The main independent variables were those describing feeding practices, i.e. breastfeeding, formula feeding, introduction of solid/semi-solid foods, and number of feeds per day as reported at the 1-year visit. These exposures were evaluated as continuous variables, e.g. number of months of exclusive/predominant/any breastfeeding or age in months at which solid/semi-solid foods (weaning) were introduced, following working definitions recommended by WHO66. There were children who received milks other than formula as part of bottle-feeding practices; we used these data to refine the duration of exclusive/predominant breastfeeding variable. As the age at initiation of formula feeding could vary for children with the same duration of total time of exposure, analyses were adjusted by the age at which formula feeding started (linear regression models) or by indicators of the timing of exposure at critical ages: hospital discharge, 3, 6, and 9 completed months (logistic regression models). The latter strategy was chosen given that adjustment for the age of the start of formula feeding produced multi-collinearity in the multivariable models. Data for duration of exclusive/predominant breastfeeding were transformed into categorical variables to study the patterns of association between the timing of feeding and neurodevelopmental scores. To determine the cut-off points for this categorisation, we first conducted a review of the developmental periods traditionally proposed as key for the acquisition of psychological milestones25,33,34,44,68–70. We then established cut-off points for exclusive/predominant breastfeeding and age at weaning from birth to the postnatal ages less than 5 months, less than 7 months, less than 13 months and less than 18 months of age. Regardless of the terminology used, it is generally accepted that fundamental neurodevelopmental processes occur within these approximate time-windows. Thus, we expected to capture, for example, the effect on early child development of being exclusively breastfed from birth to 5 months as compared to a further 2 months of exclusive/predominant breastfeeding, i.e. from birth to less than 7 months of age. Linear regression analyses (crude and adjusted) were conducted to explore associations separately for each feeding exposure as a continuous or categorical (independent) variable against each neurodevelopmental domain as a continuous (dependent) variable. Robust standard errors were estimated in all linear association models. For adjustment, the following variables were taken into account when assessing the association between feeding practices and neurodevelopment: fetal head circumference z-score between 25 and 30 weeks’ gestation based on the INTERGROWTH-21st Fetal Growth Standards, sex (girls as the baseline group), gestational age at birth (in exact weeks, corroborated by ultrasound assessment at less than 14 weeks’ gestation), birth weight (kg), birth length (cm), NICU stay (yes or no), age of the child at the time of the neurodevelopmental assessment (in exact months). Height z-score at 2 years of age based on the WHO Child Growth Standards14 was not considered a confounder but rather a possible mediator; nevertheless, it was included in the models to explore the independent, non-nutritional effect of early feeding exposures. Finally, we also evaluated the possible confounding effects of maternal age, maternal education and postnatal environmental smoking exposure71,72. The relationship between study site and infant neurodevelopment has been already studied in this population, showing the two were only marginally associated12. Furthermore, its inclusion in the regression models could lead to over adjustment particularly given the extensive adjustment for individual characteristics already in place73. We selected neurodevelopmental domains for which a consistent association or trend was observed with breastfeeding exposures, and further explored possible effect modification by indicators suggested in the literature to be related to the mother-child dyad. Hence, we repeated analyses for gross motor, runs, climbs upstairs, visual acuity, attentional score and emotional reactivity as dependent variables, and exclusive breastfeeding and age at starting solid/semi-solid foods as independent variables, all as continuous scores, stratified by the following variables considered a priori as potential effect modifiers. We constructed these indices/variables because we considered them proxy indicators of maternal closeness, i.e. indicators of emotional and/or objective physical maternal-infant interaction: (1) “Distracting mother-infant relationship index”, constructed by adding up indicators for the mother: (a) being pregnant, (b) working outside the home, and (c) not being the main person feeding the child. Each indicator contributed a value of 1 for each year present, giving the score a theoretical range from 0 to 6 units. We defined categories for low distracting score (<3) versus distracting score (≥3); (2) “Gangs of mothers” effect, defined as the mother being the main person feeding the child versus another person (Gangs of mothers); (3) “External sociability”: if the infant attended a nursery school or similar facility outside the home in the first or second year of life, defined as not attending nursery versus attending nursery either year; (4) “Competition or Intrusion”: to explore the “competitive” feelings an infant might have for its mother when it realises it has siblings, i.e. “the infant is not alone”. The number of siblings was used as a proxy, defined as no siblings versus any sibling; and (5) “Age of the infant at the time the mother returned to work outside home”: stratified as a categorical variable that took the value of 0 if the mother did not work during the first 2 post-partum years; 1 if the mother went on to work when/or after the child was 6 months old, and 2 if the mother worked outside the house before the child was 6 months old. Infant morbidity was evaluated in the second year of life by creating an unweighted score including non-severe conditions such as repeated pneumonia, urinary tract infections, glomerulonephritis, metabolic disorders, type-1 diabetes and/or ketoacidosis, i.e. any condition requiring surgery or admission to hospital. A score related to infectious/allergic/autoimmune conditions was constructed separately, involving the following diagnoses: exanthema skin disease, repeated otitis media, repeated pneumonia, urinary tract infections, recurrent fever episodes, and gastrointestinal infections including repeated diarrhoea and persistent vomiting. In adjusted models, both morbidity scores were introduced as binary indicators given the small number of children with more than one morbid event for models including categorical exposures. Associations between breastfeeding practices and the two morbidity indicators were assessed in logistic regression models adjusted for the same covariates used in the multivariable linear regression models. For all analyses, Stata 15 software was used (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC). Data were entered locally into the specially developed, online data management system (http://medscinet.com)74. The INTERGROWTH-21st Project was approved by the Oxfordshire Research Ethics Committee “C” (reference: 08/H0606/139), the research ethics committees of the individual institutions and the regional health authorities where the project was implemented. Parents gave written, fully informed consent, on behalf of their children for their participation in the study. The sponsors had no role in the study design, data collection, analysis, interpretation of the data, or writing of the paper. The following authors had access to the full raw dataset: JV, ESU and SHK. The corresponding author had full access to all the data and final responsibility for submitting the paper. All methods were performed in accordance with the relevant guidelines and regulations.