Background Maternal nutrition in preconception and early pregnancy influences fetal growth. Evidence for effects of prenatal maternal nutrition on early child development (ECD) in low-income and middle-income countries is limited.
Objectives To examine impact of maternal nutrition supplementation initiated prior to or during pregnancy on ECD, and to examine potential association of postnatal growth with ECD domains.
Design Secondary analysis regarding the offspring of participants of a maternal multicountry, individually randomised trial.
Setting Rural Democratic Republic of the Congo, Guatemala, India and Pakistan.
Participants 667 offspring of Women First trial participants, aged 24 months.
Intervention Maternal lipid-based nutrient supplement initiated preconceptionally (arm 1, n=217), 12 weeks gestation (arm 2, n=230) or not (arm 3, n=220); intervention stopped at delivery.
Main outcome measures The INTERGROWTH-21st Neurodevelopment Assessment (INTER-NDA) cognitive, language, gross motor, fine motor, positive and negative behaviour scores; visual acuity and contrast sensitivity scores and auditory evoked response potentials (ERP). Anthropometric z-scores, family care indicators (FCI) and sociodemographic variables were examined as covariates.
Results No significant differences were detected among the intervention arms for any INTER-NDA scores across domains, vision scores or ERP potentials. After adjusting for covariates, length-for-age z-score at 24 months (LAZ24), socio-economic status, maternal education and FCI significantly predicted vision and INTER-NDA scores (R2=0.11–0.38, p<0.01).
Conclusions Prenatal maternal nutrition supplementation was not associated with any neurodevelopmental outcomes at age 2 years. Maternal education, family environment and LAZ24 predicted ECD. Interventions addressing multiple components of the nurturing care model may offer greatest impact on children’s developmental potential.
Trial registration number NCT01883193.
- child development
Data availability statement
Data are available on reasonable request. On publication of study findings, de-identified study data will be available through the NICHD Data and Specimen Hub (N-DASH) at https://dash.nichd.nih.gov.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
WHAT IS ALREADY KNOWN ON THIS TOPIC
Improved maternal nutrition during the preconception and early gestation periods improves fetal and child growth in settings with high rates of stunting.
Poor physical growth and impairments in early child development (ECD) frequently co-exist.
WHAT THIS STUDY ADDS
Maternal nutrition supplementation initiated before and early in pregnancy and discontinued at delivery did not improve cognitive, language, gross motor, fine motor, positive and negative behaviour scores; visual acuity and contrast sensitivity scores and auditory evoked response potentials markers in children aged 2 years from Democratic Republic of the Congo, Guatemala, India and Pakistan.
Maternal education, family environment and child length at 24 months were associated with multiple ECD outcomes in these diverse settings, based on a multidomain, rapid, low-cost ECD assessment tool designed for low-resource settings.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Our study emphasises that a single nutritional strategy, that is, preconception and prenatal maternal nutrition supplementation, is insufficient to demonstrate positive gains in young children’s development.
Rather, a multisectoral approach is needed to maximise opportunities to improve children’s early development.
The first 1000 days of life are a well-established critical window of opportunity for improving child growth and development.1–3 While numerous early life exposures (ELEs) are associated with delays in early childhood development (ECD),4 four key risk factors (stunting, iodine deficiency, iron deficiency anaemia and inadequate cognitive stimulation) have been identified by the Lancet’s International Child Development Steering Group as urgent needs for intervention globally.3 Three of these relate specifically to maternal and child nutrition.3
Evidence from both preclinical and human studies indicate that preconception or periconception maternal nutritional status influence fetal growth and development, with life course effects on health and neurocognitive function.5 Strong associations between intrauterine and extrauterine growth, and ECD, have been demonstrated across disparate populations,6 7 so that, in some comparisons, childhood stunting is considered a proxy for neurodevelopmental risk.6 Undernutrition during early life may, therefore, be considered to be a potentially preventable cause of ECD delay. This presents a strong theoretical rationale for the initiation of maternal nutrition supplements prior to conception, to correct both maternal underweight and micronutrient deficiencies before and during sensitive periods of fetal brain development. Preconception and early pregnancy maternal nutritional supplementation are reported to improve birth outcomes and postnatal linear growth, but understanding of its impact on ECD remains limited.8
The ‘Women First’ Preconception Maternal Nutrition Trial (WF) was undertaken in four countries with high rates of childhood stunting.9 The trial resulted in significant improvements in birth length-for-age z-scores (LAZ), early postnatal growth and linear growth trajectories from birth to 24 months in the offspring of women who received nutritional supplementation initiated preconceptionally (arm 1) or at approximately 12 weeks gestation (arm 2) compared with no supplement (arm 3).8–10 WF is the first multicountry randomised controlled trial (RCT) to (i) examine associations between preconception maternal nutrition and ECD outcomes in four geographically and culturally disparate low-income and middle-income countries (LMICs) populations; (ii) measure neurocognitive outcomes using a comprehensive, rapid, low-cost ECD assessment (INTERGROWTH-21st Project Neurodevelopment Assessment (INTER-NDA))4; (iii) categorise ECD delay based on prescriptive, international standards rather than population-specific references; and (iv) include measurements of vision and cortical auditory processing.
In the present study, we compared multiple domains of neurodevelopment at age 2 years in a randomly selected subset of children in each WF intervention arm. Our aims were to (1) examine associations between WF intervention arm and ECD outcomes and (2) determine which, if any, ELE predict ECD outcomes at age 2. Our a priori trial hypothesis was that the gains previously reported for postnatal growth would be associated with gains in ECD scores at 2 years among the offspring of mothers who received nutritional supplementation.
This analysis included prospectively planned neurodevelopmental testing and anthropometry on live-born infants of WF participants. For this report, these outcomes were obtained on a randomly selected subset of the infants, representing approximately one-third of the WF offspring, evenly distributed across intervention arms and research sites. The remaining offspring were evaluated with the Bayley Scales of Infant Development (Pearson, San Antonio, Texas, USA).11
The primary WF trial was a multisite, individually randomised clinical trial of a daily 22-micronutrient fortified small-quantity lipid-based nutrient supplement formulated for pregnancy (Nutriset, Malauney, France; online supplemental material S1). The supplement was initiated at randomisation with continuation for ≥3 months (average ~9 months) before conception through delivery (arm 1), vs initiation of the same supplement late in the first trimester of pregnancy and continued through delivery (arm 2), vs no trial supplement (arm 3).12 Additionally, women in arms 1 and 2 who were underweight or had inadequate gestational weight gain were provided a balanced protein-energy lipid-based supplement (without additional micronutrients). No postnatal interventions were offered. Details on the trial’s protocol and follow-up procedures have been previously published.8 10 12
The study sites were rural communities in India (Belagavi, Karnataka), Pakistan (Thatta, Sindh), Democratic Republic of the Congo (DRC, Sud-Ubangi) and Guatemala (Chimaltenango).12
Participants and eligibility
Eligible participants for the primary WF trial were identified through the NICHD Global Network (GN) Maternal and Newborn Health Registry, household surveys and community meetings at each site.9 12 Infants who completed the birth, 6-month and at least two of the three 12–24 months follow-up visits were assessed for ECD outcomes at 24 months between August 2016 and March 2019.
Enrolment and randomisation
The central data coordinating centre (RTI International, Durham, North Carolina, USA) created the initial randomisation scheme, which included a permuted block design stratified by GN with a trial arm allocation ratio of 1:1:1 within blocks.9 12 Random assignment to neurodevelopment assessment was made before the 24-month visit and included approximately one-third of infants, evenly distributed across arms and sites.
Follow-up and ECD outcomes
Anthropometric measurements (weight, length and head circumference) were performed on children at 0 (birth), 6, 12, 18 and 24 months according to standardised procedures by trained research team members in the home or clinical environment (online supplemental material S2).10 Demographic, medical and perinatal information was collected at birth.9 Using WHO Child Growth Standards,13 z-scores, accounting for sex and age at time of measurement, were determined for length-for-age (LAZ), weight-for-age (WAZ) and head circumference-for-age (HCAZ). Home environments were assessed with family care indicator (FCI) Questionnaire at 24 months (online supplemental material S2).14 Assessors were blinded to the original randomisation assignment.
A holistic approach to ECD measurement, involving multiple outcomes, was undertaken at 24 months as follows:
Neurocognitive development: cognitive, language, fine and gross motor, positive and negative behaviour scores and corresponding risks of delay were measured on INTER-NDA . The INTER-NDA is an international, psychometrically valid, standardised, ECD assessment whose norms (online supplemental file 1) are international ECD standards, constructed according to WHO’s prescriptive guidelines.4
Vision: visual acuity (VA; measured in Logarithm of Minimum Angle of Resolution) and contrast sensitivity (CS; %) were assessed using Cardiff tests (PA Vision, UK).15 16
Cortical auditory processing: amplitudes and latencies of auditory evoked response potentials (CA-ERPs) to three types of auditory stimuli (frequent, infrequent and novel) were measured for three ERP components (P1, N2 and P3a waves) using the ‘novelty oddball’ ERP task17 (online supplemental material S4).
The administrative protocols for the ECD assessments are available at: https://www.intergrowth21.org.uk.
Sample size estimations and power calculations
Allowing for multiple comparisons (arm 1 vs arm 2 and arm 1 vs arm 3, within each site; total of 8 comparisons), a conservative sample size of 44 children/arm/site (combined site total of 176/arm) would have allowed detection of a statistically significant mean difference of 1.3 (1.5) or greater with 80% (90%) power (assuming two-sided test with overall 5% type I error). The actual number of infants who survived, were retained for follow-up, and were consented to the INTER-NDA assessment was greater than initial estimates, and included >200 children per arm (combined site).
Statistical analysis was performed in SPSS V.25.0 (IBM, Armonk, New York, USA). Prenatal, perinatal and postnatal characteristics were compared between arms and for children completing the ECD assessment and those lost to follow-up.
The distributions of ECD outcomes were inspected visually. ERP data were normally distributed; INTER-NDA and vision data were not. No transformation was identified that suited the latter; therefore, we used non-parametric tests (Kruskal-Wallis one-way analysis of variance (ANOVA)) to compare outcomes among arms. ANOVAs were used for ERP comparisons (effect sizes as eta-squared (η2)). Proportions of ECD delays between arms were compared using χ2 tests (effect sizes as Cramer’s V).
As ELEs were normally distributed, covariate analyses were undertaken to determine if any ELEs were associated (independently of maternal intervention arm) with ECD outcomes using correlations, followed by independent sample t-tests for associations identified as significant. Effect sizes were quantified using Cohen’s d and 95% CIs.18 We used generalised linear regression analysis, adjusting for child sex and age at measurement, to determine exposures that predicted ECD outcomes at 2 years independent of other ELEs. Generalised linear models were selected for their utility when outcome variables (INTER-NDA and vision outcomes) were not normally distributed or when the relationship between the exposure (ELEs) and outcome was non-linear.
Of the 730 children eligible for testing at 2 years, complete INTER-NDA data were obtained for 667 children (91.4% of eligible population), vision data for 613 children (83.9%) and ERP data of sufficient quality for analysis for 123 children (16.8%) (figure 1). Across ECD outcomes, the mean proportional contribution by study arm and site were well balanced (online supplemental material S5).
Characteristics of study population
Prenatal, perinatal and postnatal characteristics of the ECD cohort are presented in table 1. The mean (±SD) age at assessment was 24.6 months (±0.94), with 48% (n=323) male. Mean maternal education was 4.5 years (±4.2) and 31.9% of the cohort met criteria for low socio-economic status (SES). Forty-five children (6.7%) spent ≥3 days in hospital during the first 2 years, and >99% were breast fed at 6 months.
Comparisons in ECD outcomes between study arms
The associations between ECD outcomes and study arm are presented in table 2. No differences in INTER-NDA, vision and ERP outcomes were detected among treatment arms.
Overall, high rates of cognitive and motor delays and negative behaviour problems were reported: 246 (36.9%), 506 (75.9%) and 379 (56.2%) children, respectively, scored in the INTER-NDA’s range for severe delays in these domains; 444 (66.6%), 609 (91.3%) and 615 (92.2%) children scored in the range for any delay (table 2). Low VA and CS were reported in 24.0% and 19.3% of the cohort. Delays and behaviours did not consistently differ by maternal intervention arm.
Associations between ECD outcomes and early life exposures
After adjusting for infant sex, age at ECD assessment and multiple covariates (table 3), LAZ at 24 months was the only anthropometric variable that was significantly associated with ECD, including VA, gross motor, language (p<0.001 for all) and positive behaviour (p=0.01). Among other ELEs, maternal education was positively associated with vision, cognition, fine motor and language (p<0.001 for all); FCI (play activities) was associated with language (p<0.001) and SES was marginally associated with cognition, fine motor and positive behaviour (p<0.05).
Correlations between ECD outcomes and ELEs are presented in online supplemental material S6. Anthropometry z-scores (length, weight and head circumference), SES and play activities (FCI) were positively correlated with all vision and INTER-NDA outcomes. Only 3% (n=11/360) of associations between ERP outcomes and ELEs studied were significant, with no clear pattern of association detected; hence, further analyses were not undertaken. Comparisons of ELEs in children with low vision, and any INTER-NDA delay, are presented in online supplemental material S7. Higher maternal age at birth, lower anthropometry z-scores and lower SES, FCI and years of maternal schooling were associated with low VA and CS scores. Effect sizes were small to moderate for all associations (d=0.20–0.45) except for maternal schooling (d=0.99).
Lower anthropometry z-scores at all time points were associated with cognitive delay with moderate effect sizes (d=0.3–0.7). Lower LAZ24 and FCI were significantly associated with delays across all INTER-NDA domains. Lower LAZ12, WAZ12&24 and HCAZ12 were also significantly associated with delays across all INTER-NDA domains except behaviour problems. Where domains were associated with serial growth measurements, effect sizes increased as children aged. For example, for cognitive delay, LAZ effect sizes were 0.31, 0.56 and 0.69 at 0, 12 and 24 months, respectively. Similar patterns were observed for weight, and for gross motor, fine motor and language delays (online supplemental material S7).
To our knowledge, this is the first multicentre RCT to examine the effect of preconception maternal nutrition supplementation on comprehensive ECD outcomes using a standardised ECD measure developed specifically for LMICs. Our key finding was that the benefits of the maternal intervention previously reported for fetal9 and postnatal growth8 did not extend to gains in ECD scores or to reduced rates of ECD delays at 2 years among the offspring of mothers who received nutritional supplementation. Linear growth status at 24 months was a significant predictor of scores in several domains, including vision (VA), language, gross motor and positive behaviour. Additionally, indicators of family environment (play activities and play materials) and SES predicted several ECD scores, although differential associations existed between these and ECD domains. Notably, maternal education was a consistent and potent predictor for several domain scores, including vision (VA and CS), cognitive, language and fine motor. High rates of cognitive and motor delays and negative behaviours were observed, as expected in low-resource populations with rates of child stunting≥60%19 20; delayed ECD and stunting share many drivers.
Our findings differ only slightly from those of the preconceptual micronutrient supplementation trial (PRECONCEPT) from Vietnam, the only other RCT to report the effects of preconception maternal supplementation on child growth and ECD.21 PRECONCEPT reported small group differences favouring preconception iron-folate supplementation for fine motor development (effect size 1.3 SD; 95% CI 0.05 to 0.77), but not for other ECD domains at 2 years, or for any ECD outcomes at 1 year, despite gains in LAZ and lower rates of stunting at 2 years.21 A prenatal and postnatal maternal multiple micronutrient supplement (MMS) trial from Bangladesh found no impact of maternal supplementation on children’s cognitive and motor scores at 2 years.22 Likewise, a meta-analysis of prenatal MMS trials from LMICs (88 057 women) concluded that prenatal MMS did not lead to a consistent cognitive benefit for children.23
It is not clear why previously reported early gains in length and weight following maternal supplementation are not consistently associated with ECD benefits for children.21 23 24 One reason may be because extant maternal supplementation trials were powered to detect differences in child growth and that larger sample sizes are required to detect differences in ECD outcomes.23 It is also possible that ECD measures developed for high-income countries may not be sensitive indicators for LMICs.25 Nevertheless, in our study, even with the use of the INTER-NDA designed specifically for LMICs, we did not detect treatment effects. Some ECD effects may remain latent and manifest at older ages.21 Additionally, as a screening tool, the INTER-NDA is not intended to detect subtle differences. Although in the WF trial maternal supplementation was associated with improved fetal9 and postnatal growth,10 multiple critical aspects of neurological maturation occur postnatally and are influenced by environmental factors, many of which differ among settings.26 Finally, although poor compliance with the maternal intervention could theoretically explain the lack of ECD differences between arms, overall compliance was ≥80%.9 The improved birth and postnatal anthropometry reported for both intervention arms (compared with controls) make this explanation unlikely.
Our findings of the associations between general family environment, particularly maternal education and SES, the provision of stimulating environments (as assessed by FCI) and better ECD outcomes are consistent with previous reports27–29 and emphasise the importance of socio-environmental determinants in addition to biomedical determinants on long-term neurodevelopment.30 31
Key strengths of our study are the multicentre design in low-resource populations from four geographically and culturally distinct LMICs; incorporation of measures of vision and cortical auditory processing in ECD measurements; the use of the INTER-NDA and its international ECD standards and our adoption of a LMIC-centric approach to assessment (viz, low cost, rapid assessment time and non-reliance on specialists for administration).
Study limitations include sample size that was insufficient for intersite comparisons within arms. The diverse sites’ heterogeneity20 may have masked treatment-arm effects. The number of auditory ERP assessments of suitable quality for analyses was small due to the technical challenges of collecting high-quality recordings in these field conditions. Our experience emphasises the need for more refined, low-cost tools suitable for large-scale implementation in field settings.
In our study, maternal nutrition supplementation initiated either before or early in pregnancy and discontinued at delivery did not improve cognitive, language, gross motor, fine motor, positive or negative behaviour scores; VA or CS scores; or auditory ERP markers in children aged 2 years from diverse low-resource settings. These findings emphasise that a maternal nutritional intervention strategy alone was insufficient to demonstrate positive gains in young children’s development. Rather, multiple socio-environmental factors, including family environment, maternal education and children’s postnatal linear growth, were positively associated with ECD outcomes.
Data availability statement
Data are available on reasonable request. On publication of study findings, de-identified study data will be available through the NICHD Data and Specimen Hub (N-DASH) at https://dash.nichd.nih.gov.
Patient consent for publication
The Women First trial was approved by the Colorado Multiple Institutional Review Board (#13-2160); Comité de Ética Universidad Francisco Marroquin (CE-FM/UFM 059-17, Guatemala); JNMC Institutional Ethics Committee on Human Subjects Research and the Indian Council of Medical Research (KAHER/EC/2018-19/D3017, India); Comité D’Ethique, Ecole De Sante Publique, University of Kinshasa (ES/CE/102B/14, Democratic Republic of the Congo); the Aga Khan University Ethical Review Committee (2753-CHS-ERC-13, Pakistan) and RTI International (North Carolina, USA). Mothers provided written informed consent for themselves and their children.
The Women First Maternal Preconception Nutrition Trial Study Group consists of the following additional members: Carl L Bose (University of North Carolina, Chapel Hill, NC, USA); neurodevelopment assessors: Aura Arevalo, Gelen Gomez, and Marta Lidia Aguilar (INCAP, Guatemala); Zahid Abbasi, Sumera Ali and Sumaira Fatima (AKU, Pakistan); Deepa Metgud and Spurthi Mastiholi (JNMC, India); Robert Kpado, Croco Gbenge, Matthieu Gbozo, Philippe Zolia, Papy Fakadanga, and Joel Eay (KSPH, DRC); Vanessa R Thorsten, Dhuly Chowdhury, Abhik Das, and Kristen Stolka (RTI International, Durham, NC, USA); and Omrana Pasha (AKU, Pakistan).
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
MF and NFK are joint first authors.
Collaborators The Women First Maternal Preconception Nutrition Trial Study Group consists of the following additional members: Carl L. Bose (CLB) (University of North Carolina, Raleigh, North Carolina, USA); neurodevelopment assessors: Aura Arevalo (AA), Galen Gomez (GG) and Marta Lidia Aguilar (MLA) (INCAP, Guatemala); Zahid Abbasi (ZA), Sumaira Fatima (SF) (AKU, Pakistan); Deepa Metgud (DM) and Spurthi Mastiholi (SM) (JNMC, India); Robert Kpado (RK), Croco Bgenge (CB), Matthieu Gbozo (MG), PhilippeZolia (PZ), Papy Fakadanga (PF) and Joel Eay (JE) (KSPH, DRC); Vanessa R.Thorsten (VRT), Dhuly Chowdhury (DC), Abhik Das (AD) and Kristen Stolka (KS) (RTI International, Durham, North Carolina, USA) and Omrana Pasha (OP) (AKU, Pakistan).
Contributors NFK, MF and MH conceived and designed the study; MF, NFK and JW wrote the final protocol in collaboration with all members of the trial group (AT, AL, MB, ALG, LF, SS, SAA, RLG, SSG, SMD, RJD, MK-T, AS, EMMcC and members of the Women First Preconception Nutrition Trial Group listed above); MF provided expert training to research assessors in each site; AT, AL, ALG, LF, SS, SAA, SSG and SMD coordinated implementation of the study at the country level; NFK, CLB, RLG and RJD provided overall supervision of study conduct; AA, GG, MLA, ZA, SA, SF, SM, RK, CG, MG, PZ, PF and JE performed developmental assessments; MF and NFK drafted the manuscript with critical input from all authors for subsequent revisions; JFK supported data base management and statistical analyses; MF, AS, EMMcC, DC, VRT and AD provided statistical analyses. All authors read and approved the final version of the manuscript. NFK acts as guarantor for the manuscript.
Funding The Thrasher Research Foundation funded this study; the Bill & Melinda Gates Foundation, the Eunice Kennedy Shriver National Institutes of Child Health and Human Development and the NIH Office of Dietary Supplements funded the primary Women First trial. MF is supported by a Clinical Research Training Fellowship from the Medical Research Council, UK.
Competing interests None declared.
Provenance and peer review Not commissioned; internally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.