Linear Growth and Child Development in Burkina Faso, Ghana, and Malawi
OBJECTIVES: We aimed to produce quantitative estimates of the associations between 4 domains of child development and linear growth during 3 periods: before birth, early infancy, and later infancy. We also aimed to determine whether several factors attenuated these associations.
METHODS: In 3700 children in Burkina Faso, Ghana, and Malawi, growth was measured several times from birth to age 18 months. At 18 months, language, motor, socioemotional, and executive function development were assessed. In Burkina Faso (n = 1111), personal-social development was assessed rather than the latter 2 domains.
RESULTS: Linear growth was significantly associated with language, motor, and personal-social development but not socioemotional development or executive function. For language, the pooled adjusted estimate of the association with length-for-age z score (LAZ) at 6 months was 0.13 ± 0.02 SD, and with ΔLAZ from 6 to 18 months it was 0.11 ± 0.03 SD. For motor, these estimates were 0.16 ± 0.02 SD and 0.22 ± 0.03 SD, respectively. In 1412 children measured at birth, estimates of the association with LAZ at birth were similar (0.07–0.16 SD for language and 0.09–0.18 SD for motor development). These associations were weaker or absent in certain subsets of children with high levels of developmental stimulation or mothers who received nutritional supplementation.
CONCLUSIONS: Growth faltering during any period from before birth to 18 months is associated with poor development of language and motor skills. Interventions to provide developmental stimulation or maternal supplementation may protect children who are faltering in growth from poor language and motor development.
- CDI —
- communicative development inventory
- DMC-II —
- Developmental Milestones Checklist–II
- FCI —
- family care indicators
- HFIA —
- household food insecurity access
- iLiNS —
- International Lipid-Based Nutrient Supplements
- LAZ —
- length-for-age z score
- LMICs —
- low- and middle-income countries
- LNS —
- lipid-based nutrient supplements
- MMN —
- multiple micronutrients
- WHO —
- World Health Organization
What’s Known on This Subject:
One meta-analysis has examined associations between linear growth and development in low- and middle-income countries, reporting cross-sectional estimates pooling 3 to 4 studies in children aged <2 years for language and motor development but not socioemotional development or executive function.
What This Study Adds:
Linear growth was significantly associated with the development of motor and language skills but not socioemotional or executive function, with similar estimates across 4 different studies comprising 3700 children and across 3 periods of infancy from birth to 18 months.
In many populations in low- and middle-income countries (LMICs), faltering in linear growth begins before birth and continues until at least 2 years of age.1,2 Brain development occurs rapidly during this same period, from conception to age 2 years, laying the foundation for the development of cognitive, motor, and socioemotional skills throughout childhood and adulthood. The extent to which children globally are faltering in the development of these abilities during the first 2 years of life is less clear. Although stunted linear growth has been used as a proxy for children’s broader developmental status, the nature of the association between growth and development is not fully understood.3
Several qualitative reviews of studies reporting associations between growth and development in LMICs concluded that growth stunting is associated with concurrent and longer term deficits in cognition, behavior, motor skills, and school performance.3–5 The only systematic review and meta-analysis of associations between linear growth and development reported a cross-sectional unadjusted pooled estimate of 0.28 SD difference in cognitive score and 0.24 SD difference in motor score with each 1 SD difference in length-for-age z score (LAZ) in children aged <2 years.6 However, due to methodologic inconsistencies across studies or lack of data, this estimate was derived from only 3 studies in the cognitive domain and 4 studies in the motor domain; pooled estimates in other domains, such as socioemotional development and executive function, could not be generated. Similarly, pooled estimates for studies examining the associations with change in LAZ over time could not be generated due to the small number of studies and differences across studies in the timing of height measurements.
We address these gaps by using data from the International Lipid-Based Nutrient Supplements (iLiNS) Project. The iLiNS Project comprised 4 randomized trials that enrolled >7000 pregnant women or infants in Burkina Faso, Ghana, and Malawi. All trials included groups of mothers and/or children who received various doses and formulations of lipid-based nutrient supplements (LNS) and control groups who did not receive LNS. Linear growth was measured at multiple time points during infancy, and developmental assessments were conducted at 18 months of age. The main outcomes of the trials have been reported elsewhere.7–10 In brief, results differed across trials, with positive effects on 18-month linear growth in Burkina Faso9 and Ghana7 but not in the 2 trials in Malawi.8,10 Effects on 18-month child development also differed across trials, with positive effects in Burkina Faso11 but not in Ghana12 or Malawi (E.L.P., S.A.A., A.L., et al, unpublished observations).13 The focus of the present study was on the associations between linear growth and development in the 4 cohorts, regardless of intervention group (ie, controlling for intervention group as a covariate). We also investigated intervention group as a potential effect modifier (as detailed in the fourth objective).
The first objective of this study was to explore the extent to which children in each cohort had faltered in growth and development by age 18 months. The second objective was to examine the association of each developmental domain with attained length and linear growth increments during different periods of infancy. The third objective was to examine the extent to which these associations were confounded by other factors such as socioeconomic status, maternal education, and the developmental stimulation the child received. Association does not indicate causality; thus, evidence for such confounding would suggest that linear growth may not be a determinant of child development but rather a proxy for an environment that constrains both linear growth and development.
The fourth objective was to examine 3 potential modifiers of these associations. In populations with few environmental constraints on growth, the height that children attain is likely to be constrained only by genetic potential.14 In such populations, an association between linear growth and development is not observed.15 This lack of association is also likely to be observed in certain populations within LMIC contexts. For example, growth faltering in later infancy might not be associated with poor development if growth was sufficient during an earlier time period. It is also possible that growth faltering is not associated with poor development among children with high levels of developmental stimulation from the environment. For example, in 6- to 8-year-old children in Vietnam, concurrent growth stunting was associated with cognitive scores among children who had not participated in a preschool educational program at age 3 to 4 years but not among those who had participated in the program.16 Thus, we examined the following: (1) the interaction between early growth status and later growth; (2) the interaction with the amount of developmental stimulation from the environment; and (3) the interaction with trial group.
iLiNS Project Trial Designs
All children for whom complete data were available are included in the present study, including LAZ scores, 18-month development scores, and complete covariate information. Complete information was available for a total of 3700 children: 1111 in iLiNS-ZINC, 1107 in iLiNS-DOSE, 568 in iLiNS-DYAD-M, and 914 in iLiNS-DYAD-G. In the iLiNS-ZINC trial, only a random subsample of 4 of the 5 trial groups was targeted for developmental assessment.11 Baseline characteristics of the samples included in the analysis compared with those excluded are presented in Supplemental Tables 7, 8, 9, and 10.
Measurement of Linear Growth
In all trials, child length was measured to the nearest 1 mm at multiple time points by teams of 2 trained and standardized anthropometrists using length boards. Length measurements from the following time points were used in the analyses reported here: for the iLiNS-ZINC trial, age 9 and 18 months; for the iLiNS-DOSE trial, age 6 and 18 months; and for the 2 iLiNS-DYAD trials, within 6 weeks of birth and age 6 and 18 months. LAZ was calculated by using the World Health Organization (WHO) growth standards.17
Developmental Assessment in Burkina Faso
In the iLiNS-ZINC trial in Burkina Faso, motor, language, and personal-social development at age 18 months was assessed by using the Developmental Milestones Checklist–II (DMC-II).18 This tool evaluates the child's motor, language, and personal-social development through both interviewing the caregiver and observing the child. Motor items measured fine and gross motor skills; language items measured receptive and expressive language skills; and personal-social items included skills such as joining others in play, dressing, feeding, and toilet training. The score for each subscale was the sum of the item scores.
Developmental Assessment in Malawi and Ghana
In the iLiNS-DOSE and iLiNS-DYAD trials in Malawi and Ghana, the following assessments were used. Motor development was assessed by using the Kilifi Developmental Inventory (KDI), which is a tool developed in Kenya based on several standard tests.19 The child’s score was the total number of fine and gross motor skills he or she was observed to perform. Language development was assessed by using a 100-word vocabulary checklist based on the MacArthur-Bates Communicative Development Inventory (CDI).20 We developed the checklist in the local languages at each site through interviews with caregivers and pilot testing. The child’s score was the total number of words the child said, from the 100-word list, as reported by the mother. Socioemotional development was assessed by using the Profile of Social and Emotional Development, a test developed in Kenya based in part on the Brief Infant-Toddler Social and Emotional Assessment. The 19 items probe emotional regulation (eg, throwing tantrums, showing jealousy), behavioral regulation (eg, sitting still, paying attention), aggression (eg, hitting others), and social competence (eg, sharing toys). The items were summed for a total score, such that a higher score indicated fewer socioemotional problems.
Executive function was assessed by using a version of the A-not-B task, which is a widely used test of working memory and executive function in very young children that has been previously used successfully in Kenya and Uganda.21–23 In each of 10 trials, a small piece of cracker was hidden under 1 of 2 identical cups on a wooden board. The board was removed from sight for 5 seconds, during which the child was distracted with a song. The board was then returned, and the child was invited to find the cracker. Every time the child achieved 2 correct consecutive trials, the cracker was then hidden at the alternate location. The score was the total number of correct trials.
Nurturing and Developmental Stimulation in the Home Environment
The child’s stimulation from the home environment was evaluated by using the Family Care Indicators (FCI) interview.24,25 The mother was interviewed with regard to the child’s play materials and activities with caregivers in the past 3 days (Supplemental Information). The overall FCI score was calculated as the sum of the 18 item scores.
Tester Training and Reliability
The first author trained the developmental assessment teams in each trial. All data collectors were required to pass knowledge- and practice-based evaluations before administering the tests and interviews. Inter-scorer agreement and test–retest reliability were evaluated for each team and have been reported elsewhere (E.L.P., S.A.A., A.L., et al, unpublished observations).12,13,18,20 Further details are given in Supplemental Table 11.
Description of Covariates
At baseline, data collectors recorded maternal and household information. Maternal education and household assets were coded as above or below the median for that cohort. We used a cutoff of >2 to indicate a relatively higher level of food insecurity. A detailed description of these variables is given in the Supplemental Information.
For the DMC-II scores, z scores in the iLiNS-ZINC sample were calculated by standardizing to a mean of zero and an SD of 1. For the Kilifi Developmental Inventory, CDI, Profile of Socio-Emotional Development, and A-not-B task, z scores were calculated on the full sample of children from the 3 cohorts. To reduce skewness to <1, the Kilifi Developmental Inventory motor score and the DMC-II motor and personal-social scores were log-transformed and the CDI vocabulary score was square root transformed before calculating the z scores. For all other scores, skewness was <1.
The first objective was assessed in the following ways. To evaluate faltering in linear growth, mean LAZ was calculated according to WHO standards,17 which indicate the extent to which each cohort had faltered, on average, compared with the WHO sample, in units of SD. For the developmental assessments, no such norms exist. We therefore calculated the mean developmental scores in a subsample of children across the cohorts with fewer environmental constraints on development. This low-risk sample comprised all children without any of the following risk factors: asset index below the cohort median, maternal or paternal education below the cohort median, FCI score below the cohort median, stunted (LAZ less than –2) at 6/9 or 18 months, and anemic (blood hemoglobin concentration <110 g/L) at 6/9 or 18 months. We selected risk factors for poor child development that have been identified in previous studies26 and were available in all of our data sets. The distribution of scores in the low-risk sample provides an estimate of the potential scores that children could reach with fewer health and environmental constraints on development. We also compared scores in the highest and lowest wealth quartile, based on the asset index (Supplemental Information), using generalized linear models controlling for child age and sex. This method allowed us to evaluate the extent to which economically disadvantaged children had faltered compared with their more advantaged peers.
For the second objective, 2 steps were used. First, in each cohort, we examined growth status in early infancy (LAZ at 6/9 months) and growth in late infancy (ΔLAZ from 6/9 to 18 months). We calculated ΔLAZ by subtracting LAZ at 6/9 months from LAZ at 18 months. Second, in each iLiNS-DYAD cohort, we examined LAZ at birth, ΔLAZ from birth to 6 months, and ΔLAZ from 6 to 18 months.
For the third objective, 2 linear regression models were fit in each step. In model 1, the LAZ and ΔLAZ scores, child age and sex, data collector, and trial group were included (Table 1 presents the trial groups). In model 2, all additional covariates were added: baseline maternal age and education, asset index below median, household food insecurity access (HFIA) >2, HIV status, and FCI score.
For the pooled analysis of the iLiNS-DOSE and iLiNS-DYAD cohorts, trial cohort group was coded as an 8-category variable indicating both trial cohort and intervention group (2 groups in the iLiNS-DOSE cohort and 3 groups in each of the DYAD trial cohorts) (Table 1). We calculated both a fixed effects model, which assumes no heterogeneity between trial cohorts, and a random effects model, with random effects of trial cohort on intercept, slope of LAZ at 6 months, and slope of ΔLAZ from 6 to 18 months. The latter model allows the associations between linear growth and development to vary between trial cohorts.27
For the fourth objective, we examined the data from the iLiNS-ZINC cohort and the pooled data from the iLiNS-DOSE and iLiNS-DYAD cohorts. To evaluate whether the association with later growth was modified by early growth status, the interaction between LAZ at 6/9 months and ΔLAZ from 6/9 to 18 months was added to each model. To evaluate whether the association was modified by stimulation from the environment, we added to each model the interaction between the following factors: (1) FCI score and 6/9-month LAZ; and (2) FCI score and ΔLAZ from 6/9 to 18 months. To evaluate whether the association was modified according to trial group, each cohort was examined separately, and we added to the model the interaction between: (1) trial group and 6- month LAZ for the DYAD cohorts; and (2) trial group and ΔLAZ from 6/9 to 18 months for all 4 cohorts. If any interaction was significant at the P < .1 level, the association within each tertile of the effect modifier or within each trial group was examined. All statistical analyses were conducted by using SAS version 9.4 (SAS Institute, Cary, NC), except the I2 statistic, which was calculated by using Stata version 14.1 (StataCorp, College Station, TX).
Table 2 displays the mean LAZ and developmental scores in each cohort. At birth, LAZ of the cohort in iLiNS-DYAD-M was on average 1 SD below the mean of the WHO norm sample. On average, this cohort faltered an additional 0.2 SD between birth and 6 months and 0.3 SD between 6 and 18 months. Mean LAZ of the cohort in iLiNS-DYAD-G was 0.6 SD below the mean of the WHO norm sample at birth, and they faltered an additional 0.1 SD from birth to 6 months and 0.1 SD from 6 to 18 months. The iLiNS-ZINC cohort was 1.2 SD below the mean of WHO norms at age 9 months and faltered an additional 0.4 SD from 9 to 18 months. The iLiNS-DOSE cohort was on average 1.4 SD below the mean of WHO norms at age 6 months and faltered an additional 0.5 SD from 6 to 18 months.
The distributions of motor, language, socioemotional, and executive function scores were similar in the 3 cohorts in Ghana and Malawi (Table 2). The low-risk sample included 5 children in the iLiNS-ZINC cohort, 56 children in the iLiNS-DOSE cohort, 17 children in the iLiNS-DYAD-M cohort, and 80 children in the iLiNS-DYAD-G cohort. For the motor, socioemotional, and executive function scores, the mean and SD of the low-risk sample were within the same range as the means and SDs of each cohort. For language development, the mean vocabulary score of the low-risk sample was 40 words, whereas the mean vocabulary scores of the full cohorts in Ghana and Malawi were 25 to 28 words.
Figure 1 displays the differences in z scores between the highest and lowest wealth quartile in each cohort. In all 4 cohorts, children in the lowest wealth quartile had significantly lower LAZ scores than children in the highest wealth quartile, with differences ranging from 0.2 to 0.3 SD. Significant differences between the lowest and highest wealth quartiles were found for motor development in the iLiNS-ZINC (0.3 SD) and iLiNS-DOSE (0.4 SD) cohorts, and for language development in the iLiNS-DOSE (0.4 SD) and iLiNS-DYAD-G (0.2 SD) cohorts. No significant differences between wealth quartiles were found for personal-social, socioemotional, or executive function.
Table 3 presents the estimates of the association of early growth status (LAZ at 6/9 months) and growth in later infancy (ΔLAZ from 6/9 to 18 months) with each developmental score. In all cohorts, both measures of linear growth were significantly associated with language and motor development, and neither measure was associated with socioemotional development. Both measures of linear growth were significantly associated with personal-social development, which was only measured in the iLiNS-ZINC cohort. Executive function was not associated with early growth status in any cohort and was associated with growth from 6 to 18 months only in the iLiNS-DOSE cohort.
For language, motor, and personal-social development, the coefficients for early growth status were similar in magnitude to the coefficients for later infant growth. The coefficients for model 1 were similar to the coefficients for model 2, indicating that the household and family factors included in model 2 did not greatly confound the associations between linear growth and development. The associations of both measures of growth with language and motor development were somewhat smaller in Ghana, compared with Malawi and Burkina Faso. I2 statistics indicated substantial heterogeneity between trials for the motor and language coefficients; however, when combining only 3 trials, heterogeneity between trials cannot be estimated with precision, and the confidence intervals were very large. The coefficients were not significantly different between trials except for the association between 6-month LAZ and language score (for the interaction with trial, P = .04; all other P values >.1). The adjusted estimates pooling the data from Ghana and Malawi indicate that a 1 SD difference in 6-month LAZ was associated with a 0.13 SD difference in language scores and a 0.16 SD difference in motor scores. Each 1 SD difference in ΔLAZ from 6 to 18 months was associated with a difference of 0.11 SD in language scores and 0.22 SD in motor scores. In the random effects models, which accounted for heterogeneity, the estimates were similar and the P values were larger; the confidence intervals took into account the variance both within and between trials, and the variance between only 3 trials cannot be estimated with precision.27
The analysis of the 3 measures of growth (LAZ at birth, ΔLAZ from birth to 6 months, and ΔLAZ from 6 to 18 months) in the iLiNS-DYAD cohorts revealed a similar pattern (Table 4). Significant associations were found for language and motor development but not for socioemotional development or executive function. The coefficients for the 3 measures of growth were similar to each other, except growth from birth to 6 months in Ghana, which was not significantly related to any developmental score. However, the coefficients were not significantly different between Ghana and Malawi (for the interactions with trial, all P values >.1). The cross-sectional estimates of the associations between 18-month LAZ and each developmental score are reported in Supplemental Table 12.
We examined the potential effect modifiers only for the 3 developmental scores for which significant associations with growth were found: language, motor, and personal-social scores. No significant interactions were found between LAZ at 6/9 months and ΔLAZ from 6/9 to 18 months (P values >.1). Three interactions were found with the FCI score: in the pooled data, the interaction with LAZ at 6 months for motor development (P = .07) and in the iLiNS-ZINC cohort, the interaction with LAZ at 9 months for motor development (P = .09) and with ΔLAZ from 9 to 18 months for language development (P = .04). Table 5 displays these associations between growth and development stratified according to FCI tertile. As expected, stronger associations were found in the lowest tertile of FCI score, compared with the mid and high tertiles.
Two significant interactions were found with trial group: for language development, the interaction with 6-month LAZ in the iLiNS-DYAD-M (P = .02) and iLiNS-DYAD-G (P = .08) cohorts. Table 6 displays these associations according to trial group. In both iLiNS-DYAD cohorts, the association between 6-month LAZ and language development was significant in the iron and folic acid group but not in the LNS group. In Ghana, there was also no association in the multiple micronutrients (MMN) group.
We examined associations between linear growth from birth to 18 months and 4 domains of child development in 3700 children in 4 cohorts in Burkina Faso, Ghana, and Malawi. With regard to the extent to which children in these cohorts had faltered in growth by age 18 months, this factor ranged from 0.8 SD below the mean of the WHO standard in Ghana to 1.8 SD below the mean in the iLiNS-DOSE cohort in Malawi. In contrast, the difference in attained length between the highest and lowest wealth quartile within each cohort was only 0.2 to 0.3 SD. This finding suggests that even the most economically advantaged children within these samples were faltering in linear growth compared with global standards.
Conclusions are more difficult to draw regarding the extent to which children in these cohorts had faltered in development because no global standards exist. The socioeconomic disparities in motor development in the iLiNS-DOSE and iLiNS-ZINC cohorts and in language development in the iLiNS-DOSE and iLiNS-DYAD-G cohorts were similar to the disparities in linear growth (0.2 to 0.4 SD). This finding suggests that for these cohorts and developmental domains, economically disadvantaged children were faltering to a similar extent as in linear growth. The socioeconomic disparities for all other cohorts and domains were smaller and nonsignificant, which suggests that for these cohorts and domains, even the most disadvantaged children were not greatly falling behind their more advantaged peers. However, if the most advantaged children were not achieving their potential in linear growth, it is possible that they also were not achieving their potential in development. Even the subsample with the fewest environmental constraints on development, which included only 158 of 3700 children, had faltered in linear growth 0.7 SD below the WHO mean by age 18 months. This low-risk sample scored on average within the same range in motor, socioemotional, and executive function as the 4 cohorts. Their median vocabulary score was 40 words, compared with a median of 25 to 28 words in each cohort. A difference of 12 to 15 words is equivalent to ∼0.5 to 0.8 SD. These findings suggest that with fewer constraints on development, a higher level of vocabulary could have been achieved.
Linear growth was associated with language, motor, and personal-social development but not socioemotional development or executive function, with the exception that the association between ΔLAZ from 6 to 18 months and executive function was significant in the iLiNS-DOSE cohort in Malawi. This finding suggests that from birth to 18 months, children’s growth in height is generally related to the development of language, motor, and personal-social skills (feeding, dressing, and toilet training) but not executive function or socioemotional skills (paying attention, regulating anger and aggression). This lack of association is consistent with 1 study that found no association between child height and scores on the A-not-B task in 15-month-old children in Uganda23 and 1 study that found no association with attention in 6- to 9-month-old infants.28 Few studies examining linear growth in children aged <2 years have assessed socioemotional or executive function, partly because these factors are difficult to assess at this age. Children aged <2 years have only begun to develop these skills, and few assessments exist, particularly for executive function, whereas even fewer have been evaluated for psychometric properties, such as predictive validity.29 A few studies in older children have reported associations between socioemotional skills and height-for-age, but even in older children, few studies of linear growth have examined these domains.6
For language development, the pooled adjusted estimate of the association with LAZ at 6 months was 0.13 SD; with ΔLAZ from 6 to 18 months, it was 0.11 SD (model 2). For motor development, these estimates were 0.16 SD and 0.22 SD, respectively. The sums of these estimates in each domain were the same as the cross-sectional pooled adjusted estimates reported by Sudfeld et al6 for children aged <2 years, which were 0.24 SD in the cognitive domain and 0.38 SD in the motor domain. When we examined growth status separately at birth and from birth to 6 months, both growth measures showed significant associations with language and motor development in Malawi; in Ghana, only the associations for LAZ at birth were significant. Together, these findings show that growth faltering during any period before birth, from birth to 6 months, and from 6 to 18 months is similarly associated with the motor and language skills children have attained by 18 months, suggesting that no period is more important than another but rather that this entire period is important for intervention.
Associations between all measures of growth and development were similar with and without adjusting for maternal age and education, household asset index and food insecurity, HIV status, and developmental stimulation in the home. Thus, we found no evidence that linear growth is a proxy for these aspects of the environment but rather the associations between growth and development remained consistent when controlling for these factors. However, it is possible that other unmeasured factors confound the association. Convincing evidence for causality would be an intervention that positively affects both growth and development, in which the effect on development is mediated by the effect on growth. We are unaware of such a study. In the iLiNS-ZINC trial, the intervention positively affected both linear growth9 and development11; however, the effect on development was not mediated by the effect on growth.30 The finding that the intervention had independent effects on growth and development suggests that these effects occurred through different mechanisms and does not support a causal link between growth and development.
Although the associations of linear growth with language and motor development were remarkably consistent across sites, they were weaker or absent in certain subsamples of children, specifically children with a high level of developmental stimulation from the environment and children whose mothers received LNS or MMNs during pregnancy. The lack of association between growth and aspects of development in these subsamples is consistent with samples of children with few environmental constraints on growth.14 This scenario suggests that both developmental stimulation and maternal supplementation can protect children from the commonly observed association between growth faltering and poorer language and motor skills. The interaction with maternal supplementation was only evident for the associations with growth in early infancy (age 6 months). We found no evidence that child supplementation with LNS attenuated the association between linear growth and development from 6 to 18 months. We also found no evidence that sufficient growth during an earlier time period (before 6/9 months) protected children from an association between growth faltering and poor development in later infancy. This outcome further supports the conclusion that healthy growth during each period before 6 months and from 6 to 18 months is similarly associated with motor and language development.
Linear growth faltering from birth to 18 months is associated with deficits in the development of language, motor, and personal-social skills but not socioemotional or executive function skills, at least as measured in this study. Linear growth during different periods (before birth, in early infancy, and in later infancy) is similarly associated with the development of these skills, indicating that children who falter in growth during any of these periods have poorer motor and language skills at age 18 months. Developmental stimulation from the environment and maternal supplementation with LNS or MMNs may protect children who are faltering in growth from poor development of language and motor skills.
We thank the families and communities who participated in the iLiNS trials and the iLiNS teams who executed the studies. Rosemonde Guissou, Zinewendé Ouédraogo, Harriet Okronipa, Martin Ndelemani, Thokozani Phiri, Nozgechi Phiri, Chiza Kumwenda, Jaden Bendabenda, and Andrew Matchado contributed to the coordination of the studies. Boateng Bannerman, Joy Thakwalakwa, Lotta Alho, and Basho Poelman contributed to data cleaning and database management. Charles Arnold and Rebecca Young provided statistical support. Mamane Zeilani served on the iLiNS project Steering Committee.
- Accepted May 24, 2016.
- Address correspondence to Elizabeth L. Prado, PhD, Program in International and Community Nutrition, University of California at Davis, 3253 Meyer Hall, One Shields Ave, Davis, CA 95616. E-mail:
The findings and conclusions contained within are those of the authors and do not necessarily reflect positions or policies of the Bill & Melinda Gates Foundation.
Dr Prado designed data collection instruments, conceptualized the analyses of linear growth and development, conducted data analysis, and drafted the manuscript; Drs Abbeddou, Phuka, Somé, and Yakes Jimenez and Mss Arimond and Ocansey designed the data collection instruments, coordinated and supervised data collection, and critically reviewed the manuscript; Drs Adu-Afarwuah, Ashorn, Ashorn, Brown, Hess, Lartey, Maleta, Ouédraogo, Vosti, and Dewey designed and supervised the iLiNS trials, designed the data collection instruments, served on the iLiNS Project steering committee, and critically reviewed the manuscript; and all authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: Dr Brown worked as a consultant and later as an employee for the Bill & Melinda Gates Foundation. The other authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Based on research funded in part by a grant to the University of California Davis from the Bill & Melinda Gates Foundation, with additional funding from the Office of Health, Infectious Diseases, and Nutrition, Bureau for Global Health, US Agency for International Development under terms of Cooperative Agreement No. AID-OAA-A-12-00005, through the Food and Nutrition Technical Assistance III Project, managed by FHI 360. The sponsors of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Adu-Afarwuah S,
- Lartey A,
- Okronipa H, et al
- Ashorn P,
- Alho L,
- Ashorn U, et al
- Hess SY,
- Abbeddou S,
- Jimenez EY, et al
- Maleta KM,
- Phuka J,
- Alho L, et al
- Prado EL,
- Abbeddou S,
- Yakes Jimenez E, et al
- Prado EL,
- Adu-Afarwuah S,
- Lartey A, et al
- Prado EL,
- Maleta K,
- Ashorn P, et al
- Watanabe K,
- Flores R,
- Fujiwara J,
- Tran LT
- WHO Multicentre Growth Reference Study Group
- Fenson L,
- Marchman VA,
- Thal D,
- Dale PS,
- Reznick JS,
- Bates E
- Copyright © 2016 by the American Academy of Pediatrics