Prenatal and postnatal lipid-based nutrient supplementation and
cognitive, social-emotional, and motor function in preschool-aged
children in Ghana: a follow-up of a randomized controlled trial
Maku E Ocansey,1 Seth Adu-Afarwuah,2 Sika M Kumordzie,1 Harriet Okronipa,1 Rebecca R Young,1 Solace M Tamakloe,2
Brietta M Oaks,1,3 Kathryn G Dewey,1 and Elizabeth L Prado1
1Program in International and Community Nutrition, Department of Nutrition, University of California, Davis, CA; 2Department of Nutrition and Food Science,
University of Ghana, Legon, Ghana; and 3Department of Nutrition and Food Sciences, University of Rhode Island, Kingston, RI
ABSTRACT Keywords: cognitive development, preschool development, lipid-
Background:Adequate nutrition is necessary for brain development based nutrient supplementation, multiple micronutrients, prenatal
during pregnancy and infancy. Few randomized controlled trials supplementation, postnatal supplementation
of supplementation during these periods have measured later
developmental outcomes.
Objective: Our objective was to investigate the effects of provision
of prenatal and postnatal lipid-based nutrient supplements (LNS) on Introduction
child development at preschool age. The initial years of life are critical for the formation of brain
Methods: We conducted a follow-up study of 966 children aged 4– structure and capacity (1, 2). Neurodevelopmental processes
6 y in 2016, born to women who participated in the International occur rapidly during gestation and the first 2 y of life. Adequate
Lipid-Based Nutrient Supplements-DYAD trial conducted in Ghana nutrition is important to support these processes and for the
in 2009–2014, representing 79% of eligible children. Women ≤20 long-term development of cognitive, motor, and social-emotional
weeks of gestation were randomized to daily LNS or multiple skills. In animal models, gestational and early postnatal nutrient
micronutrient (MMN) capsules during pregnancy through 6 mo deficiencies result in impairments such as reduced and truncated
postpartum or iron and folic acid (IFA) capsules during pregnancy dendritic aborization, and alterations in myelin composition
and calcium placebo capsules during 6 mo postpartum. Children in and synapse structure (2). In humans, many studies have
the LNS group received LNS from 6 to 18 mo. Primary outcomes shown associations between indicators of undernutrition and
of this follow-up study were (1) a cognitive factor score based on micronutrient deficiency, such as stunted growth and anemia,
a test battery adapted from several standard tests, 2) fine motor and developmental and cognitive function in early infancy and
score (9-hole pegboard test), and (3) social-emotional difficulties childhood (2). Results from randomized controlled trials (RCTs)
(Strengths and Difficulties Questionnaire; SDQ). Eight secondary of the effects of supplementation with specific micronutrients and
outcomes were calculated in specific domains (e.g., language, SDQ
prosocial). Analysis was by a complete case intention to treat in a
2-group comparison: LNS compared with non-LNS (MMN + IFA). This study was funded by a grant to the University of California, Davis,
Results: Children in the LNS group had significantly lower social- from the Bill & Melinda Gates Foundation, and the Society for Research in
emotional difficulties z-scores than children in the non-LNS group Child Development, 2015 Patrice L Engle Dissertation Grant in Global Early
(adjusted for child age β =−0.12, 95% CI:−0.25, 0.02, P= 0.087; Child Development.
fully adjusted β = −0.16, 95% CI: −0.29, −0.03, P = 0.013). The SupplementalMethods and Supplemental Tables 1–5 are available from the
“Supplementary data” link in the online posting of the article and from the
effect of LNS on social-emotional difficulties score was larger among
same link in the online table of contents at https://academic.oup.com/ajcn/.
children living in households with lower home environment scores Address correspondence to MEO (e-mail: emocansey@ucdavis.edu).
(P-interaction = 0.081). No other outcomes differed between the 2 Abbreviations used: ALA, α-linolenic acid (18:3n−3); HOME, Home
intervention groups. Observation for theMeasurement of the Environment Inventory; IFA, iron and
Conclusions: Provision of LNS during the first 1000 d of folic acid; iLiNS, International Lipid-BasedNutrient Supplement; LNS, lipid-
development improved behavioral function, particularly for children based nutrient supplement;MMN,multiple micronutrients; RCT, randomized
from low nurturing and stimulation households, but did not affect controlled trial; SDQ, Strengths and Difficulties Questionnaire.
cognition at preschool age in this setting. Trial Registration: Received April 13, 2018. Accepted for publication October 3, 2018.
clinicaltrials.gov, Identifier NCT00970866. Am J Clin Nutr First published online February 5, 2019; doi: https://doi.org/10.1093/
2019;109:322–334. ajcn/nqy303.
322 Am J Clin Nutr 2019;109:322–334. Printed in USA. © 2019 American Society for Nutrition.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which
permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Prenatal and postnatal LNS and child development 323
fatty acids during pregnancy and infancy on child development of providing 2 types of multiple micronutrient supplements
have, however, been mixed (2–5). compared with iron and folic acid, for preventing malnutrition
Home fortification interventions including the use of multiple in pregnant and postpartum women and their infants, and
micronutrient powders and small-quantity lipid-based nutrient evaluated their effects on maternal nutritional status, child
supplements (LNSs), have been evaluated to assess their potential growth, micronutrient status, and neurobehavioral development
to ameliorate the negative effects of undernutrition (6) on child at age 18 mo. At the time of the original trial, iron and folic
growth and development in low- and middle-income countries. acid supplementation was the standard practice and WHO and
Although 5 RCTs of supplementation with LNS during the Ghana Health Service recommendation for antenatal care in
postnatal period (7–11) and 3 RCTs providing LNS during both Ghana, and multiple micronutrient supplements were already
prenatal and postnatal periods (12–14) have reported effects being evaluated as a likely alternative standard of care in many
on developmental outcomes, results in early childhood up to countries. A detailed description of the design and methods of
age 2 y have been mixed, and none has examined longer-term the original trial (clinicaltrials.gov; NCT00970866) has been
developmental effects of LNS. Recently, we reported the results published elsewhere (15). In brief, pregnant women attending
of the International Lipid-Based Nutrient Supplements (iLiNS) antenatal clinics in 4 health facilities in the area at ≤20 weeks
DYAD trial in Ghana showing that provision of LNS to women of gestation were recruited into the study if they were ≥18 y old
from pregnancy to 6 mo postpartum and to their infants from 6 to and agreed to participate by signing or thumb-printing informed
18 mo positively affected linear growth (15), but did not affect consent after screening. Exclusion criteria were HIV infection,
motor, cognitive, or social-emotional development at age 18 asthma, epilepsy, tuberculosis, any malignancy, known milk or
mo (13). However, global behavioral developmental assessments peanut allergies, intention to move from the study area during the
before age 2 y may not be sensitive enough to detect effects. study period, unwillingness to receive fieldworkers or take study
For example, a group of children who experienced thiamine supplement, participation in another trial, or gestational age>20
deficiency in infancy did not show neurological symptoms at the wk before completion of the enrollment process.
time of deficiency, but showed language impairment at age 5–7 A total of 1320 pregnant women were randomly assigned
y (16). Similarly, in an RCT in the United States, infants who to 1 of 3 intervention arms daily from enrollment to delivery:
received formula containing DHA and arachidonic acid showed 1) 60 mg of iron plus 400 μg of folic acid [iron and folic
higher vocabulary and IQ scores at age 5–6 y than infants who acid (IFA) group: n = 441]; 2) multiple micronutrient capsule
received formula without these fatty acids, even though they did containing 18 vitamins and minerals [multiple micronutrients
not differ in vocabulary or Bayley Scales of Infant Development (MMN) group: n= 439]; and 3) LNS with similar micronutrients
scores at age 18 mo (17). These examples show that later effects as the MMN supplement, plus other minerals and macronutrients
of supplementation on cognitive and behavioral development (LNS group: n = 440) (15). The nutrient and energy contents of
may be observed, even if such effects were not detected on the supplements provided in the main trial are shown in Table 1.
global developmental measures at a younger age. Here, we report After birth, the MMN and LNS groups continued to receive the
a follow-up study assessing the effects of the intervention on same supplements until 6 mo postpartum, whereas the control
cognitive, motor, and social-emotional development at age 4– IFA group received calcium placebo capsule (200 mg/d) during
6 y, using a comprehensive battery of cognitive, motor, and that period. Children in the LNS group received LNS designed
social-emotional tests adapted in the local Ghanaian setting. To for children from 6 to 18 mo of age, whereas children in the other
our knowledge, our study is the first long-term follow-up of an 2 groups received no supplement.
RCT of both prenatal and postnatal LNS supplementation. We
are aware of only 2 other RCTs that have conducted long-term
follow-up assessment following nutritional supplementation in Follow-up study.
both the prenatal and postnatal periods, 1 of which randomized
only 4 villages to intervention and control groups (18) and 1 of Between January and December 2016, all parents or caregivers
which had a high rate of attrition, re-enrolling only 55% of the of children who had participated in the iLiNS-DYAD-Ghana trial,
original sample in the follow-up study (19, 20). including those who had relocated from the study site, for whom
residential information and/or telephone numbers were available,
were contacted for enrollment in the follow-up study. During
Methods this period, children who had been born during the trial were4–6 y of age. Figure 1 shows the trial profile. We re-enrolled
Study design and participants 1014 children whose mothers or caregivers provided informed
consent to participate, and obtained developmental data from 966
The study reported here was a follow-up study of children (79% of 1222 children whose vital status at 18 mo was alive or
and mothers who participated in the iLiNS-DYAD-Ghana unknown). We excluded children if they had moved to a new
randomized trial (15). location where a round trip to the study site would cost >60
Ghana cedis (∼US$15) at the time. In Ghana, these were children
who could not make a round trip to the study site and complete the
Design of the original trial. neurobehavioral assessments on the same day. Ethical approval
Between 2009 and 2014, the iLiNS-DYAD-Ghana trial was for this follow-up study was obtained from the ethics committees
conducted in semiurban communities in the Yilo and Manya of the University of California, Davis, the Ethics Committee for
Krobo Districts of the Eastern Region in Ghana, located about the College of Basic and Applied Sciences of the University of
70 km north of the capital, Accra. The trial tested the efficacy Ghana, and the GhanaHealth Service Ethical ReviewCommittee.
324 Ocansey et al.
TABLE 1 Nutrient and energy contents of the supplements used in the International Lipid-Based Nutrient Supplement Dyad-Ghana randomized controlled
trial in Ghana1
Chemical form used in
Nutrient supplement formulation IFA MMN Maternal LNS Child LNS
Ration per day — 1 tablet 1 tablet 20-g sachet 20-g sachet
Total energy, kcal — — 0 118 118
Protein, g — — 0 2.6 2.6
Fat, g — — 0 10 9.6
Linoleic acid, g — — 0 4.59 4.46
α-Linolenic acid, g — — 0 0.59 0.58
Vitamin A, μg RE Retinyl acetate — 800 800 400
Vitamin C, mg l-Ascorbic acid — 100 100 30
Vitamin B1, mg Thiamin hydrochloride — 2.8 2.8 0.3
Vitamin B2, mg Riboflavin — 2.8 2.8 0.4
Niacin, mg Niacinamide — 36 36 4
Folic acid, μg Pteroyl monoglutamic acid 400 400 400 80
Pantothenic acid, mg Calcium pantothenate — 7 7 1.8
Vitamin B6, mg Pyridoxine hydrochloride — 3.8 3.8 0.3
Vitamin B12, μg Cyanocobalamin (0.1%) — 5.2 5.2 0.5
Vitamin D, mg Cholecalciferol (D3) — 10 10 5
Vitamin E, mg dl-α-Tocopherol acetate — 20 20 6
Vitamin K, μg Phylloquinone 5% — 45 45 30
Iron, mg Encapsulated ferrous sulfate 60 20 20 6
Zinc, mg Zinc sulfate — 30 30 8
Cu, mg Encapsulated copper sulfate — 4 4 0.34
Calcium, mg Tricalcium phosphate — 0 280 280
Phosphorus, mg Dipotassium phosphate — 0 190 190
Potassium, mg Potassium chloride — 0 200 200
Magnesium, mg Magnesium citrate — 0 65 40
Selenium, μg Sodium selenite 1.5% — 130 130 20
Iodine, μg Potassium iodate — 250 250 90
Manganese, mg Manganese sulfate — 2.6 2.6 1.2
1IFA, iron and folic acid capsule; MMN, multiple micronutrient supplement capsule; LNS, lipid-based nutrient supplement for pregnant and lactating
women. Information from table previously published (13).
Data-collection procedures additional home visit was conducted by 8 trained fieldworkers
At enrollment into the original trial, maternal and household to interview parents or caregivers about the child’s behavior
information including maternal age, birth order, education, and and to assess nurturing and stimulation provided from the home
household assets was collected by trained fieldworkers using a environment. A laboratory visit by mothers and children was
questionnaire. Trained laboratory personnel measured maternal also done for biological sample collection. Data collection for a
blood hemoglobin concentration at a scheduled clinic visit using mother–child dyad in the follow-up study typically took 6–8 wk
a digital Hemocue (HemoCue model 301, AG, Switzerland) . to complete.
Field staff visited homes to explain the follow-up study,
obtain consent for participation, and collect sociodemographic
information. A second home visit was conducted typically Developmental assessment measures
within 7 d to obtain reports on child health and household We assessed neurobehavioral development by several mea-
investments and to invite participants to visit the test center sures, as described in Table 2. Language ability was assessed by
for anthropometric, body composition, and neurobehavioral the body-part naming and identification and comprehension of
assessments. Phone interviews were conducted beforehand to instructions subtests of the Developmental Neuropsychological
ascertain that the child was in good health, and appointments Assessment II (21). Preacademic skills were assessed using
were rescheduled in cases of reported illness from parents or the Parent’s Evaluation of Developmental Status developmental
caregivers. Procedures at the test center typically involved a milestones test (22) administered by trained testers rather than by
4-h visit including a snack break and lunch. Neurobehavioral parent report. Executive function was assessed using the head–
assessment took about 1 h and 30 min. toe, delay of gratification, and visual search tasks. The head–
For neurobehavioral assessment, mothers and children were toe task was drawn from the International Development and
assessed privately in a test room to reduce distractions. In addition Early Learning Assessment (23), the delay of gratification from
to the snack break, children were given short breaks in between Noble (24), and the visual search task from the Supplementation
test sessions to reduce tiredness. Five trained data collectors who withMultipleMicronutrients Intervention Trial (25).We assessed
were blind to treatment group conducted the neurobehavioral visuospatial ability using a block design test based on the
assessments, with 1 data collector assessing 1–7 children/d. An British Ability Scales II pattern construction subtest (26) and
Prenatal and postnatal LNS and child development 325
FIGURE 1 Study profile showing infants whose mothers were enrolled into the trial, and the reasons some infants were lost to follow-up. IFA, iron and
folic acid; LNS, lipid-based nutrient supplement; LNS group, women received 20 g LNS daily during pregnancy and 6 mo postpartum; infants received 20
g LNS daily from 6 to 18 mo of age; MMN, multiple micronutrients; Non-LNS group, women received either IFA during pregnancy and placebo for 6 mo
postpartum or MMN capsules during pregnancy and 6 mo postpartum. Infants did not receive any supplement. Groups shown are based on supplements women
received at enrollment.
326 Ocansey et al.
TABLE 2 Motor, cognitive, and social-emotional measures of assessment
Developmental
domain Developmental test Test description and scoring
Motor
Fine motor NIH Toolbox 9-Hole Pegboard We recorded the time required for children to accurately place and remove 9 plastic
pegs from a pegboard, first with the dominant hand, followed by the other hand. The
score was the mean time in seconds taken to complete the task with each hand.
Cognitive
Language ability Developmental Neuropsychological Children were asked to say aloud or point to body parts on a line drawing of a person or
Assessment II Body Part Naming on their own body. The 2 scores were the sum of body parts correctly named and
and Identification identified.
Developmental Neuropsychological Children were instructed to point to a picture, e.g., “Show me a puppy that is big and
Assessment II Comprehension of blue and happy.” The score was the number of items indicated correctly.
Instructions
Visuospatial ability Block Design Children were asked to copy increasingly complex patterns of models built by the
instructor, using wooden block in 30 s. The score was the number of structures
correctly copied within the time limit.
Declarative memory Paired-Associate Learning and Children were first taught new words for pictures of 8 objects and were asked to point
Recall Task to them as the instructor mentioned them aloud. They were later asked to recollect
the words learned after a delay of median (IQR) 7 (6–11) min. We calculated the
score as the mean number of correct responses on a set of 8 learning trials and 2
delayed recall trials.
Executive function
Visual selective Visual Search Test Based on the Developmental Neuropsychological Assessment subtest adapted in the
attention Supplementation with Multiple Micronutrients Intervention Trial, children were
asked to identify all instances of a target picture (chicken or kitten) printed on a page
with other distracter pictures as quickly as possible in 2 min. The score was the total
time in seconds per correct target item identified.
Inhibition/reward Delay of Gratification The test was carried out 3 times for each child in between other tests. After each test,
children were asked to make the choice of having 1 candy out of a tray of colorful
candies immediately or having 2, 3, or 4 at the end of the second, third, and fourth
test, respectively. Scoring was based on the number of candies chosen.
International Development and Early Children were asked to inhibit the normal response to touch their head, when instructed
Learning Assessment Head/Toes to do so, by touching their toes instead. This was repeated 5 times, interchanging the
Test touch head or toes instruction in a particular order. The score was the sum of correct
responses.
Pre-academic skills Parent’s Evaluation of In 14 items, children were asked to perform skills such as counting, reading aloud
Developmental Status words, or identifying letters of the alphabet. The score was the total of correct
pre-academic subscale responses.
Social-emotional
Behavior regulation Behavior Rating Scale Based on the Preschool Self-Regulation Assessment Data Collector Report. The data
collector assessed children’s behavior or temperament at the test center and filled
reports immediately after the test session for each child. Items on children’s
emotions, attention, behavior, defiance, noncompliance, and anxiety throughout the
data collector–child interaction were scored on a Likert scale ranging from 0 to 3
(e.g., child is cooperative; child waits patiently for new tasks to begin) and some
items were reverse-coded to minimize automatic responding. The score was the sum
of data collector ratings on the 10-item scale.
Psychosocial and Strengths and Difficulties Parents or caregivers were asked during an interview to describe their child’s behavior
prosocial Questionnaire within the past 6 mo, based on a set of 25 questions divided between 5 scales: 1)
characteristics emotional symptoms, 2) conduct problems, 3) hyperactivity/inattention, 4) peer
relation problems, 5) prosocial behavior. Responses were scored on a Likert scale
ranging from 0 to 2 (not true, somewhat true, and certainly true, respectively).
Attributes 1–4 were summed up to generate a “Total difficulties score.”
the Wechsler Primary and Preschool Scale of Intelligence block screening questionnaire for child mental health problems in
design subtest (27). We assessed declarative memory using a children aged 3–16 y. The SDQ comprises 25 items on
paired associate memory task from Baddeley and colleagues psychological attributes divided into 5 subscales: emotional
(28). Motor function was assessed by the NIH Toolbox 9-hole symptoms, conduct problems, hyperactivity/inattention, peer
pegboard test (29). We assessed social-emotional competence relations problems, and prosocial behavior. We calculated the
by caregiver interview using the Strengths and Difficulties total difficulties score, which assesses behavioral problems,
Questionnaire (SDQ) (30, 31), which is a widely used brief as the sum of the first 4 subscales. We additionally assessed
Prenatal and postnatal LNS and child development 327
social-emotional development by direct observation using the be involved in the follow-up at age 4–6 y, or approximately 316
Behavior Rating Scale adapted from the School Transition and per intervention group (LNS, MMN, IFA). With this sample size
Readiness study (32) based on the Preschool Self-Regulation and a power of 80% at a 0.05 level of significance, we expected
Assessment data collector report (33). The tests were adapted to be able to detect a difference of ≥0.25 SD in each of the 3
to the local setting in Ghana and evaluated for test–retest and continuous primary outcomes.
internal reliability through 2 rounds of pilot studies conducted in We posted a statistical analysis plan with prespecified po-
the study area before the follow-up study. For details of the test tential covariates and effect modifiers to the project website
selection criteria and test adaptation process, see Supplemental (www.ilins.org) before study investigators were unblinded to
Methods. The test–retest reliability of the adapted tests ranged children’s intervention group assignments. All analyses were
from r= 0.61 to 0.94, and internal reliability (Cronbach’s alpha) conducted using SAS version 9.4 (SAS Institute).
ranged from 0.61 to 0.9. We examined whether children in the 3 intervention groups
were similar regarding a number of baseline and other character-
istics, using ANOVA for continuous variables and chi-square for
Additional measures at follow-up categorical (binary) variables. We also examined differences in
baseline characteristics between children tested at follow-up and
We assessed the stimulation or caregiving available to children those lost to follow-up.
in the home environment using the Early Childhood version of We calculated z-scores for each test score based on the
the Home Observation for the Measurement of the Environment distribution of scores in our sample. All z-scores were computed
(HOME) Inventory (34), which we adapted to the local context. in 3-mo age bands, with a mean of 0 and a SD of 1 in each age
The test–retest reliability was 0.63. band. Standard norming guidelines state that a sufficient sample
Maternal depressive symptoms were assessed using the size is 75–200 per age group. Age bands are expected to be
Edinburgh Postnatal Depression Scale, which has been validated smaller in the first year of age (1 mo) and larger at later ages (2–3
among both postpartum and nonpostpartum women (35). mo for toddlers and 6 mo or 1 y for school children) (37). Using
We used the General self-efficacy scale (36) to assess the 3-mo age bands, our sample included about 100 per age group,
general sense of perceived self-efficacy among mothers or ranging from 4.25 y to 6.53 y.
caregivers. The 3 primary outcomes were the cognitive, motor, and social-
emotional domain scores. We calculated an overall cognitive
factor score as the first factor of a factor analysis using the
Training of data collectors and quality control
principal-axis factoring method including 7 outcome measures
During the 1-y data-collection period, we conducted quarterly listed in Table 2: body part naming and identification, comprehen-
knowledge and practice-based evaluations to ensure that data sion of instructions, preacademic skills, visual search, head–toe,
collectors were standardized. At the beginning and during the block design, and paired associate memory scores, comprising all
last quarter of data collection, we evaluated the inter-rater cognitive z-scores except the delay of gratification score, which
agreement. For each of the 5 trained data collectors, we video- was the only score that was not strongly associated with the other
recorded 1 child testing session and 1 caregiver interview. All scores. The motor domain score was calculated as the mean of
trained data collectors and their supervisors watched each video the NIH Toolbox 9-hole pegboard scores for the dominant and
and independently scored the test or interview. For each data nondominant hands. The social-emotional domain score was the
collector, the percentage of item scores that agreed with the total difficulties score from the SDQ.
supervisor was calculated. Overall agreement was calculated as Eight secondary outcomes were also calculated. We calculated
the average agreement across data collectors for each test or z-scores for the following: language, declarative memory,
interview. The inter-rater accuracy was high (>90%) for all tests visuospatial ability, prosocial skills, delay of gratification, head–
except the visual search task (74%), due to slight differences toe inhibition, preacademic skills, and behavior rating scale. We
between data collectors and their supervisor in regulating timers. estimated the prevalence of severe and moderate to severe delay
The mean time to complete the task was 80 s, and the mean as the bottom 10% (lowest decile) and 25% (lowest quartile)
time difference between data collectors and the supervisor of our sample, respectively, of scores in each domain. For
was 2.4 s. the HOME inventory questionnaire, 3.6% of item scores were
missing. The method described by Raghunathan and colleagues
(38) was adopted to impute these missing items based on other
Sample size and statistical analyses items in the HOME.
For the follow-up study, we hypothesized that: a) children in We first tested the null hypothesis of no difference between
the LNS group will have better scores on motor, cognitive, and the 3 treatment groups using ANCOVA for continuous outcomes
social-emotional function tests at preschool age than children in (each domain z-score) and logistic regression for binary out-
the MMN or IFA groups, and b) the percentage of children with comes. For categorical multilevel or non-normally distributed
severe and moderate to severe delays in motor, cognitive, and count outcomes, we used ordinal probit regression and the
social-emotional development will be lower in the LNS group negative binomial regression model, respectively, for analysis.
at preschool age than in the MMN or IFA groups. We estimated For the ordinal probit regression, predicted probabilities were
the effect size based on the number of mother–child pairs who estimated to describe the relation between different levels of the
completed the main trial (n = 1185) and presumed that attrition response variable. Probabilities modeledwere cumulated over the
or refusal for the follow-up activities will be no more than 20% lower ordered values, assuming the same relation exists between
of that number. We expected that at least 948 participants would sequential levels.
328 Ocansey et al.
For all analyses, post-hoc pairwise comparisons of the children (79% of 1222 eligible children and 73% of the 1320
3 intervention groups were performed using Tukey–Kramer women enrolled) (Figure 1). Children with neurobehavioral
adjustment. We defined significant pairwise comparisons as P assessment data did not differ significantly in most background
< 0.05. If there were no significant differences between the IFA characteristics from those lost to follow-up. However, mothers
and MMN groups, we combined these 2 groups into a non-LNS of children included in this analysis were less likely to be
(control) group to estimate 2-group comparisons (LNS compared nulliparous at enrollment (P = 0.032), and they had higher self-
with non-LNS). The premise for combining the 2 control groups reported adherence to supplement use throughout pregnancy up
was that only the children in the LNS group were provided with to 6 mo postpartum (P < 0.0001) (Supplemental Table S1).
supplements directly, from 6 to 18 mo of age. The proportion of children lost to follow-up was significantly
We compared groups using 3 models. The first model was greater in the IFA group (32%) than in the LNS group (23%;
adjusted for child age at follow-up only (model 1). The second P = 0.002) and significantly greater in the non-LNS group (IFA
model was additionally adjusted for gender, developmental group + MMN group) (29%) than in the LNS group (23%;
assessment data collector, and any of the following prespecified P = 0.018).
baseline variables that were significantly associated at the
P < 0.1 level with the outcome in correlation analysis:
maternal age, maternal education, maternal prepregnancy BMI,
maternal hemoglobin concentration, household assets score, and Group characteristics comparisons
parity (model 2). Third, we adjusted for any factors collected We did not find any significant differences between the IFA
after enrollment (birth weight) or at follow-up (preschool and MMN groups in any outcome, so we primarily report the 2-
quality, home stimulation score, maternal agency, and maternal group comparisons combining the IFA and MMN groups, with
depression) that were significantly associated at the P < 0.1 analyses performed by a complete case intention-to-treat. The 3-
level with the outcome in correlation analysis (model 3). group comparisons are presented in Supplemental Table S2.
For any covariates that were collected after baseline, we first The baseline and other selected maternal and child character-
checked whether they were different between groups before istics of the developmental sample are shown in Table 3. There
including them in the model because they could be potential were no significant differences between participants in the 2
mediators. intervention groups in 10 of the 12 background characteristics
We evaluated potential effect modification by 8 prespecified described in Table 3 (P> 0.05), but there were slight differences
maternal (age, education, parity, hemoglobin concentration), in 2 characteristics. Participants in the LNS group were from
household (household assets score, HOME score), and child households with a lower mean asset score (P = 0.021) and were
(gender) variables for each outcome. We tested the interaction less likely to self-report adherence to supplement use during
between each potential effect modifier and intervention group. pregnancy through to 6 mo postpartum that was ≥80% than the
Significant interactions (P < 0.1) were further examined with non-LNS group (P = 0.003).
stratified analyses, or estimation of adjusted intervention group
means at the 10th, 50th, and 90th percentile of the effect
modifier, in order to understand the nature of the effect
modification. Effects of the intervention
Maternal adherence to supplement use was determined by self- Table 4 shows the mean motor, cognitive, and social-
report, with data collected biweekly. In addition, fieldworkers emotional development z-scores in the 2 intervention groups. In
collected any unused LNS sachets at each visit and reconciled the age-adjusted analysis (model 1), there were no significant
the number of sachets remaining since the last visit. We differences overall in any of the 3 primary outcomes measured
calculated adherence as the percentage of follow-up days (e.g., at 4–6 y. However, the LNS group showed a trend for a lower
during pregnancy or from enrollment to 6 mo postpartum) that mean social-emotional difficulties score than the non-LNS group
the supplement was reportedly consumed. We conducted per (P = 0.087). With additional adjustment for baseline covariates
protocol analyses in 2ways: 1) only including children ofmothers (model 2: see footnote to Table 4), we found a significant
who self-reported greater than or equal to 80% adherence (based difference between the 2 groups in social-emotional difficulties
on previous main trial analyses) to supplement consumption (P = 0.044). With adjustment for additional covariates collected
during pregnancy and 2) only including children of mothers who after baseline (model 3: see footnote to Table 4), this difference
self-reported greater than or equal to 80% adherence during the remained significant (P = 0.013). The 3-group comparisons are
period of pregnancy up to 6 mo postpartum. presented in Supplemental Table S3. Results from a post-hoc
exploratory analysis on the 4 subscales that make up the SDQ
total difficulties score are reported in Supplemental Methods.
Results We found no significant differences between groups in
any of the continuous secondary outcomes—namely, language,
Participants at follow-up visuospatial ability, declarative memory, preacademic skills, and
Out of 1320 women enrolled in the original trial, 1222 children behavior rating scale in unadjusted or adjusted analysis (Table 5).
were eligible for re-enrollment in this follow-up study, when The incident rate ratio (95% CI) for the head–toe inhibition task
excluding misdiagnosed pregnancies (n = 5), miscarriages and was 1.03 (0.85–1.25), P = 0.738 (data not shown). For prosocial
stillbirths (n = 66), and children who died before the end of skills, the incident rate ratio (95% CI) was 0.99 (0.95–1.05),
the main trial (n = 27). We re-enrolled 1014 mother–child P= 0.968. This means that the mean counts of the LNS and non-
dyads at follow-up and obtained developmental data from 966 LNS groups were almost equal for both tasks (data not shown).
Prenatal and postnatal LNS and child development 329
TABLE 3 Selected characteristics of women and children by intervention group at baseline and follow-up
LNS1 n = 340 Non-LNS1 n = 626
Variable Mean ± SD (n) or % (n/total) Mean ± SD (n) or % (n/total) P value
Baseline maternal age, y 26.9 ± 5.5 (340) 26.8 ± 5.4 (626) 0.767
Pre-pregnancy BMI2, kg/m2 24.8 ± 4.4 (336) 24.4 ± 4.5 (613) 0.073
Gestational age at enrollment, wk 16.1 ± 3.3 (340) 16.1 ± 3.2 (626) 0.947
Baseline maternal education, y 7.6 ± 3.7 (340) 7.6 ± 3.4 (626) 0.466
Baseline maternal hemoglobin, g/L 111.2 ± 11.3 (340) 111.3 ± 12.4 (625) 0.898
Baseline household asset score3 − 0.09 ± 1.0 (334) 0.06 ± 1.0 (619) 0.021
Nulliparous, % 32.4 (110/340) 32.0 (200/626) 0.943
Gestational age at delivery, wk 39.4 ± 0.10 (338) 39.3 ± 0.1 (623) 0.932
Child male, % 48.2 (164/340) 47.4 (297/626) 0.814
Child age at follow-up, y 5.0 ± 0.0 (340) 4.9 ± 0.0 (626) 0.096
Mean maternal adherence from pregnancy 67.2 (225/335) 76.3 (472/619) 0.003
through 6 mo postpartum (percentage of
supplements consumed)
Home stimulation score at follow-up 27.9 ± 4.5 (329) 27.9 ± 4.9 (599) 0.989
LNS; lipid-based nutrient supplement. Non-LNS; Iron & folic acid + multiple micronutrient capsules (control group).
1Results are based on ANOVA (SAS PROC GLIMMIX) or chi-square (SAS PROC FREQ).
2Estimated pre-pregnancy BMI was calculated from estimated pre-pregnancy weight (based on polynomial regression with gestational age, gestational
age squared, and gestational age cubed as predictors) (39) and height at enrollment.
3Proxy indicator for household socioeconomic status constructed for each household based on ownership of a set of assets (radio, television, etc.),
lighting source, drinking water supply, sanitation facilities, and flooring materials. Household ownership of this set of assets is combined into an index (with a
mean of zero and SD of 1) using principal components analysis. Higher values represent higher socioeconomic status.
There were no significant differences across groups in response when including only children of mothers who self-reported
to the delay of gratification task (Supplemental Table S4). ≥80% adherence to supplement consumption during pregnancy
Estimating the percentage of children in our sample with or during pregnancy through 6 mo postpartum, the pattern of
severe (lowest decile) or moderate-to-severe delay (lowest results was consistent with that shown for the full sample, with
quartile), we found no significant differences between groups significant effects on social-emotional difficulties but not on
in any developmental outcome, except for a lower percentage cognitive and motor scores (data not shown).
of children in the LNS group than in the non-LNS group in
the highest quartile of the social-emotional difficulties score
(LNS 20.8%; non-LNS 27.8%; P = 0.016). This difference Effect modification
remained significant in model 2 and model 3 (Table 4). In For each of the 8 continuous outcomesmeasured, we examined
the 3-group comparisons, these differences were also consistent 8 potential effect modifiers: child sex, HOME score, household
across models (data not shown). In the per protocol analysis asset score, and the following maternal factors collected at
TABLE 4 Primary outcomes: motor, cognitive, and social-emotional z-scores at 4–6 y by intervention group1 and adjusted model
Adjusted for child age at Adjusted for baseline and other
follow-up Adjusted for baseline covariates covariates
Developmental LNS3 Non-LNS3 Difference in mean Difference in mean Difference in mean
domain N2 Mean (95% CI) Mean (95% CI) (95% CI) P value (95% CI) P value (95% CI) P value
Cognitive z-score 951 0.03 (−0.06, 0.13) −0.01 (−0.08, 0.06) 0.04 (−0.08, 0.16) 0.510 0.06 (−0.06, 0.18) 0.3024 0.05 (−0.08, 0.19) 0.4325
Motor z-score 963 0.00 (−0.10, 0.09) 0.00 (−0.11, 0.07) 0.00 (−0.13, 0.12) 0.935 0.06 (−0.12, 0.12) 0.9786 0.03 (−0.09, 0.16) 0.5967
Social-emotional 959 −0.08 (−0.18, 0.03) 0.04 (−0.04, 0.12) −0.12 (−0.25, 0.02) 0.087 −0.14 (−0.27, −0.00) 0.0448 −0.16 (−0.29, −0.03) 0.0139
difficulties z-score
LNS; lipid-based nutrient supplement. Non-LNS; Iron & folic acid + multiple micronutrient capsules (control group).
1We first tested the null hypothesis of no difference between the 3 treatment groups, and combined the iron and folic acid/multiple micronutrients groups because there were
no significant differences between those 2 groups.
HOME, Home Observation for the Measurement of the Environment Inventory.
2Sample size based on model 1 adjusted for child age at follow-up.
3Results are based on ANCOVA (SAS PROC GLIMMIX).
4Adjusted for child age at follow-up, data collector, maternal education, maternal age, household asset score, and maternal hemoglobin.
5Additionally adjusted for exposure to multiple languages, type of preschool, and HOME score.
6Adjusted for child age at follow-up, data collector, and child sex.
7Additionally adjusted for exposure to multiple languages and HOME score.
8Adjusted for child age at follow-up, data collector, maternal education, prepregnancy BMI, maternal hemoglobin, maternal age, and household asset score.
9Additionally adjusted for maternal agency, maternal depression, and HOME score.
330 Ocansey et al.
TABLE 5 Secondary outcomes: Selected developmental outcomes at 4–6 y by intervention group1 and adjusted model
Adjusted for baseline and other
Adjusted for child age at follow-up2 Adjusted for baseline covariates3 covariates4
Developmental LNS6 mean Non-LNS6 mean (95% CI) Difference in mean or Difference in mean or Difference in mean or
domain N5 (95% CI) or OR (n/total) or OR (n/total) OR (95% CI) P value OR (95% CI) P value OR (95% CI) P value Covariates
Language z-score 963 0.00 (−0.08, 0.09) 0.00 (−0.06, 0.06) 0.00 (−0.10, 0.11) 0.936 0.02 (−0.08, 0.12) 0.702 0.01 (−0.09, 0.11) 0.796 A, B, F, G, H, M, N
Visuospatial z-score 963 − 0.02 (−0.08, 0.04) 0.00 (−0.04, 0.04) − 0.02 (−0.09, 0.06) 0.675 − 0.02 (−0.09, 0.05) 0.612 − 0.01 (−0.09, 0.08) 0.853 B, C, G, I, K, N
Preacademic z-score 958 − 0.01 (−0.12, 0.10) 0.00 (−0.08, 0.08) − 0.01 (−0.15, 0.12) 0.852 0.00 (−0.12, 0.13) 0.965 − 0.01 (−0.15, 0.14) 0.927 A, B, C, D, E, G, H,
I, J, K, M, N, O
Declarative memory 958 0.04 (−0.06, 0.14) − 0.03 (−0.10, 0.04) 0.07 (−0.06, 0.19) 0.285 0.08 (−0.04, 0.20) 0.209 0.04 (−0.09, 0.17) 0.568 A, B, E, F, H, I, N
z-score
Behavior rating 962 0.00 (−0.11, 0.10) 0.0 (−0.08, 0.08) − 0.01 (−0.14, 0.12) 0.910 − 0.01 (−0.14, 0.12) 0.922 − 0.01 (−0.15, 0.13) 0.884 B, G, H, I, M, N
z-score
Cognitive lowest 951 10.2 (34/335) 10.1 (62/616) 0.99 (0.64, 1.55) 0.972 0.95 (0.60, 1.51) 0.834 1.05 (0.59, 1.86) 0.880 B, E, G, J, L, N
decile
Cognitive lowest 951 24.5 (82/335) 25.3 (156/616) 0.95 (0.70, 1.30) 0.761 0.93 (0.67, 1.28) 0.637 1.01 (0.71, 1.44) 0.952 A, B, C, E, H, I, M,
quartile N
Social-emotional 958 8.9 (30/336) 10.9 (68/622) 0.80 (0.51, 1.26) 0.332 0.74 (0.47, 1.18) 0.208 0.64 (0.39, 1.06) 0.085 H, M, N, O
problems highest
decile
Social-emotional 958 20.8 (70/336) 27.8 (173/622) 0.68 (0.49, 0.93) 0.016 0.65 (0.47, 0.90) 0.010 0.61 (0.43, 0.88) 0.008 A, B, C, E, H, M, N,
problems highest O
quartile
Motor lowest decile 963 11.2 (38/338) 9.4 (59/625) 1.23 (0.80, 1.89) 0.355 1.24 (0.80, 1.89) 0.333 1.23 (0.75, 2.02) 0.404 G, I, N, O
Motor lowest 963 25.4 (86/338) 24.8 (155/625) 1.02 (0.75, 1.39) 0.879 1.02 (0.75, 1.39) 0.885 1.03 (0.73, 1.46) 0.853 B, G, I, N, O
quartile
LNS; lipid-based nutrient supplement. Non-LNS; Iron & folic acid + multiple micronutrient capsules (control group). 1We first tested the null hypothesis of no difference between the 3 treatment groups and combined the iron and folic
acid/multiple micronutrients groups because there were no significant differences between those 2 groups.
2All models were adjusted for child age at follow-up.
3Baseline covariates = A–F.
4Other covariates collected after baseline = G–O. A = maternal age; B = maternal education; C = maternal hemoglobin; D = maternal prepregnancy BMI; E = household asset score; F = primiparity; G = child sex; H = data
collector; I = exposure to multiple languages by 18 mo; J = type of preschool; K = teacher’s education; L = mean time in instructions at preschool; M = maternal depression; N = home stimulation score; O = maternal agency.
5Sample size based on model 1 adjusted for child age at follow-up.
6Results are based on ANCOVA (SAS PROC GLIMMIX).
Prenatal and postnatal LNS and child development 331
baseline: age, prepregnancy BMI, education, hemoglobin level, main trial, and the intervention groups in this sample were similar
and parity. Six out of 64 (9%) interactions between the effect to each other, suggesting a low risk of bias.
modifier and group were found to be significant at P< 0.1, which To our knowledge, this study is the first to examine the
is the proportion that would be expected due to chance. For any long-term developmental effects of LNS supplementation during
effect modifier, the maximum number of interactions found to be both pre- and postnatal periods. Two earlier RCTs examined
significant was 3 out of 8 outcomes; thus, none of the variables long-term developmental effects of early supplementary feeding
was a consistent effect modifier across outcomes. The interaction during both pregnancy and early childhood. In Guatemala, 2
between intervention group and HOME score was significant villages were assigned to Atole (a high-energy, high-protein
for the primary social-emotional development outcome that supplement) and 2 villages to Fresco (a low-energy, nonprotein
significantly differed between groups (P interaction = 0.081). supplement) (18). Both supplements were fortified with mi-
The effect of the LNS intervention on the behavioral problem z- cronutrients and were targeted at pregnant and lactating women
score was larger among children from households with HOME and children up to age 7 y. In Colombia (19), nutritionally
score below the median (β = 0.22 SD± 0.09; P= 0.019) than for at-risk families were randomly assigned to early (pregnancy
children from households with higher HOME scores (β = 0.12 through 6 mo postpartum) and/or late (6–36 mo postpartum)
SD ± 0.09; P = 0.204) (Figure 2). food supplementation with or without a social stimulation
intervention. Effects of supplementation on cognitive function
in both studies were observed immediately after the intervention
period, with cognitive gains being sustained through school age
Discussion and beyond. In Guatemala, higher supplement consumption,
In this randomized trial, children who received LNS from regardless of assigned group, was associated with higher affect
6 to 18 mo of age, and whose mothers received LNS during and social involvement, whereas low supplement consumption
pregnancy and through 6 mo postpartum, had significantly was associated with passive, despondent, and anxious behavior in
lower mean scores for behavioral problems, indicating better children at 6–8 y (40). Social-emotional and behavior problems
social-emotional behavior at 4–6 y of age than control children were not assessed in Colombia. In contrast, whereas we found
whose mothers received IFA or MMN. Intervention group no immediate positive effects of LNS at age 18 mo on cognitive
differences in behavioral problem scores were larger among or social-emotional function, we found differences in social-
children who received less nurturing and stimulation from the emotional and behavioral problems at preschool age. At least
home environment. Cognitive and motor scores at age 4–6 y, 2 other studies have found developmental effects of early
the other primary outcomes, did not differ significantly between supplementation on later outcomes, even without finding earlier
intervention groups, nor did any secondary outcomes, which effects on global developmental assessments (16, 17), consistent
included language, visuospatial ability, declarative memory, with the pattern of results in our cohort. The social-emotional
inhibitory control, preacademic skills, behavior rating scale, assessment we used at age 18 mo may not have been sensitive
prosocial skills, and delay of gratification scores. enough to detect effects at that early age, when children’s
The strengths of this study include the randomized study emotional and behavioral regulation is immature. Our finding of
design, which allowed us to examine long-term effects of nutrient lower behavior problems in the LNS group at 4–6 y is consistent
supplementation during most of the first 1000 d from conception with the association of higher supplement intake with positive
to age 24 mo. We used a wide range of carefully selected tests social-emotional and behavioral function at 6–8 y in Guatemala,
adapted to the Ghanaian context, which showed high reliability in although those findings were correlational and could have been
the local context. We implemented standardized test procedures confounded by unmeasured factors.
and rigorously trained a team of data collectors who achieved It is uncertain why we found no effects on cognition in the
high inter-rater agreement (>90%). Developmental outcomes present study. In our cohort, the prevalence of baseline maternal
were collected from 79% of children eligible for the follow- anemia (13%) and underweight (2.4%) during pregnancy was
up from the main trial, indicating a relatively low attrition rate. low (41). Thus it is plausible that the Ghanaian children were at
One limitation is that, during the main trial, participants could lower risk of malnutrition during fetal life than the study children
not be blinded to receipt of LNS compared with MMN or IFA in the other 2 trials, and therefore less likely to respond to a
because of the differences in their appearance, although data nutritional supplement (2). This is consistent with other studies
collectors who conducted the neurobehavioral assessments at of maternal supplementation during pregnancy and lactation (25,
follow-up and data analysts were kept blinded. Lack of blinding 42, 43) and supplementation of children in early infancy (44, 45)
of parents or caregivers could have biased their responses in which treatment effects were found only in subgroups at risk
for the neurobehavioral assessments based on the caregiver of undernutrition or poor social-economic status. However, the
report, i.e., the SDQ, which includes perceptions of children’s positive effects of the iLiNS-DYAD intervention on birth weight
behavioral problems. Another limitation is that we evaluated (15) and linear growth at 18 mo (46) show that this population did
multiple aspects of child development using multiple tests, which have potential to benefit from supplementation in some outcomes.
could result in false-positive significant results. However, we The potential to benefit in social-emotional outcomes may
prespecified only 3 primary outcomes and found a significant be related to the high prevalence of reported social-emotional
effect on 1 of these 3, reducing the likelihood that this was due difficulties in this sample: about 25% of children had a total
to chance. Also, we observed a differential loss to follow-up difficulties score in the abnormal range based on SDQ standard
between intervention groups, with a higher loss in the non-LNS cutoff scores (≥17) (30, 47), which is high compared with other
group. Nonetheless, most maternal baseline characteristics were studies: 3.6% in Denmark (48), 7.1% in Norway (49), and 9.9%
similar between this sample and the full sample enrolled into the in the United Kingdom (47). To our knowledge, our study is
332 Ocansey et al.
Low HOME High HOME
0.2
0.15
0.1
0.05
0
-0.05
-0.1
-0.15
-0.2
-0.25
LNS Non-LNS
FIGURE 2 Mean social-emotional difficulties score by intervention group and HOME score. Error bars show the standard error of the mean. High HOME,
HOME score above the sample median; HOME, Home Observation for the Measurement of the Environment Inventory; LNS, lipid-based nutrient supplement
for mothers and children; Low HOME, HOME score below the sample median; Non-LNS, iron and folic acid or multiple micronutrient capsules for mothers
only (control group). P for interaction between HOME score (categorized by the median population score as low HOME or high HOME score) and intervention
group = 0.081.
the first to document the prevalence of behavioral problems problems in later preschool and school age (58, 59). This suggests
among preschool children in Ghana, and to show that nutritional that the additional iron provided to the children in the LNS group
supplementation decreased the prevalence of parental report of from 6 to 18 mo is a plausible biological mechanism for the
such problems. observed effects at 4–6 y.
Two potential biological explanations for the lower behavioral A third potential explanation is that parents in the LNS
problem scores in the LNS arm than in the non-LNS arm could be group might have had greater expectations for their children’s
the essential fatty acids provided by LNS and the iron provided development, given that they were aware of the supplement
to children in the LNS group from 6 to 18 mo. In animal models, received, which could have led to biased reports of children’s
DHAdeficiency affects brain regions involved in the regulation of behavioral problems on the SDQ. However, we found no
emotional status, such as the prefrontal cortex, striatum (50), and differences between intervention groups in the prosocial skills
dopamine pathways, with accompanying deficits in behavioral subscale of the SDQ (also based on the caregiver report), which
and learning tasks (51). Among boys with attention deficit we would have expected if biased reporting was a factor. In
hyperactivity disorder, DHA deficiency has been associated with addition, in our cohort, parents’ perceptions of the impacts of
behavioral and learning problems (52). The LNS provided in the supplements on the index child were equally positive in the
our study contained 0.5 g of α-linolenic acid (18:3n−3) (ALA), LNS and non-LNS groups at follow-up (60), and both groups
the omega-3 essential fatty acid precursor to DHA, and 4.6 g had high expectations regarding the supplement’s impact on the
of linoleic acid (18:2n−6); the omega-6:omega-3 ratio was thus child’s future cognitive development and school performance.
within the recommended 4–10:1 omega-6:omega-3 fatty acid In conclusion, the provision of LNS during most of the
ratio (53). However, the conversion rate of ALA to DHA is first 1000-d window decreased behavioral problems reported by
typically low (9% in young women) (54) and we found no effect caregivers at preschool age, especially for children from low-
of maternal LNS supplementation on maternal plasma fatty acid stimulation households, but did not affect cognition or fine motor
status at 36 weeks of gestation, although breastmilk ALA at 6 function at preschool age in this Ghanaian cohort. Follow-up of
mo postpartum was higher in the LNS group than in the non- this cohort is needed to investigate whether the behavioral effects
LNS group among women in this study (55). Thus it is unclear persist and influence other functional outcomes through school
whether the essential fatty acids provided by LNS in early life age or young adulthood.
could account for the difference in child behavior problems at
4–6 y. Iron is needed for neurodevelopmental processes, such We thank the entire iLiNS-DYAD-Ghana team who conducted the follow-
as myelination, and for the synthesis of the neurotransmitter up study especially Edna Tetteh, MandyMcCarthy, Phoebe Ametepey, Vivian
dopamine, which is involved in social-emotional regulation (56). Abayavor, and Rhodaline Maulepe for neurobehavioral assessments. We
Iron deficiency may alter dopamine pathways, which may lead to thank Anna Lartey for leadership and conceptualization of the original trial;
socioemotional behavioral abnormalities including hyperactivity Charles D. Arnold for statistical advice; Mary Arimond for support with
overall project management; and Ebenezer Adjetey and Richard Azumah for
and inattentiveness (57). Iron deficiency in infancy and early data management.
childhood is associated with negative affective behavior, emotion The authors’ responsibilities were as follows—MEO, SA-A, KGD, and
regulation, temperament control, and social and attentional ELP: designed the research; MEO, SA-A, SMK, HO, SMT, BMO, and
SDQ Total difficules z-score 
 
Prenatal and postnatal LNS and child development 333
ELP: conducted the research; RRY, ELP, and MEO: performed the statistical supplementation on childhood cognitive outcomes. Am J Clin Nutr
analysis; KGD: advised on the analysis; MEO, ELP, and KGD: wrote the 2013;98(2):403–12.
manuscript; and SA-A, SMK, HO, BMO, RRY, and SMT: reviewed the draft 18. Pollitt E, Gorman K, Engle PL, Martorell R, Rivera J. Early
manuscript; and all authors read and approved the final manuscript. None of supplementary feeding and cognition—effects over 2 decades. Monogr
the authors report a conflict of interest related to research presented in this Soc Res Child Dev 1993;58(7):R5–R98.
19. Waber DP, Vuori-Christiansen L, Ortiz N, Clement JR, Christiansen
article. NE, Mora JO, Reed RB, Herrera MG. Nutritional supplementation,
maternal education, and cognitive development of infants at risk of
malnutrition. Am J Clin Nutr 1981;34(4):807–13.
20. Super C, Herrera M, Mora J. Cognitive outcomes of early nutritional
References intervention in the Bogota study. In: Meeting of the Society for
Research in Child Development, Seattle, WA; 1991.
1. Fox SE, Levitt P, Nelson CA, III. How the timing and quality of early 21. Brooks BL, Sherman EM, Strauss E. NEPSY-II: A developmental
experiences influence the development of brain architecture. Child Dev neuropsychological assessment. Child Neuropsychol 2009;16(1):80–
2010;81(1):28–40. 101.
2. Prado EL, Dewey KG. Nutrition and brain development in early life. 22. Brothers KB, Glascoe FP, Robertshaw NS. PEDS: Developmental
Nutr Rev 2014;72(4):267–84. milestones—an accurate brief tool for surveillance and screening. Clin
3. Bougma K, Aboud FE, Harding KB, Marquis GS. Iodine and mental Pediatr (Phila) 2008;47(3):271–79.
development of children 5 years old and under: A systematic review 23. Pisani L, Borisova I, Dowd AJ. International evelopment
and meta-analysis. Nutrients 2013;5(4):1384–416. and early learning assessment: technical working paper. Save
4. Brown KH, Peerson JM, Baker SK, Hess SY. Preventive zinc the Children, Washington (DC): 2015, Retrieved from
supplementation among infants, preschoolers, and older prepubertal https://www.savethechildren.org/content/dam/global/reports/education-
children. Food Nutr Bull 2009;30(1 Suppl):S12–40. and-child-protection/IDELA-tech-wrkppr-15.pdf.
5. Makrides M, Collins CT, Gibson RA. Impact of fatty acid status on 24. Noble KG, Norman MF, Farah MJ. Neurocognitive correlates of
growth and neurobehavioural development in humans. Maternal Child socioeconomic status in kindergarten children. Dev Sci 2005;8(1):74–
Nutr 2011;7(Suppl 2):80–88. 87.
6. Arimond M, Zeilani M, Jungjohann S, Brown KH, Ashorn P, 25. Prado EL, Alcock KJ, Muadz H, Ullman MT, Shankar AH. Maternal
Allen LH, Dewey KG. Considerations in developing lipid-based multiple micronutrient supplements and child cognition: A randomized
nutrient supplements for prevention of undernutrition: Experience from trial in Indonesia. Pediatrics 2012;130: peds. 2012-0412.
the International Lipid-Based Nutrient Supplements (iLiNS) Project. 26. Elliott CD, Smith P,McCulloch K. British Ability Scales second edition
Maternal Child Nutr 2015;11(S4):31–61. (BAS II): Administration and scoring manual. London: NFER-Nelson
7. Adu-Afarwuah S, Lartey A, Brown KH, Zlotkin S, Briend A, Dewey 1996.
KG. Randomized comparison of 3 types of micronutrient supplements 27. Wechsler D. TheWechsler Intelligence Scale for Children. SanAntonio,
for home fortification of complementary foods in Ghana: Effects on TX: The Psychological Corporation; 2002.
growth and motor development. Am J Clin Nutr 2007;86(2):412–20. 28. Baddeley A, Gardner JM, Grantham-McGregor S. Cross-cultural
8. Phuka JC, Gladstone M, Maleta K, Thakwalakwa C, Cheung YB, cognition: Developing tests for developing countries. Appl Cogn
Briend A, Manary MJ, Ashorn P. Developmental outcomes among 18 Psychol 1995;9(7):S173–95.
- month - old Malawians after a year of complementary feeding with 29. Bauer PJ, Zelazo PD. The National Institutes of Health Toolbox
lipid - based nutrient supplements or corn - soy flour. Maternal Child for the assessment of neurological and behavioral function: A
Nutr 2012;8(2):239–48. tool for developmental science. Child Dev Perspect 2014;8(3):
9. Iannotti LL, Dulience SJL, Green J, Joseph S, François J, Anténor M- 119–24.
L, Lesorogol C, Mounce J, Nickerson NM. Linear growth increased in 30. Goodman R. The Strengths and Difficulties Questionnaire: A research
young children in an urban slum of Haiti: A randomized controlled trial note. J Child Psychol Psychiatry 1997;38(5):581–86.
of a lipid-based nutrient supplement. Am J Clin Nutr 2014;99(1):198– 31. Goodman A, Goodman R. Strengths and difficulties questionnaire as a
208. dimensional measure of child mental health. J Am Acad Child Adolesc
10. Mangani C, Cheung YB, Maleta K, Phuka J, Thakwalakwa C, Dewey Psychiatry 2009;48(4):400–403.
K, Manary M, Puumalainen T, Ashorn P. Providing lipid - based 32. Obradović J, Yousafzai AK, Portilla XA, Rasheed MA, Memon U,
nutrient supplements does not affect developmental milestones among Siyal S, Tirado-Strayer N, Keehn B; School Readiness and Transition
Malawian children. Acta Paediatr 2014;103(1):e17–26. (STAR) Study: Executive functions battery adaptation manual. Early
11. Prado EL, Abbeddou S, Jimenez EY, Somé JW, Ouédraogo ZP, Vosti childhood cognitive stimulation and successful transition to preschool
SA, Dewey KG, Brown KH, Hess SY, Ouédraogo J-B. Lipid-based in a disadvantaged population in rural Pakistan. Aga Khan University,
nutrient supplements plusmalaria and diarrhea treatment increase infant Stanford University, & Harvard University, 2014.
development scores in a cluster-randomized trial in Burkina Faso. J Nutr 33. Smith-Donald R, Raver CC, Hayes T, Richardson B. Preliminary
2016;146(4):814–22. construct and concurrent validity of the Preschool Self-regulation
12. Prado EL, Maleta K, Ashorn P, Ashorn U, Vosti SA, Sadalaki J, Dewey Assessment (PSRA) for field-based research. Early Child Res Q
KG. Effects of maternal and child lipid-based nutrient supplements on 2007;22(2):173–87.
infant development: A randomized trial in Malawi. Am J Clin Nutr 34. Caldwell B, Bradley R. Home Observation for Measurement of the
2016;103(3):784–93. Environment: Administration manual. Tempe, AZ: Family & Human
13. Prado EL, Adu-Afarwuah S, Lartey A, Ocansey M, Ashorn P, Vosti Dynamics Research Institute, Arizona State University Google Scholar;
SA, Dewey KG. Effects of pre-and post-natal lipid-based nutrient 2003.
supplements on infant development in a randomized trial in Ghana. 35. Cox JL, Chapman G, Murray D, Jones P. Validation of the Edinburgh
Early Hum Dev 2016;99:43–51. Postnatal Depression Scale (EPDS) in non-postnatal women. J Affect
14. Matias SL, Mridha MK, Tofail F, Arnold CD, Khan MSA, Siddiqui Disord 1996;39(3):185–89.
Z, Ullah MB, Dewey KG. Home fortification during the first 1000 36. Schwarzer R, Jerusalem M. Generalized Self-Efficacy scale. In:
d improves child development in Bangladesh: A cluster-randomized Weinman J,Wright S, JohnstonM . eds.Measures in health psychology:
effectiveness trial. Am J Clin Nutr 2017;105(4):958–69. A user’s portfolio causal and control beliefs. Windsor, UK: NFER-
15. Adu-Afarwuah S, Lartey A, Okronipa H, Ashorn P, Zeilani M, Peerson Nelson; 1995:35–37.
JM, Arimond M, Vosti S, Dewey KG. Lipid-based nutrient supplement 37. Glascoe FP, Marks KP, Poon JK, Macias MM. Identifying and
increases the birth size of infants of primiparous women in Ghana. Am addressing developmental-behavioral problems. PEDStest.com, LLC;
J Clin Nutr 2015;101(4):835–46. 2016.
16. Fattal I, Friedmann N, Fattal-Valevski A. The crucial role of thiamine 38. Raghunathan TE, Lepkowski JM, Van Hoewyk J, Solenberger P. A
in the development of syntax and lexical retrieval: A study of infantile multivariate technique for multiply imputing missing values using a
thiamine deficiency. Brain 2011;134(6):1720–39. sequence of regression models. Surv Methodol 2001;27(1):85–96.
17. Colombo J, Carlson SE, Cheatham CL, Shaddy DJ, Kerling EH, 39. Adu-Afarwuah S, Lartey A, Okronipa H, Ashorn P, Ashorn U, Zeilani
Thodosoff JM, Gustafson KM, Brez C. Long-term effects of LCPUFA M, Arimond M, Vosti SA, Dewey KG. Maternal Supplementation
334 Ocansey et al.
with Small-Quantity Lipid-Based Nutrient Supplements Compared and identification of predictors from birth to preschool age in a Danish
with Multiple Micronutrients, but Not with Iron and Folic Acid, birth cohort: The Copenhagen Child Cohort 2000. Eur Child Adolesc
Reduces the Prevalence of Low Gestational Weight Gain in Semi- Psychiatry 2010;19(9):725–35.
Urban Ghana: A Randomized Controlled Trial. J Nutr 2017;147:697– 49. WichstrømL, Berg-Nielsen TS, AngoldA, Egger HL, SolheimE, Sveen
705 10.3945/jn.116.242909 28275100. TH. Prevalence of psychiatric disorders in preschoolers. J Child Psychol
40. Barrett DE, Radke-Yarrow M, Klein RE. Chronic malnutrition and Psychiatry 2012;53(6):695–705.
child behavior: effects of early caloric supplementation on social 50. Carrié I, Clément M, de Javel D, Francès H, Bourre JM. Phospholipid
and emotional functioning at school age. Dev Psychol 1982;18(4): supplementation reverses behavioral and biochemical alterations
541. induced by n-3 polyunsaturated fatty acid deficiency in mice. J Lipid
41. Adu-Afarwuah S, Lartey A, Okronipa H, Ashorn P, Zeilani M, Res 2000; 41:473–80 10706595.
Baldiviez LM, Oaks BM, Vosti S, Dewey KG. Impact of small- 51. Innis SM. Dietary omega 3 fatty acids and the developing brain. Brain
quantity lipid-based nutrient supplement on hemoglobin, iron status Res 2008;1237:35–43.
and biomarkers of inflammation in pregnant Ghanaian women. Matern 52. Stevens LJ, Zentall SS, Abate ML, Kuczek T, Burges JR. Omega-3
Child Nutr 2017;13(2):e12262. fatty acids in boys with behavior, learning, and health problems. Physiol
42. Joos SK, Pollitt E, Mueller WH, Albright DL. The Bacon Chow Behav 1996;59(4):915–20.
study: Maternal nutritional supplementation and infant behavioral 53. Dietary reference intakes for energy, carbohydrate, fiber, fat, fatty acids,
development. Child Dev 1983;54(3):669–76. cholesterol, protein, and amino acids. Washington (DC): Institute of
43. Prado EL, Sebayang SK, Apriatni M, Adawiyah SR, Hidayati N, Medicine, 2005.
Islamiyah A, Siddiq S, Harefa B, Lum J, Alcock KJ. Maternal 54. Burdge GC, Wootton SA. Conversion of α-linolenic acid to
multiple micronutrient supplementation and other biomedical and eicosapentaenoic, docosapentaenoic and docosahexaenoic acids
socioenvironmental influences on children’s cognition at age 9–12 years in young women. Br J Nutr 2002;88(4):411–20.
in Indonesia: Follow-up of the SUMMIT randomised trial. Lancet 55. Oaks BM, Young RR, Adu-Afarwuah S, Ashorn U, Jackson KH,
Global Health 2017;5(2):e217–28. Lartey A, Maleta K, Okronipa H, Sadalaki J, Baldiviez LM. Effects
44. GranthamMcGregor SM, Walker SP, Chang SM, Powell CA. Effects of a lipid-based nutrient supplement during pregnancy and lactation
of early childhood supplementation with and without stimulation on on maternal plasma fatty acid status and lipid profile: Results of two
later development in stunted Jamaican children. Am J Clin Nutr randomized controlled trials. Prostaglandins Leukot Essent Fatty Acids
1997;66(2):247–53. 2017;117:28–35.
45. Tofail F, Persson LA, El Arifeen S, Hamadani JD, Mehrin F, Ridout D, 56. Kim J, Wessling-Resnick M. Iron and mechanisms of emotional
Ekstrom EC, Huda SN, Grantham-McGregor SM. Effects of prenatal behavior. J Nutr Biochem 2014;25(11):1101–7.
food and micronutrient supplementation on infant development: A 57. Lozoff B. Early iron deficiency has brain and behavior effects consistent
randomized trial from the Maternal and Infant Nutrition Interventions, with dopaminergic dysfunction. J Nutr 2011;141(4):740s–46s.
Matlab (MINIMat) study. Am J Clin Nutr 2008;87(3):704–11. 58. Lozoff B, Jimenez E, Hagen J, Mollen E, Wolf AW. Poorer
46. Adu-Afarwuah S, Lartey A, Okronipa H, Ashorn P, Peerson JM, behavioral and developmental outcome more than 10 years after
Arimond M, Ashorn U, Zeilani M, Vosti S, Dewey KG. Small-quantity, treatment for iron deficiency in infancy. Pediatrics 2000;105(4):
lipid-based nutrient supplements provided to women during pregnancy e51.
and 6 mo postpartum and to their infants from 6 mo of age increase 59. Lozoff B, Beard J, Connor J, Felt B, Georgieff M, Schallert T. Long-
the mean attained length of 18-mo-old children in semi-urban Ghana: lasting neural and behavioral effects of iron deficiency in infancy. Nutr
A randomized controlled trial. Am J Clin Nutr 2016;104(3):797–808. Rev 2006;64(suppl 2):S34–43.
47. Goodman A, Rowe R, Gan Y. SDQ: Information for researchers and 60. Adams KP, Okronipa H, Adu-Afarwuah S, Arimond M, Kumordzie S,
professionals about the Strengths and Difficulties questionnaires 2010, Oaks BM, Ocansey ME, Young RR, Vosti SA, Dewey KG. Ghanaian
Retrieved from http://www.sdqinfo.org/. parents’ perceptions of pre and postnatal nutrient supplements and
48. Elberling H, Linneberg A, Olsen EM, Goodman R, Skovgaard AM. The their effects. Matern Child Nutr 2018;14:e12608 10.1111/mcn.12608
prevalence of SDQ-measured mental health problems at age 5–7 years 29656569