Prenatal Micronutrient Supplementation and Intellectual and Motor Function in Early School-aged Children in Nepal FREE

Micronutrient inadequacy is a critical concern among pregnant women and young children throughout the world. Gestation and the early postnatal period are considered sensitive periods for brain development, and nutritional deprivation during this period may lead to functional impairments.

Early iron deficiency can alter neuroanatomy, biochemistry, and metabolism, leading to changes in neurophysiologic processes that support cognitive and sensorimotor development.^1- 2 It also has an adverse effect on neurogenesis, altering brain morphology in specific regions such as the hippocampus and striatum.^3- 4 In addition, the timing of iron availability for myelin production, especially to oligodendrocytes, is critical⁵; hypomyelination due to iron deficiency persists later in life.^6- 7 Iron deficiency also alters neurochemistry, specifically monoamine transmission, reception, and metabolism.⁶ Finally, iron is important for energy metabolism in the brain.^2,8 Abnormal cerebral energy metabolism, as measured via loss of cytochrome c oxidase in the hippocampus and frontal cortex⁹ and via magnetic resonance spectroscopy studies,^3- 4 has been linked to iron deficiency. Energy-dependent processes such as dendritic arborization and synaptogenesis are thereby impaired²; MAP2 (microtubule-associated protein 2) expression required for dendritic scaffolding may be involved.¹⁰ Also in the hippocampus, down-regulation of several genes involved in synaptic plasticity occurs with perinatal iron deficiency.¹¹

Evidence from primates or humans that can demonstrate the effects of gestational iron deficiency on cognitive and motor functioning is limited. In rhesus monkeys, infants prenatally deprived of iron exhibited 20% reduced spontaneous activity level, lower inhibitory response to novel environments, and frequent behavioral changes,¹² whereas human infants with cord serum ferritin concentrations below 35 μg/L had lower auditory recognition memory.¹³ In a controlled trial in Australia, prenatal iron supplementation (20 mg/d) had no effect on IQ test performance in 4-year-old offspring,¹⁴ but iron deficiency anemia was prevalent in only 11% of the mothers.

The evidence for similar effects associated with zinc supplementation in pregnancy is even more inconclusive. A trial in Peru¹⁵ reported improvements in fetal neurobehavioral development associated with maternal supplementation, whereas in Bangladesh, infants of zinc-supplemented mothers had lower scores on the Bayley Scales of Infant Development than those whose mothers received a placebo during pregnancy; this may be due to an inhibitory effect of zinc on iron status.¹⁶ Zinc supplementation in African American women had no effect on cognitive or psychomotor functioning of their 5-year-old offspring.¹⁷

This study was designed to assess intellectual, executive, and motor functioning in a cohort of Nepalese children aged 7 to 9 years in 2007-2009 whose mothers received daily prenatal and postnatal micronutrient supplements in a controlled, cluster-randomized, double-blind trial conducted between 1999 and 2001.^18- 19 Additionally, in 2001-2005, these children were enrolled in a randomized, placebo-controlled trial of iron and/or zinc supplementation during preschool age.^20- 21 To isolate the effect of maternal supplementation alone, data from children in the placebo group of the preschool supplementation trial were included in the present analysis.

Ethical approval for the study was obtained from institutional review boards at the Johns Hopkins Bloomberg School of Public Health, the Pennsylvania State University, and the Institute of Medicine, Tribhuvan University.

From June 2007 to April 2009, we prospectively followed up 7- to 9-year-old offspring of women who had participated in a trial of prenatal supplementation with different combinations of micronutrients in a southeastern plains district (Sarlahi) of Nepal (Figure and eFigure). The study was conducted in 30 Village Development Committees (VDCs) of the district. Of a total of 48 VDCs, 30 VDCs were defined as the study area based on their geographic location. The VDCs located in the hills toward the north, where access was harder, and toward the border with India, where the population was expected to be transient, were not included. Village Development Committees in the far east or west of the district were also excluded. The entire 30-VDC study area was then divided into 426 sectors, or smaller community clusters, to create randomization units. Each sector had an area small enough to be covered by a local female worker who walked to all the households with participants. The parent 5-group, double-masked, cluster-randomized controlled trial had as its goal to examine effects on birth weight and infant mortality of daily supplementation with alternative combinations of micronutrients during pregnancy through 12 weeks postpartum.^18- 19 The mean gestational age at enrollment was 11 weeks (SD, 5.1 weeks) and adherence to supplements was high (88% [interquartile range, 67%-97%] of all possible doses); neither of these differed by supplementation group. The intervention groups (with micronutrient amounts per daily supplement) were folic acid (400 μg), folic acid plus iron (60 mg), folic acid/iron plus zinc (30 mg), and folic acid/iron/zinc plus vitamins D (10 μg), E (10 mg), B₁ (1.6 mg), B₂ (1.8 mg), B₆ (2.2 mg), B₁₂ (2.6 μg), C (100 mg), and K (65 μg); niacin (20 mg); copper (2.0 mg); and magnesium (100 mg), all with 1000-μg retinol equivalents of vitamin A. Vitamin A supplementation alone was the control. Children born to those in the folic acid–only group were not included in this follow-up study because of the low likelihood of folic acid alone having any effect on study outcomes. Also, to isolate effects of maternal supplementation, only children from the placebo group of a subsequent iron and/or zinc supplementation trial among preschoolers^20- 21 were included in this analysis. Per Nepalese government policy, children received 200 000 IU of vitamin A biannually from 6 to 60 months of age. To our knowledge, the study children had no other exposure to other forms of micronutrient supplementation.

Households with eligible children were invited to participate in the follow-up study. The purpose of the study was explained and parental oral consent and child assent were obtained.

Eligible households were visited to collect information about household demographics and socioeconomic status, a detailed history of the child's enrollment in school, type of school, and number of completed and repeated years of schooling.

Local terms were used to obtain histories of morbidity symptoms in study children during the previous 7 and 30 days. Assessment of diet at the time of follow-up was performed using a 7-day food frequency questionnaire. Household salt used for cooking was tested with a semiquantitative kit (MBI Kits, Madras, India). Iodine content was recorded as 0, less than 15 ppm, or 15 ppm or more.

Four master’s-level graduate students in psychology were the psychometrists who administered the Universal Nonverbal Intelligence Test (UNIT) and the Movement Assessment Battery for Children (MABC). They were trained by methods used in child clinical and school psychology PhD programs. When each performed a fully accurate test administration and scoring, they were certified to collect data. Test sessions were video-recorded and approximately 20% of the video records of each psychometrist's work were selected randomly and reviewed for test administration and scoring accuracy by Pennsylvania State University graduate students supervised by the Pennsylvania State University coinvestigators.

Each test was selected for its capacity to measure aspects of cognitive or motor functioning previously known to be sensitive to brain function changes attributed to nutritional influences. The UNIT²² is a nonverbal test of general intelligence using hand gestures to maximize fairness for all children irrespective of race, ethnicity, sex, language, country of origin, or hearing status.²³ The UNIT has 6 subtests assessing symbolic memory (0-30 possible raw score points), cube design (0-53), spatial memory (0-27), analogic reasoning (0-31), object memory (0-30), and mazes (0-93). The analogic reasoning subtest was culturally inappropriate because the stimuli include pictures of objects that are unfamiliar to most Nepalese children; it was therefore not administered.

Three tasks of executive function were administered that assessed inhibitory control and processing speed: a Stroop numbers test, backward digit span, and go/no-go. The Stroop test is commonly used to test inhibitory control,²⁴ and the backward digit span from the Wechsler memory scale²⁵ is used to test working memory and inhibitory control. A computerized go/no-go task was used to further assess inhibitory control. The task was a series of 210 “go” trials (press computer key when “go” stimulus is present) and 70 “no-go” trials (do not press if a “no-go” stimulus is present).²⁶

Motor development was assessed with the MABC,²⁷ which consists of a series of tasks to assess manual dexterity, ball skills, and balance. As an additional assessment of fine motor skills, a finger-tapping test was performed to assess index finger motor speed of each hand.

Children were brought to a central site for testing. On arrival, they were given a snack and a drink, and the psychometrist spent 30 minutes building rapport before taking the children to a testing room.

At the central site, trained anthropometrists measured height, weight, and mid upper arm circumference using standard procedures. Hemoglobin was estimated using fingerstick with a B-Hemoglobin Analyzer (HemoCue, Lake Forest, California). Children with anemia (hemoglobin <115 g/L) received iron/folic acid supplements as treatment.

Environmental influences on the selected child outcomes were indexed with the Middle Childhood Home Observation for the Measurement of the Environment (HOME) Inventory²⁸ and the Raven Coloured Progressive Matrices,²⁹ administered to mothers at the central site visit.

In this study, the sample size was fixed by the prior study designs. Hence, we calculated the smallest detectable differences in means of tests of cognition and motor functioning planned for the study between groups of interest that could be detected with a given power (80% and 90%) and type I error (α = .05). Testing was 2-sided and differences are estimated for varying levels of estimated correlations in outcomes within a cluster given the cluster-randomized design of the study as shown in eTable 1.

Weight-for-age, height-for-age, and body mass index z scores were calculated using the international reference standard.³⁰ Following basic exploratory analysis and asset score creation for socioeconomic status (asset score ranged from 0-11 and was made up of any household ownership of goats, cattle, cart, bicycle, mortorcycle, electricity, radio, television, telephone, mobile phone, or watches), baseline comparisons of the 4 treatment groups were performed with analysis of variance to assess imbalance on potentially confounding variables. Variables that were unbalanced between the treatment groups and associated with outcomes were controlled for in the adjusted analyses. Intellectual functioning outcomes were UNIT total scores, failure on the Stroop practice test, longest number of digits remembered backward correctly (backward digit span), and no-go percentage correct. Motor function was assessed with the MABC total and the mean number of finger taps of both hands.

Although UNIT is a standardized IQ test, there are no norms for Nepal, where it has never been used. Its factor structure for the Nepal sample was assessed using exploratory and confirmatory factor analyses of UNIT subscales. After omitting items with no variability, α was calculated for each subtest total raw score. The coefficients indicated satisfactory reliabilities, ranging from 0.76 for symbolic memory to 0.88 for mazes. Exploratory factor analyses suggested a 2- or 3-factor structure. Confirmatory analyses using EQS, version 6.0 (Multivariate Software Inc, Encino, California) compared the 2- and 3-factor solutions and found the 2-factor model to have a superior fit, with fit indexes of 0.995 (vs 0.927 for the 3-factor model). The subtests and their raw mean scores comprising the UNIT factors were as follows: for factor 1, symbolic memory (3.64 [SD, 2.73]), object memory (9.11 [SD, 4.56]), cube design (5.94 [SD, 5.13]), and spatial memory (4.36 [SD, 3.62]); and for factor 2, mazes subtest (22.16 [SD, 13.25]). Raw scores were converted to T scores (mean, 50 [SD, 10]) based on a child's age (7, 8, or 9 years) and were used as the outcome. The MABC is scored for failures so that higher scores reflect higher motor impairment. Raw scores on each task were converted to scaled scores provided in the manual, based on the standardization sample. Scaled score ranges on each of the subtests were as follows: manual dexterity, 0 to 15; ball skills, 0 to 10; and balance skills, 0 to 15. The total scaled score had a range of 0 to 40. Children whose score is less than the fifth percentile are considered to definitively have a motor problem. Those who score between the fifth and 15th percentiles are considered to be borderline in terms of motor problems. For the total scaled score, the 15th percentile cutoff is greater than 10. Generally, the above-mentioned cutoffs are used for clinical assessment.

Data were analyzed as for intention to treat and testing was 2-sided. We estimated the difference between test scores for each of the 3 active treatment groups compared with the placebo group using multivariate analysis of variance (MANOVA) across outcomes to account for correlated multiple outcomes. Treatment groups were coded as indicator variable. In addition, we applied a Bonferroni correction to the P values to account for the multiple comparisons being made between 3 treatment groups vs a single control. Only 643 records were used in the multivariate analysis as records with missing data for individual outcomes were excluded. As part of the multivariate analysis of covariance procedure, a bootstrap method was used to estimate 95% confidence intervals (CIs) and P values accounting for independent correlation of responses among children from the same sector due to the cluster-randomized design.^31- 32 Because age, sex, schooling, and asset score were strongly associated with UNIT and MABC scores, these variables were included in the adjusted models. In addition, because diet and morbidity were different by treatment group, we ran models adjusted for these variables as confounders along with other variables. We also examined age and sex for effect modification across the outcomes using MANOVA with bootstrapped 95% CIs. Data were analyzed using SAS software, version 9.2 (SAS Institute Inc, Cary, North Carolina) and STATA software, version 11 (Stata Corp, College Station, Texas).

In 2000-2001, 3351 live births were recorded in the 4 treatment groups of the maternal trial of micronutrient supplementation^18- 19 (Figure and eFigure). By March 2007, 129 had moved out of the study area and 289 had died, leaving a total of 2933 known to be alive and residing in the study area. In this analysis, we excluded all children (500-600 per group) who received active supplements as part of a preschool iron and zinc supplementation trial carried out by our group from 2001 to 2005^20- 21 (Figure). Of the remaining 708 children eligible for inclusion in this analysis, the number of participants by treatment group ranged between 108 and 228. Of this, a total of 676 consented to participate and were enrolled, corresponding to a loss to follow-up of approximately 5%.

The mean age of children at follow-up was 8.4 (SD, 0.65) years and differed by treatment group (P = .003) (Table 1). A majority (80%) had started school. Treatment groups differed with respect to dairy and meat intake in the past 7 days, percentage reporting symptoms of lower respiratory tract and gastrointestinal tract infection, and socioeconomic status. The percentages of children with stunting, wasting, or anemia did not differ by treatment group. Maternal age, schooling, literacy, and Raven score did not differ by treatment group.

In the control group, the mean UNIT T score was 48.2 (SD, 10.2); the proportion who failed the Stroop test was 0.45 (SD, 0.50); the mean longest number of digits correctly recalled in a backward digit span was 1.72 (SD, 0.96); the percentage correct no-go trials was 45.2 (SD, 21.0); the mean MABC standard score was 9.82 (SD, 6.99); and the mean number of finger taps was 35.3 (SD, 5.7) (Table 2). MANOVA revealed a significant difference (Bonferroni-adjusted P <.001) across tests for the iron/folic acid group compared with the control group. On individual tests, scores were better for intellectual function, tests of executive function (except for the go/no-go test), and motor function among children in the iron/folic acid group relative to the control group (Table 3). These differences were attenuated or absent in the iron/folic acid/zinc and multiple micronutrient groups across tests. Adjusting for age, sex, having ever been sent to school, asset score, diet, and morbidity did not change the results of the MANOVA test for the iron/folic acid group (P < .002) but P values were higher and no longer significant for some individual tests (Table 4). Adjusted mean differences in the iron/folic acid group relative to the control group were 2.38 (95% CI, 0.06-4.70; P = .04) for the UNIT T score, −0.14 (95% CI, −0.23 to −0.04; P = .005) for failure on the Stroop test; 0.36 (95% CI, 0.01-0.71; P = .02) for the backward digit span; −1.47 (95% CI, −3.06 to 0.12; P = .07) for the MABC (lower scores indicating better motor function); and 2.05 (95% CI, 0.87-3.24; P = .001) for the finger-tapping test.

Stratified analysis by age and sex and adjusted for confounders showed some differences, with older and female children benefiting more from the iron/folic acid intervention, but interaction effects were largely not statistically significant (eTable 2).

In a rural South Asian population, overall outcomes of general intellectual test performance and aspects of executive and motor function in 7- to 9-year-old children were better among those whose mothers had received prenatal iron and folic acid supplementation compared with controls. We previously reported high prevalence rates of iron deficiency and anemia during pregnancy in this rural area of Nepal.^33- 34 In general, the differences in test scores between the other intervention groups and controls were not statistically significant.

We found evidence linking prenatal iron/folic acid supplementation with working memory and inhibitory control test results. On the Stroop and backward digit span tests, children in the iron/folic acid group did the best and significantly better than control children, although children scored similarly across treatment groups on correct response to the no-go stimulus, which measures inhibitory control. The development of executive functioning is protracted, beginning in infancy and extending into early adulthood. Thus, future examination of this domain at an older age may reveal larger differences that may have been too early to detect in this analysis.

General intellectual functioning was higher with prenatal iron/folic acid supplementation. We observed an adjusted 2.4-point or one-quarter SD difference overall in the UNIT score, which, although of low clinical significance at the individual level, may be a meaningful difference at the population level. Examination of the UNIT score distributions for the control and iron/folic acid groups suggests a larger difference in the lower tail. The differences in intellectual and executive function tasks suggest that iron/folic acid, when provided during critical periods of development, may make an important difference in children's ability to learn tasks that are related to their academic achievement.

The higher scores of motor function, specifically related to fine motor control and speed, that are apparent in this study also show the possible importance of iron in utero. Both myelination and dopamine functioning may be related to the development of the motor system. Furthermore, motor-related areas of the brain also contain high concentrations of iron.³⁵ There is not much information concerning meaningful differences in MABC scores, but the mean adjusted non–statistically significant difference of 1.5 in the total score represents a one-quarter SD difference.

The combination of iron/folic acid/zinc was not different from controls. As noted, previous studies of prenatal zinc supplementation have failed to establish any benefit to mental or motor outcomes in children.^16- 17 Addition of zinc to iron and folic acid attenuated or negated the positive association with outcomes, which may be related to the inhibitory role of zinc in iron absorption described in the literature.^36- 37 We also noted a similar attenuation with added zinc of the positive effect of iron and folic acid on the original trial outcomes of birth weight¹⁸ and maternal anemia.³⁴ Outcomes in the multiple micronutrient supplement group were also not different than the control. Other studies have found small changes in cognitive outcomes with this intervention, mostly in infants or young children. In China, prenatal multiple micronutrient supplementation was associated with increases in mental development scores of 1.00 and 1.22 points compared with folic acid and folic acid/iron supplementation, respectively, although not in psychomotor scores at 1 year of age.³⁸ In a stratified analysis, infants of mothers with low body mass index who received multiple micronutrients had small but significant increments in motor scores and activity ratings than those whose mothers received iron/folic acid in Bangladesh.³⁹ Similarly, among children of women infected with human immunodeficiency virus, small albeit significant increases were observed with multivitamin supplementation relative to a placebo in psychomotor but not mental development indexes.⁴⁰

It is unlikely that outcome differences in the study were due to vitamin A present in the supplement. In a separate study in which we followed up children of women who received either vitamin A or placebo before, during, and after pregnancy, we found no differences between supplementation with vitamin A and placebo in either UNIT or MABC scores.⁴¹ However, a positive interaction between these 2 nutrients (iron and vitamin A) cannot be ruled out. We are unable to separate the folic acid–induced effects from those of iron. However, a systematic review that included 4 randomized clinical trials of the effects of dietary supplements of folic acid found no beneficial effect of 750 μg/d of folic acid on measures of cognition or mood in older healthy women,⁴² although these data are not directly relevant for school-aged children. Prenatal folic acid use in a US population was positively associated with gross motor development but negatively associated with personal-social outcomes.⁴³ Similarly, maternal folic acid use was associated with fewer child behavioral problems.⁴⁴ Data from these studies are unconvincing regarding the role of folic acid; however, a synergistic interaction between iron and folic acid may exist.

Our study has several strengths and some limitations. Because of the original randomized study design and a high rate of follow-up, we can report some linkage between the role of iron supplementation in utero and cognitive and motor outcomes. However by excluding the folic acid–alone group and including only the placebo group of the child supplementation trial, we deviated from the original designs of the trials and, therefore, have refrained from making any causal inferences. We used tests of general intelligence and motor function that are well known and standardized for use in different cultures and settings. Yet, factors such as a child's sex and ever having been sent to school were strong predictors of the outcomes, especially for the UNIT score. It is possible that in this setting, having been to school better prepares a child for testing. Separating out the effect of better achievement on test scores due to schooling is important, but there were some differences by treatment group in the proportion of children sent to school, and the adjustment for this variable did alter the study results somewhat, suggesting that schooling may have been in the causal pathway; this warrants further study. Sex difference in test scores may be related to the emphasis on proficiency in household chores and taking care of younger siblings among girls, not allowing them sufficient time for schoolwork and study. Schoolwork proficiency may itself be valued differently by sex.

In conclusion, our study found evidence that maternal prenatal supplementation with iron and folic acid was positively associated with general intellectual ability, some aspects of executive function, and motor function, including fine motor control, in offspring at 7 to 9 years of age in a rural area of Nepal where iron deficiency is highly prevalent. Antenatal iron/folic acid use per international guidelines should be expanded in many low- and middle-income settings where program coverage continues to be poor. Further follow-up studies are required to examine whether the observed benefits in early school age persist into adolescence and adulthood.