Evaluation of multiple micronutrient supplementation and medium-quantity lipidbased nutrient supplementation in pregnancy on child development in rural Niger: A secondary analysis of a cluster randomized controlled trial

Background It is estimated that over 250 million children under 5 years of age in low- and middle-income countries (LMICs) do not reach their full developmental potential. Poor maternal diet, anemia, and micronutrient deficiencies during pregnancy are associated with suboptimal neurodevelopmental outcomes in children. However, the effect of prenatal macronutrient and micronutrient supplementation on child development in LMIC settings remains unclear due to limited evidence from randomized trials. Methods and findings We conducted a 3-arm cluster-randomized trial (n = 53 clusters) that evaluated the efficacy of (1) prenatal multiple micronutrient supplementation (MMS; n = 18 clusters) and (2) lipidbasedAU : Pleasenotethat}lipid – basedmicronutrientsupplementation}hasbeenchangedto}lipid -basednutrientsupplementation}nutrient supplementation (LNS; n = 18 clusters) as compared to (3) routine iron-folic acid (IFA) supplementation (n = 17 clusters) among pregnant women in the rural district of Madarounfa, Niger, from March 2015 to August 2019 (ClinicalTrials.gov identifier NCT02145000). Children were followed until 2 years of age, and the Bayley Scales of Infant and Toddler Development III (BSID-III) were administered to children every 3 months from 6 to 24 months of age. Maternal report of WHO gross motor milestone achievement was assessed monthly from 3 to 24 months of age. An intention-to-treat analysis was followed. Child BSID-III data were available for 559, 492, and 581 singleton children in the MMS, LNS, and IFA groups, respectively. Child WHO motor milestone data were available for 691, 781, and 753 singleton children in the MMS, LNS, and IFA groups, respectively. Prenatal MMS had no effect on child BSID-III cognitive (standardized mean difference [SMD]: 0.21; 95% CI: -0.20, 0.62; p = 0.32), language (SMD: 0.16; 95% CI: -0.30, 0.61; p = 0.50) or motor scores (SMD: 0.18; 95% CI: -0.39, 0.74; p = 0.54) or on time to achievement of the WHO gross motor milestones as compared to IFA. Prenatal LNS had no effect on child BSID-III cognitive (SMD: 0.17; 95% CI: -0.15, 0.49; p = 0.29), language (SMD: 0.11; 95% CI: -0.22, 0.44; p = 0.53) or motor scores (SMD: -0.04; 95% CI: -0.46, 0.37; p = 0.85) at the 24-month endline visit as compared to IFA. However, the trajectory of BSID-III cognitive scores during the first 2 years of life differed between the groups with children in the LNS group having higher cognitive scores at 18 and 21 months (approximately 0.35 SD) as compared to the IFA group (p-value for difference in trajectory <0.001). Children whose mothers received LNS also had earlier achievement of sitting alone (hazard ratio [HR]: 1.57; 95% CI: 1.10 to 2.24; p = 0.01) and walking alone (1.52; 95% CI: 1.14 to 2.03; p = 0.004) as compared to IFA, but there was no effect on time to achievement of other motor milestones. A limitation of our study is that we assessed child development up to 2 years of age, and, therefore, we may have not captured effects that are easier to detect or emerge at older ages. Conclusions There was no benefit of prenatal MMS on child development outcomes up to 2 years of age as compared to IFA. There was evidence of an apparent positive effect of prenatal LNS on cognitive development trajectory and time to achievement of selected gross motor milestones. A randomized, double-blind, placebo-controlled trial to assess the safety and efficacy of Rotasiil (Serum Institute of India), a live, oral rotavirus vaccine for infants, was conducted in Madarounda, Niger (clinicaltrials.gov identifier: {"type":"clinical-trial","attrs":{"text":"NCT02145000","term_id":"NCT02145000"}}NCT02145000) [22,23]. The rotavirus vaccine trial protocol and the primary results and adverse events have been published elsewhere [22]. Given the low immunogenicity of oral vaccines in high mortality settings, a nested cluster-randomized controlled trial of nutritional interventions in pregnancy (referred hereafter as the nutrition-immunogenicity trial) was conducted to evaluate whether prenatal MMS and LNS can increase the immunogenicity of rotavirus vaccine as compared to prenatal IFA, which is standard of care in Niger and most LMIC settings [24]. The nutrition-immunogenicity trial protocol and the primary results and adverse events have been published elsewhere [23]. The nutrition-immunogenicity trial was conducted concurrently with the rotavirus vaccine trial; pregnant women were first enrolled in the nutrition-immunogenicity trial and then infants of mothers who were enrolled in the nutrition-immunogenicity trial were screened at 6 to 8 weeks of age for enrollment in the rotavirus vaccine trial. We have previously reported that there was no difference in the primary outcome of infant immune response to rotavirus vaccine between prenatal LNS, MMS, and IFA supplementation [24]. In this study, we present an analysis of the nutrition-immunogenicity trial to assess the effect of prenatal MMS and LNS as compared to prenatal IFA on the secondary outcome of child development up to 2 years of age. Pregnant women were enrolled in the nutrition-immunogenicity trial from March 2015 and March 2016, and the last child 2-year follow-up visit was conducted in August 2019. There were no important changes to the nutrition-immunogenicity trial after enrollment commenced. This study is reported as per the “CONSORT extension for Cluster Trials” guideline (see S1 CONSORT Checklist). Briefly, the nutrition-immunogenicity trial randomized 53 village clusters in a 1:1:1 allocation scheme to 1 of 3 prenatal supplementation groups: IFA (routine care), MMS, or LNS [24]. Randomization was performed by having the head of each village cluster select a piece of paper indicating the randomization group from an opaque jar. Villages were stratified by population size: <100; 100 to 249; ≥250 nonpregnant women of reproductive age. A total of 17 villages were randomized to IFA, 18 villages to MMS, and 18 villages to LNS. Women of reproductive age in participating villages provided written informed consent for community-based monthly pregnancy surveillance, which included an at-home urine pregnancy test. Women with a confirmed pregnancy were then screened for trial enrollment at the health facility. The trial inclusion criteria were (i) <30 weeks gestation at the time of enrollment based on maternal report of the last menstrual period; (ii) intended to remain in the study area until 2 years postpartum; and (iii) did not have a chronic health condition, severe illness at screening, pregnancy complications (moderate to severe edema, hemoglobin (Hb) <7 g/dL, or diastolic blood pressure >90 mm Hg); and (iv) no self-reported peanut allergy [24]. Pregnant women who meet all inclusion criteria and provided written informed consent were enrolled in the nutrition-immunogenicity trial. A total of 3,332 pregnant women were enrolled in the nutrition-immunogenicity trial, of which 1,105 were in the IFA group, 1,083 in the MMS group, and 1,144 in the LNS group. At 6 to 8 weeks after birth, infants born to women enrolled in the nutrition-immunogenicity trial were screened for enrollment in the randomized, double-blind, placebo-controlled vaccine trial of a live, oral rotavirus vaccine [22]. The inclusion criteria for infants in the rotavirus vaccine trial were (i) 6 to 8 weeks of age; (ii) able to swallow and have no history of vomiting within the past 24 hours; (iii) intended to remain in the study area for 2 years; and (iv) parent/guardian provided written informed consent. The analytic population for the current study consists of 2,551 children whose mothers completed prenatal supplementation in the nutrition-immunogenicity trial and were also enrolled in the rotavirus vaccine trial. There were 860 children in the IFA group, 777 children in the MMS group, and 874 children in the LNS group. Pregnant women received nutritional supplements based on their village cluster from the time of randomization until delivery. The composition of the IFA, MMS, and LNS is detailed in the Table A in S1 Appendix. Pregnant women in the IFA standard of care group received tablets containing 60 mg iron and 400 μg folic acid (Remedica; Limassol, Cyprus). Pregnant women in the MMS group were provided capsules containing 30 mg iron, 400 μg folic acid, and 20 other micronutrients (DSM Nutritional Products; Isando, South Africa). Pregnant women in the LNS group received a 40-g fortified, ready-to-use food made of peanuts, oil, dried skimmed milk powder, and sugar (Nutriset S.A.S; Malaunay, France), which contained the same micronutrient content as the MMS. Based on nutritional composition, the LNS would be classified as a medium-quantity LNS [25]. Due to the inability to manufacture LNS completely indistinguishable from IFA tablets and MMS capsules, it was not possible to blind participants or field staff to their randomized group. The statistical analysis was conducted blinded to the randomized group using coded labels. During pregnancy, home visits were conducted by research assistants every 7 days until delivery to distribute nutritional supplements. At each visit, the research assistants also obtained a count of the number of consumed nutritional supplements since the last visit. Adherence percentage for each participant was calculated as the total number of nutritional supplements consumed from enrollment to delivery as assessed at home visits divided by the expected total number of nutritional supplements the woman should have consumed from enrollment to delivery. At the time of enrollment in the nutrition-immunogenicity trial, study midwives administered a standardized questionnaire to pregnant women, which assessed maternal and household sociodemographic characteristics and conducted a physical exam, and assessed maternal anthropometry (height, weight, and mid-upper arm circumference (MUAC)). Maternal Hb concentration was assessed from a finger-prick blood sample (Hemocue Hb 301, Angelholm, Sweden), and pregnant women received a malaria rapid diagnostic test (Biolin Malaria Ag Pf (HRP-2), Abbott Diagnostics, Scarborough, USA). Food security was assessed using the household hunger scale [26]. Improved sanitation was defined as a household having access to a flush toilet, improved pit latrine, or slab latrine. Improved water source was defined as households using covered or protected ground well for drinking water. Infants were screened for enrollment in the vaccine trial at 6 to 8 weeks of age. Community health assistants conducted monthly home visits from 3 to 24 months of age to assess child morbidity and growth. A culturally adapted BSID-III with the cognitive, language, and motor scales was administered at the health facility at 6 months of age and every 3 months thereafter until 24 months of age [27]. To ensure cultural appropriateness, BSID-III items were adapted based on the consensus of a panel of local health staff and a child development expert. Adaptations included the replacement of unfamiliar images or terminology with more culturally relevant stimuli (e.g., changing a picture of an apple to a mango or using a simple wooden doll as are commonly found in villages of the study area). To maintain functional equivalence, replacement stimuli were selected to be of similar size, style, and complexity to the original stimuli. Eight female research nurses were selected and trained as dedicated BSID-III data collectors. Recruitment for the BSID-III assessment cohort was stopped on September 17, 2017, primarily due to resource constraints; the cohort of infants who turned 6 months of age after this date did not receive BSID-III assessments. Field-based supervision and weekly staff meetings were used to prevent assessor drift and ensure the continued quality of implementation. The BSID-III was administered in quiet and dedicated evaluation rooms at each health facility. The BSID-III showed high internal consistency for all domains (Cronbach’s alphas ≥0.85) (Table B in S1 Appendix). In addition, data collectors assessed maternal reports of achievement of the 6 WHO gross motor milestones every 4 weeks from 4 to 24 months of age: sitting without support, standing with assistance, hands-and-knees crawling, walking with assistance, standing alone, walking alone [28]. There were no changes to the developmental outcomes after the trial commenced. The nutrition-immunogenicity trial sample size was based on the primary infant immunogenicity endpoint of seroconversion at 28 days post-oral rotavirus vaccine dose 3 assuming 90% power and a 20% absolute difference in the proportion of children that seroconvert between the randomized vaccine and placebo groups [23]. The intention-to-treat (ITT) principle was used for all analyses and all analyses accounted for clustering by the village. Generalized linear regression models with cluster-robust standard errors were used to assess standardized mean differences (z-score differences) in BSID-III cognitive, language, and motor scores at 24 months of age (endline visit) using study-specific z-scores that were calculated from internal means and standard deviations among the full child population. We also assessed differences in BSID-III composite scores (US norms) for the 24-month visit by randomized groups [29,30]. We analyzed differences in the longitudinal trajectory of BSID-III cognitive, language, and motor raw scores from 6 to 24 months of age between randomized treatment arms with generalized linear mixed models (GLMMs). Multilevel models were constructed with the GLIMMIX procedure in SAS version 9.3 to take into account correlation within village clusters and correlation within children who were nested within the village clusters with random intercepts. GLMMs included randomized treatment arm, child age at assessment (6-, 9-, 12-, 15-, 18-, 21-, and 24-month time bins), an interaction term between treatment arm and infant age, and used a compound symmetry structure for within-subject correlation. If the overall test for difference in the trajectory of development domain scores between randomized groups was statistically significant, then differences in mean scores between groups at each child age were assessed using least-square means with Tukey–Kramer adjustment for multiple comparisons. In addition, Cox proportional hazard models with cluster-robust standard errors were used to assess the time to acquisition of each of the 6 WHO gross motor milestones; hazard ratios (HRs) >1.0 indicated the earlier achievement of the milestones (beneficial effect). Interaction terms between randomized group and time were used to assess the proportional hazards assumption. As a sensitivity analysis to address the potential for baseline imbalance between randomized groups, we constructed multivariate models for all outcomes that included baseline covariates that may be associated with development outcomes based on a literature review including household wealth quintile, household size, food security, maternal age, maternal education, maternal anemia (Hb <11 g/dL), maternal underweight (body mass index (BMI) <18.5 kg/m2), malaria infection, the season of enrollment and child age and sex. Missingness for baseline variables was <5%, and missing indicators were used to retain participants in multivariable models. Further, we assessed potential bias due to dependent censoring (missing outcome data) on BSID-III scores at 24 months using inverse probability of censoring weights (IPCW) [31]. Stabilized censoring weights were constructed in models that included household wealth quintile, household size, food security, maternal age, maternal education, maternal anemia, maternal underweight, maternal malaria, and season of enrollment in the trial. Missing indicators were used to retain all participants in the calculation of censoring weights. In addition, we conducted exploratory analyses that assessed potential differences in the magnitude of the effect (effect modification) of MMS and LNS on BSID-III domain scores at 24 months of age as compared to IFA by baseline maternal education, maternal anemia, maternal underweight, and child sex. The effect modifiers were selected based on evidence that the effect of MMS on birth outcomes and child mortality differs by maternal nutritional status and child sex [13]. The Wald test was used to assess the statistical significance of interaction. Analyses were conducted in Stata Version 16 and SAS version 9.3. The trial received ethical approval from the Comité Consultatif National d’Ethique in Niger, the Comité de Protection des Personnes in France, the Commission d’Ethique de la Recherche sur l’Etre Humain, Hôpitaux Universitaires de Genève in Switzerland, the Research Ethics Review Committee of the World Health Organization in Switzerland, and the Western Institutional Review Board in Olympia, WA. The trial was overseen by an independent Data and Safety Monitoring Board.