Does early Vitamin B12 supplementation improve neurodevelopment and cognitive function in childhood and into school age: A study protocol for extended follow-ups from randomised controlled trials in India and Tanzania

Introduction: As many as 250 million children under the age of 5 may not be reaching their full developmental potential partly due to poor nutrition during pregnancy and the first 2 years of life. Micronutrients, including vitamin B12, are important for the development of brain structure and function; however, the timing, duration and severity of deficiencies may alter the impact on functional development outcomes. Consequently, to fully explore the effect of vitamin B12 on cognitive function, it is crucial to measure neurodevelopment at different ages, in different populations and with vitamin B12 supplementation at different times during the critical periods of neurodevelopment. Methods and analysis: In this project, we follow up children from four recently completed randomised placebo-controlled trials of oral vitamin B12 supplementation, two in India and two in Tanzania, to explore the long-term effects on neurodevelopmental outcomes and growth. All the included trials provided at least two recommended dietary allowances of oral vitamin B12 daily for at least 6 months. Vitamin B12 was supplemented either during pregnancy, early infancy or early childhood. Primary outcomes are neurodevelopmental status, cognitive function and growth later in childhood. We apply validated and culturally appropriate instruments to identify relevant developmental outcomes. All statistical analyses will be done according to intention-to-treat principles. The project provides an excellent opportunity to examine the effect of vitamin B12 supplementation in different periods during early life and measure the outcomes later in childhood. Ethics and dissemination: The study has received ethical approvals from all relevant authorities in Norway, USA, Tanzania and India and complies fully with ethical principles for medical research. Results: will be presented at national and international research and policy meetings and published in peer-reviewed scientific journals, preferably open access. This study is a follow-up of four recently completed double-blind randomised placebo-controlled trials conducted in lower socioeconomic, semiurban and urban populations in Delhi (India),21 30 Bangalore (India)31 and Dar es Salaam (Tanzania)32 33 in 2001– 2011. All trials provided at least two recommended dietary allowances (RDA) of vitamin B12 daily (2×2.6 µg for adults and 9 µg for small children) for at least 6 months, with no vitamin B12 supplementation in the placebo groups. The two studies that enrolled pregnant women provided the highest dose of vitamin B12 (50 µg daily, which is approximately 20 times the RDA). An overview of the original trials and the follow-ups is presented in table 1. Overview of study populations, study period and exposure in the original trials and in the Vitabeginning follow-up study *From <14 weeks gestational age to 6 weeks postpartum vitamin B12 50 µg daily. In addition, all women were given iron and folic acid supplementation throughout pregnancy. †From 6 to 30 months for 6 months; four groups: vitamin B12, vitamin B12 +folic acid, folic acid and placebo, doses >12 months: vitamin B12 1.8 µg and folic acid 150 µg (half doses for <12 months). ‡From 12 to 27 weeks gestational age until delivery: multivitamin B12 50 µg, vitamin B1 20 mg, vitamin B2 20 mg, vitamin B6 25 mg, niacin 100 mg, vitamin C 500 mg, vitamin E 30 mg, folic acid 0.8 mg. §From 6 weeks to 18 months, four groups; multivitamin + zinc, vitamin Z and placebo. Multivitamin <6m: vitamin B12 1 mg, vitamin B1 0.5 mg, vitamin B2 0.6 mg, vitamin B 60.6mg, niacin 4 mg, folic acid 130 µg, vitamin C 60 mg, vitamin E 8 mg. Zinc 5 mg >6 months multivitamins and zinc doses were double. In Bangalore, South India, HIV-negative pregnant women recruited before or at 14 weeks gestational age were randomised to receive a daily dose of oral vitamin B12 (50 µg) or a placebo through 6 weeks postpartum. The primary objective was to determine the effect of vitamin B12 supplementation in improving maternal vitamin B12 status. Enrolment was completed in September 2010, and the last enrolled infant was born in August 2011. Oral vitamin B12 supplementation of women throughout pregnancy and early lactation, in combination with standard prenatal care with routine supplementation of iron and folate, significantly increased the vitamin B12 status of women and their offspring.31 Neurodevelopment was measured at 9 and 30 months. In this follow-up study, we will measure the effect of maternal vitamin B12 supplementation on neurodevelopment and cognitive function several times 5–6.5 years after supplementation and on neurophysiological outcomes using ERP. In Delhi, North India, children 6–30 months of age were randomised to receive daily (1) vitamin B12, (2) vitamin B12 and folate, (3) folate or (4) placebo for 6 months. The supplementation included vitamin B12 (1.8 µg) and/or folate (150 µg) and with half doses for children <12 months. The primary objective was to measure the effect of these interventions on the incidence of diarrhoea and pneumonia. Neurodevelopment was a predefined secondary outcome and was measured in a subsample of 422 children. In total, 1000 children were enrolled between January 2010 and September 2011. Vitamin B12 and folic acid-supplemented children scored significantly higher on neurodevelopment scores at the age of 12–36 months, compared with those who received placebo.21 In this follow-up study, we will measure to what extent early supplementation of folic acid and/or vitamin B12 improves neurodevelopment and cognitive function 5–6 years after supplementation. The study is powered to measure the effect of vitamin B12, folic acid and the two vitamins combined on neurodevelopmental outcomes. The follow-up study will also measure to what extent vitamin D status in early life is associated with neurodevelopmental scores in early school years. In Dar es Salaam, Tanzania, infants 6–10 weeks of age born to HIV-negative mothers were randomised to receive daily (1) zinc, (2) multivitamins, (3) zinc + multivitamin or (4) placebo. Multivitamins included vitamin B121 mg, B10.5 mg, B20.6 mg, B60.6 mg, niacin 4 mg, folic acid 130 µg, vitamin C 60 mg and vitamin E 8 mg and was provided alongside zinc 5 mg. Doses were doubled after 6 months. Children were followed for 18 months (ie, until age 19.5 months). The primary objective was to measure the incidence of diarrhoea and respiratory tract infections. Enrolment was completed in December 2009, and follow-up ended in May 2011. Neurodevelopmental outcomes were assessed in a subset of children at 15 months.32 Daily zinc supplementation lowered the burden of diarrhoea and respiratory tract infections. No added benefit was seen from the provision of multivitamins.33 In another trial in Dar es Salaam, Tanzania on micronutrients and adverse pregnancy outcomes, HIV-negative pregnant women at 12–27 weeks gestational age were randomised to receive daily oral supplementation of multivitamins including vitamin B12 or placebo. Multivitamins included vitamin B1250 µg, vitamin B120 mg, vitamin B220 mg, vitamin B625 mg, niacin 100 mg, vitamin C 500 mg, vitamin E 30 mg and folic acid 0.8 mg. All women received prenatal iron and folate supplementation. The primary objective was to measure the effect of vitamin supplementation on fetal loss, low birth weight and severe preterm birth. Enrolment was completed in July 2004, and the last enrolled infant was born in February 2005. Multivitamin supplementation reduced the incidence of low birth weight and small for gestational age births but had no significant effects on prematurity or fetal death.34 The two studies from Tanzania are well suited for follow-up studies on potential impact of micronutrient supplementation on child health and neurodevelopmental outcomes in older children. Neurodevelopment, cognitive function and linear growth will be key outcomes in the different studies. Neurodevelopmental status in young children has been assessed by the comprehensive assessment tool Bayley Scales of Infant and Toddler Development third edition35 and the easily administered screening tool Ages and Stages Questionnaire third edition in the original studies.21 36 In the follow-up, each site use their own unique collections of tests and questionnaires (table 2). General intellectual functioning is assessed by a modified Kaufmann ABC II (Bangalore and Dar es Salaam), the Wechsler Preschool and Primary Scale of Intelligence–III (Bangalore) or the Wechsler Intelligence Scale for Children VI, Indian version (Delhi). These are complemented by tests on specific cognitive functions by subtests from the developmental neuropsychological test battery A Developmental NEuroPSYchological Assessment, version 2 (NEPSY-II) including attention and executive functioning, language, social perception, sensorimotor and visuospatial processing (Delhi) and adaptive functioning by the Vineland Social Maturity Scale (Bangalore). Mental health and behaviour problems are measured by the parent-reported screening instrument Strengths and Difficulties Questionnaire and executive functions by the parent-reported questionnaire Behaviour Rating Inventory of Executive Function (all sites). Finally, in one study (Bangalore), we measure neurophysiological functions using event-related potentials as this may yield additional information on the effects of nutritional deficiencies on brain function. In the present study, we propose to use two well-characterised ERPs, P-300 and Mis-match Negativity that are known to reflect higher cognitive functions of attention and memory (see table 2 for details on the assessments at each site). Overview over inventories and data collection tools used in the different studies and the age of assessment *In Bangalore: Atlantis, number recall, word order, pattern reasoning and triangles from KABC and Koh’s Block Design Test and Verbal Fluency in addition; in Dar es Salaam: the EACABT with permission from Savings Brain Multisite Study (WHO/BMGF) including Atlantis, hand movements, footsteps, story completion, Kilifi Naming Test, ROCF, NOGO, shift, people search, literacy and numeracy, HOME, Brief-P (modified from MAL-ED), SDQ, BQP, in addition to verbal fluency and Koh’s Block. †ENOBIO 32 (Device name)—brain monitoring and stimulation technologies, mismatch negativity, P300 (Neuroelectrics, Boston, MA). ‡Different versions used in different sites: in Bangalore: Brief Parent; in Delhi: Brief Second Edition; in Dares es Salaam: Saving Brains/Gates/WHO modified version from the malaria study. §On stored serum samples from Child II cohort (subject to funding). BMGF, Bill & Melinda Gates Foundation; BQP, Behavior Questionnaire for Parents; CBC, complete blood count; CRP, C reactive protein; EACBT, African Cognitive Assessment Battery; HcY, homocysteine; HOME, Home Observation Measurement of the Environment; KABC, Kaufman Assessment Battery for Children; MAL-ED, Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development, MMA, methylmalonic acid; NEPSY-II, A Developmental NEuroPSYchological Assessment; NOGO, go/no go test for sustained attention and response control; RBC folate, red blood cell folate analysis; ROCF, Rey–Osterrieth complex figure; SDQ, Strengths and Difficulties Questionnaire; VSMS, Vineland Social Maturity Scale; WISC-IVINDIA, Wechsler Intelligence Scale for Children VI, Indian version. Weight and height are measured using standard methods. In Delhi, height, weight and skin fold thickness were measured in children, using Seca scales (hight/weight) and Holtain Calipers. In Bangalore, weights of mothers and children were recorded using a digital balance (Salter’s 9016; Tonbridge) to the nearest 100 g, and the heights were measured using a stadiometer to the nearest 0.1 cm. The data for our most relevant tests are expected to be normally distributed. We have decided that a standardised effect size of 0.35 SD is the minimally clinically relevant effect for the main psychometric tests. If the smallest group in any of our comparisons includes at least 130 children in each arm, and assuming a two-sided alpha error of 0.05 we will have 81%, 89% and 98% power to detect differences for effect sizes of 0.35, 0.4 and 0.6, respectively. As a result, all of the described studies have sufficient power to detect important differences between vitamin B12 and placebo recipients between individuals with poor and adequate vitamin B12 status or between other groups of other exposures as long as each of the groups contain at least 130 children. Statistical powers according to standardised mean effect sizes of .35, 0.4 and 0.6 and total sample sizes are depicted in figure 1. This graph was generated using the ‘power two means’ command in STATA V.13; it assumes equal group sizes and equal variances and a significance level of 0.05. Estimated required total sample sizes based on relevant effect sizes. Data collection for the primary and secondary outcomes are synchronised and selected to capture the same domains of development using different tools. Assessment tools for cognitive functions such as general abilities, achievement, verbal, visual spatial and sensorimotor skills, memory, executive functioning, general behaviour and social perception and maturity are dependent on the child’s age. We have carefully selected well-validated and developmentally sensitive instruments to ensure detection of the relevant predictors according to age and research questions. Instruments are adapted to ensure psychometric qualities, as well as cultural and linguistic appropriateness of the test at each site. Clinically useful tests will be prioritised to improve sustainability of test material and knowledge of developmental assessment at the specific sites. Assessments are administered by trained psychologists in the Indian sites and by trained healthcare workers in the Tanzania site. A group of experienced scientists with expertise in developmental, neuropsychological and neurophysiological assessments are responsible for the training and standardisation in the different sites. Several domains will be measured and compared between the study groups within each of the studies. In these analyses, we will initially use the Student’s t-test for the crude analyses and multiple linear regression models to adjust for potential baseline differences and when measuring effect modification. Each of the studies has several outcomes as we will compare both linear and ponderal growth, in addition to all the above-mentioned neurodevelopmental measures between the study groups. Thus, there will be several comparisons from each study, and negative and positive effects will be reported to avoid focusing on spurious positive findings. For each of the planned publications, we will make a detailed plan of analysis before commencing the analysis. In these plans, we will include sections on how to deal with multiple comparisons and whether post hoc adjustments will be done. In addition to these standard per-protocol analysis, we will consider instrumental variable analysis in an attempt to estimate the true effect of vitamin B12 had it been given to all participants in the scheduled doses and intervals. The random allocation will be the instrument in these analyses. For per-protocol analysis, participants who received less than 50% of the projected doses during the period of intervention will not be included in the analyses, well acknowledging that the ensuing effect estimates may not only be biased but will certainly represent an effect higher than what can be achieved even in our well-resourced study setting. For our subgroup analyses, we will include interaction terms to measure whether or not the subgrouping variable significantly modifies the effect of the exposure of interest. All of our analyses will initially be done according to intent to treat. We will not be able to retain the complete number of children from these studies. We will compare the features of the population that is included in this analysis with the population that we failed to re-enrol into the study. We will also detect risk factors for poor neurodevelopment in multiple linear or binomial regression models. We will include socioeconomic and seasonal factors and dietary intake as exposures in the analyses. A significance level of 0.05 will be used.