Vitamin B12 supplementation for growth, development, and cognition in children

Objectives

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To determine the effects of oral vitamin B₁₂ supplementation on growth, development, and cognition in children under 12 years of age.

Background

Description of the condition

Vitamin B₁₂ is a water‐soluble vitamin that is required for DNA synthesis and methylation, and folate metabolism. Vitamin B₁₂ is needed to convert folate to its active form (tetrahydrofolate) to synthesise purines (Banerjee 1990; Shane 1985; Stabler 2020; Stover 2004); and to convert homocysteine to methionine to produce of S‐adenosylmethionine, which is needed in over 100 methylation reactions (Froese 2019; Green 2017; Pourié 2022). As a result, vitamin B₁₂ is critical for cell division, DNA synthesis and methylation, red blood cell formation (erythropoiesis), and neurotransmitter synthesis (Reynolds 2006; Stover 2004; Winter‐Vann 2003).

Vitamin B₁₂ is found predominantly in animal‐source foods, including meat, fish, poultry, eggs, and dairy products (Stabler 2020). Vitamin B₁₂ is also found in nutritional yeast, fortified foods, and cereals (Green 2017), and in some plant‐based foods, such as shiitake mushrooms, dried purple laver (nori), and tempeh (Watanabe 2014). Daily requirements of vitamin B₁₂ vary by age and pregnancy or lactation status. In adults, the recommended intake is 2.4 µg/day to 2.8 µg/day, and in children this ranges from 0.4 µg/day to 1.8 µg/day (Institute of Medicine 1998; Institute of Medicine 2006).

Vitamin B₁₂ deficiency

Vitamin B₁₂ deficiency is an important public health problem (Allen 2009; Brito 2015; Green 2017; McLean 2008), with a higher prevalence in the elderly (Allen 2010; Green 2017), pregnant women (Finkelstein 2015), and young children (Green 2017), particularly in resource‐limited settings where dietary intake or bioavailability is low and infectious diseases that impair gastrointestinal absorption are common. The reported prevalence of vitamin B₁₂ deficiency ranges from 30% to 40% in Latin America (Anaya‐Loyola 2019; Arazo‐Rusindo 2021; Brito 2015), to up to 50% to 80% in South Asia (Finkelstein 2021b; Gonmei 2018). In children, the reported prevalence of vitamin B₁₂ deficiency ranges from 5% to 64% in studies in Mexico (5%; De la Cruz‐Góngora 2021), Kenya (10%; Williams 2018), Nepal (17%; Ulak 2016), Tanzania (26%; Bellows 2017), Guatemala (30%; Jones 2007), and India (64%; Kadiyala 2021), although there is limited population representative data.

Vitamin B₁₂ is synthesised by bacteria and is obtained in the diet predominantly through consumption of animal‐source foods (Allen 2008; Allen 2009; Stabler 2020). In addition to inadequate dietary intake, vitamin B₁₂ deficiency can result from low bioavailability or impaired absorption, due to pernicious anemia (an autoimmune disease affecting intrinsic factor, required for vitamin B₁₂ absorption), gastrointestinal infections (e.g. Helicobacter pylori, intestinal helminths) (Allen 2008; Finkelstein 2015; Green 2017; Layden 2018), or other gastrointestinal diseases (e.g. inflammatory bowel disease) (Finkelstein 2015). Infants and young children are at increased risk of vitamin B₁₂ deficiency, due in part to inadequate vitamin B₁₂ supply in utero and status at birth (Bergen 2016; Finkelstein 2019; Finkelstein 2021a; Greibe 2013), and low vitamin B₁₂ content in breast milk (Green 2017; Hay 2010); children born to mothers with vitamin B₁₂ deficiency and exclusively breastfed infants are at increased risk of vitamin B₁₂ deficiency (Allen 2008; Anaya‐Loyola 2020). Vitamin B₁₂ deficiency may also develop in young children due to low vitamin B₁₂ content in complementary foods and low intake of animal‐source foods (Allen 2008), and as dietary requirements increase with age (i.e. 0 to 6 months: 0.4 μg/day; 7 to 17 months: 0.5 μg/day; 1 to 3 years: 0.9 μg/day; 4 to 8 years: 1.2 μg/day; 9 to 13 years: 1.8 μg/day) (Institute of Medicine 1998; Institute of Medicine 2006).

The classical manifestation of vitamin B₁₂ deficiency is hematological, as pernicious anemia and megaloblastic anemia (enlarged red blood cells, that results from impairments in DNA synthesis during erythropoiesis) (Green 2017; Stabler 2020). Other symptoms of vitamin B₁₂ deficiency, including neurologic manifestations and cognitive impairment (Green 2017), can develop in the absence of anemia (Lindenbaum 1988). In children, vitamin B₁₂ deficiency has been associated with adverse health outcomes, including anemia (Kose 2020; Mullikin 2018; Qasrawi 2022), and impaired growth (Strand 2015), cognitive function (Dror 2008; Venkatramanan 2016), and development (Venkatramanan 2016), which may be irreversible (Black 2008; Pepper 2011). This Cochrane Review focuses on the effects of vitamin B₁₂ supplementation among apparently healthy children, regardless of baseline vitamin B₁₂ status.

Vitamin B₁₂ biomarkers

Vitamin B₁₂ status is evaluated through circulating (i.e. total vitamin B₁₂, holo‐transcobalamin) or functional (i.e. methylmalonic acid, MMA) biomarkers. Total vitamin B₁₂ is the most used biomarker of vitamin B₁₂ status, and vitamin B₁₂ deficiency and insufficiency are commonly defined as total vitamin B₁₂ less than 148 pmol/L and less than 221 pmol/L, respectively (Allen 2018; Yetley 2011). Vitamin B₁₂ concentrations in circulation peak approximately seven hours after administration of a supplement (Devi 2020; Miller 2020); its half‐life in plasma is estimated to be approximately six days, in contrast to approximately 12 months storage in the liver (Adams 1963).

Holo‐transcobalamin (approximately 30% of vitamin B₁₂ in circulation) is the form of vitamin B₁₂ that is transported into cells via transcobalamin receptors (Quadros 2010; Quadros 2013). Holo‐transcobalamin (bound to cyanocobalamin) increases approximately three‐fold 24 hours after administration of vitamin B₁₂ supplementation (Hardlei 2010).

Methylmalonic acid is a functional and specific biomarker of vitamin B₁₂ status and is elevated in the context of vitamin B₁₂ deficiency (Allen 2018; Yetley 2011). In contrast, total homocysteine is a non‐specific biomarker for vitamin B₁₂ status, as it is influenced by nutritional (e.g. folate status) and non‐nutritional (e.g. renal function) factors (Green 2017). The use of both circulating (e.g. total vitamin B₁₂) and functional (e.g. MMA) biomarkers is recommended for vitamin B₁₂ assessment at the population level (Allen 2018; Yetley 2011). However, specific cut‐offs for vitamin B₁₂ deficiency have not been established for children, and the use of different biomarkers and cut‐offs constrains comparability of findings across studies.

Description of the intervention

Supplementation with oral vitamin B₁₂ is a common strategy to improve vitamin B₁₂ status and treat vitamin B₁₂ deficiency (Stabler 2013). Cyanocobalamin and methylcobalamin are commonly used formulations (Obeid 2015), with standard doses of oral vitamin B₁₂ ranging from 2.0 µg/day to 10.0 µg/day to prevent deficiency in healthy or higher risk (e.g. vegan) individuals (Stabler 2013). Daily supplementation is usually recommended (Carmel 2008; Stabler 2013) for a duration of four to six weeks, although intermittent administration (e.g. weekly or every two weeks) has also been reported (Bhowmik 2021). Intramuscular injection or intravenous administration of vitamin B₁₂ are more commonly used in patients with malabsorption disorders; oral vitamin B₁₂ supplementation at higher doses (e.g. 1000 µg/day to 2000 µg/day) may need to be continued indefinitely (Carmel 2008; Stabler 2013).

There is no tolerable upper limit for vitamin B₁₂, and no toxic level has been established (Allen 2018; Green 2017; Institute of Medicine 1998). A water‐soluble vitamin, vitamin B₁₂ is stored in the liver and any excess that is not required by the body is excreted in urine.

In children, vitamin B₁₂ can be administered via oral supplements (i.e. capsules, tablets, dispersible tablets); liquids, drops or syrups; and micronutrient powders, ready‐to‐use therapeutic foods, or consumption of fortified foods.

This review focuses on oral vitamin B₁₂ supplementation (in the form of capsules, tablets, dispersible tablets, liquids, drops, soft gels/pastes, or fortified foods), with any formulation (e.g. cyanocobalamin, methylcobalamin), dose, frequency (i.e. daily or intermittently), and duration of supplementation. This review will not include other types of interventions such as intravenous or intramuscular injections, which are primarily administered to patients with malabsorption disorders (i.e. medical conditions that may affect vitamin B₁₂ absorption or metabolism) at pharmaceutical doses.

How the intervention might work

The goal of vitamin B₁₂ supplementation is to improve vitamin B₁₂ status and health outcomes. Dietary vitamin B₁₂ is tightly bound to proteins which must be degraded to bind to intrinsic factor for absorption (20% to 40%) via receptor‐mediated endocytosis in the ileum (Green 2017; Rashid 2021; Stabler 2020). In supplements, approximately 50% of vitamin B₁₂ is absorbed with a 1.0 µg oral dose in the ileum, and absorption decreases with increasing dosages, as active transport is saturable (Allen 2018); additionally, up to 1% of free cobalamin is passively absorbed in the terminal ileum (Green 2017; Stabler 2013).

The mechanisms of vitamin B₁₂ in growth, development, and cognition have not been fully elucidated. Evidence from laboratory studies of mouse models of vitamin B₁₂ deficiency have suggested DNA hypomethylation (Fernàndez‐Roig 2012), decreased myelination (Arora 2019), and anatomical (Arora 2017) and behavioral alterations in the affected mice (Arora 2017). In humans, vitamin B₁₂ deficiency — and resulting impairments in DNA synthesis and methylation, cell division, folate metabolism, erythropoiesis, and neurotransmitter synthesis (Green 2017; Green 2022; Stover 2004) — may impair brain development (Pepper 2011; Pourié 2022), cognitive function (Ars 2019; D'souza 2021), growth (Strand 2015), and anemia (Green 2022; Kose 2020; Mullikin 2018; Qasrawi 2022), although the mechanisms involved have not been fully established.

In observational studies in children, higher vitamin B₁₂ status has been associated with improved cognitive and development outcomes (e.g. mental developmental scores) (Strand 2013), cognition function (Venkatramanan 2016), and school performance (Duong 2015; Louwman 2000). Vitamin B₁₂ supplementation in children has been shown to improve vitamin B₁₂ status (Bahadir 2014). Evidence from randomised trials in children suggest that vitamin B₁₂ supplementation may confer longer term benefits to child health outcomes, including improved growth (Strand 2015) and gross motor development (Kvestad 2015).

Why it is important to do this review

Vitamin B₁₂ deficiency is an important public health problem, with a high burden in pregnant women (Finkelstein 2015) and young children (Green 2017). Inadequate supply of vitamin B₁₂in utero (Finkelstein 2015; Molloy 2018) and during childhood (Doyle 1989; Dror 2014) can impair child growth and development. Emerging evidence suggests that vitamin B₁₂ deficiency is a risk factor for adverse health outcomes in children (Duong 2015; Louwman 2000; Venkatramanan 2016); and higher vitamin B₁₂ status (Duong 2015; Strand 2013) and vitamin B₁₂ supplementation (Kvestad 2015; Strand 2015) have been associated with improved child health outcomes. The high burden of vitamin B₁₂ deficiency in children and associated risk of adverse health outcomes have stimulated interest in fortification and interventions with vitamin B₁₂ (Allen 2010). However, vitamin B₁₂ screening or interventions are not included in routine pediatric care, and the efficacy of vitamin B₁₂ supplementation on child health outcomes has not been established.

No systematic reviews have been conducted to date to examine the effects of vitamin B₁₂ supplementation on child health outcomes — including growth, development, and cognition. This review will complement findings of previous Cochrane Reviews, including: vitamin B₁₂ (Finkelstein 2020) and multiple micronutrient supplementation during pregnancy (Keats 2019); vitamin D (Huey 2020), zinc (Gogia 2012; Lassi 2020), iron (De‐Regil 2011), and long‐chain polyunsaturated fatty acid supplementation in children (Delgado‐Noguera 2015).

Objectives

To determine the effects of oral vitamin B₁₂ supplementation on growth, development, and cognition in children under 12 years of age.

Methods

Criteria for considering studies for this review

Types of studies

We will include the following types of studies.

1. Randomised controlled trials (RCTs), with randomisation at the individual or cluster level.

2. Quasi‐randomised RCTs (i.e. in which treatment was allocated by a predictable method such as alternation, birthdate, alphabetical order, or case record number).

We will exclude cross‐over trials or observational studies.

Types of participants

Participants will include children from birth to less than 12 years of age, regardless of geographic location and baseline vitamin B₁₂ status. We will only consider trials where the intervention is administered directly to the child; thus, we will exclude studies of interventions targeted only to pregnant or breastfeeding women. We will also exclude studies where the intervention is targeted to participants with critical illnesses or severe co‐morbidities (e.g. HIV, tuberculosis, cancer), including malabsorption disorders or other medical conditions that may affect vitamin B₁₂ absorption or metabolism (e.g. inflammatory bowel disease, atrophic gastritis, H pylori infection).

If a study includes participants above 12 years of age, we will attempt to contact study authors to obtain the specific data for the participants younger than 12 years of age.

Types of interventions

We will include studies of oral supplementation of vitamin B₁₂, alone or combined with other vitamins or minerals (e.g. iron, folic acid, other B vitamins).

We will include vitamin B₁₂ supplementation delivered via a tablet, capsule, dispersible tablet, syrup, drops, micronutrient powders, soft gels/pastes, and fortified foods. No restrictions will be placed on formulation, dose, frequency, or duration of supplementation. We will exclude interventions of intravenous or intramuscular injections of vitamin B₁₂.

We will include studies with co‐interventions (e.g. nutrition education) if the administered co‐interventions were the same in both the intervention and control groups.

Comparators

We will include studies where the comparators are placebo, no intervention, or other micronutrient supplements that do not contain vitamin B₁₂.

Main comparisons

1. Supplementation with vitamin B₁₂ alone versus placebo.

2. Supplementation with vitamin B₁₂ alone versus no intervention.

3. Supplementation with vitamin B₁₂ with other micronutrients, compared to supplements with the same formulation without vitamin B₁₂ (e.g. vitamin B₁₂ with iron‐folic acid versus iron‐folic acid alone).

Types of outcome measures

Primary outcomes

1. Growth (height or length‐for‐age, reported continuously as height‐for‐age Z score (HAZ) or length‐for‐age Z score (LAZ), according to the World Health Organization (WHO) child growth standards.

2. Cognitive function using validated instruments, including general cognitive ability or specific cognitive domains, such as the Bayley Scales of Infant Development (BSID) (Weiss 2010), Wechsler Intelligence Scales for Children (Wechsler 1994), Digit Span Test (Wechsler 1994), or Developmental Neuropsychological Assessment (Brooks 2009).

3. Development outcomes, such as motor skill development, language, and socio‐emotional and adaptive behavior, using validated instruments such as BSID (Weiss 2010).

4. Any adverse effects, as defined by trial authors.

Secondary outcomes

1. Vitamin B₁₂ status: serum/plasma vitamin B₁₂ (pmol/L), MMA (μmol/L), holo‐transcobalamin (μmol/L); vitamin B₁₂ deficiency/insufficiency, as defined by trial authors (e.g. less than 148 pmol/L, less than 221 pmol/L) (Allen 2018; Yetley 2011).

2. Hemoglobin (g/L) adjusted for altitude (and smoking if reported), when applicable; anemia, defined using age‐specific cut‐offs

3. Megaloblastic anemia, as defined by trial authors.

4. Body composition, as defined by trial authors.

5. Health‐related quality of life, as defined by trial authors, including role limitations due to physical, personal, or emotional health problems, physical functioning, emotional well‐being, social functioning, bodily pain, energy or fatigue, and general health perceptions, evaluated by validated instruments (e.g. 36‐Item Short Form Health Survey (SF‐36) (Ware 1992), EuroQol‐5 dimensions (EQ‐5D) (EuroQol 1990)).

6. Morbidity (proportion of children with at least one reported illness), as defined by trial authors.

7. All‐cause mortality, defined as death from any cause.

Timing of outcome assessment

• Adverse events, all‐cause mortality, morbidity will be extracted and analysed at any time during the study period after participants were randomised to intervention or comparator groups.

• Growth, cognitive function, development outcomes, vitamin B₁₂ status, hemoglobin, anemia, body composition, and health‐related quality of life: group outcomes will be extracted and analysed over the short (immediately after intervention), medium (one to six months), and long (more than six months) term.

Search methods for identification of studies

Electronic searches

An information specialist (Erin Eldemire) at Cornell University will search international and regional electronic databases and trial registries listed below without date or language restrictions.

1. Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library, which includes the Cochrane Developmental, Psychosocial and Learning Problems Specialised Register.

2. PubMed (1946 onwards) National Library of Medicine (www.ncbi.nlm.nih.gov/pubmed).

3. Embase Ovid (1980 onwards).

4. CINAHL EBSCO (Cumulative Index to Nursing and Allied Health Literature; 1982 onwards).

5. Centre for Agriculture and Biosciences International (CABI): CAB Abstracts and Global Health Web of Science (1973 onwards).

6. PsycINFO EBSCO (1887 onwards).

7. ERIC EBSCO (Educational Resources Information Center; 2004 onwards).

8. Web of Science Core Collection (Clarivate; 1900 onwards), we will search the following databases: Science Citation Index Expanded (SCI‐EXPANDED), Social Sciences Citation Index (SSCI), Conference Proceedings Citation Index‐Science (CPCI‐S), Conference Proceedings Citation Index‐Social Science & Humanities (CPCI‐SSH), and Emerging Sources Citation Index (ESCI).

9. Cochrane Database of Systematic Reviews (CDSR; current issue), part of the Cochrane Library.

10. Proquest Dissertations & Theses A&I (all available years).

11. Global Index Medicus (www.globalindexmedicus.net; we will search African Index Medicus (AIM), Latin America and the Caribbean Literature on Health Sciences (LILACS), Index Medicus for Eastern Mediterranean Region (IMEMR), Index Medicus for South‐East Asia Region (IMSEAR), and Western Pacific Region Index Medicus (WPRIM).

12. IBECS (ibecs.isciii.es).

13. Scientific Electronic Library Online (SciELO; www.scielo.br).

14. WHO International Clinical Trials Registry Platform (ICTRP; apps.who.int/trialsearch).

15. EU Clinical Trials Register (www.clinicaltrialsregister.eu/ctr‐search).

16. Epistemonikos (limited to systematic reviews; www.epistemonikos.org).

17. Scopus Elsevier (1788 onwards).

The PubMed search strategy is in Appendix 1, and will be translated for other databases. No restrictions will be applied for language or publication date.

Searching other resources

We will search reference lists of relevant publications (including trials, reviews, meta‐analyses, reports) identified by the electronic searches, to identify additional eligible studies.

We will attempt to obtain information on relevant unpublished, planned, and ongoing trials by contacting the authors of included trials and other organisations: Nutrition International (www.nutritionintl.org), the International Micronutrient Malnutrition Prevention and Control Programme (IMMPaCt; www.cdc.gov/nutrition/micronutrient-malnutrition/index.html) from the US Centers for Disease Control and Prevention (CDC), the WHO (www.who.int), and the United Nations Children’s Fund (UNICEF; www.unicef.org).

Data collection and analysis

Selection of studies

Two review authors (CEGL, SV) will independently screen titles and abstracts of records identified in the search strategy (Electronic searches) for inclusion using Covidence. We will retrieve full‐text records for abstracts identified, and review them in duplicate (CEGL, SV) to evaluate eligibility (Criteria for considering studies for this review). Discrepancies will be resolved through discussion and consultation with the senior author (JLF). We will attempt to contact study authors for additional information if studies are published only as abstracts or conference proceedings, or if there is limited information on the study design or methods. We will present a flow diagram (PRISMA) to summarise the number of records identified, included, and excluded in the review (Page 2021).

Data extraction and management

We will develop a data extraction form for this review and pilot test it using three included studies. For eligible studies, two review authors (CEGL, SV) will independently extract information using this form. Extracted information will include study participants (age, sex, location, baseline vitamin B₁₂ status), intervention (type, formulation, dose, duration, comparison groups), study methods (method of randomisation, unit of randomisation, sample size), outcomes relevant to the review, methods of outcome assessment, trial dates, sources of funding, adverse effects, the declaration of interests of study authors, and information to allow for risk of bias judgments. Any discrepancies will be resolved through discussion and consultation with the senior author (JLF). We will enter data extracted from included studies into Review Manager software (Review Manager 2020), and will check for accuracy.

Assessment of risk of bias in included studies

Two review authors (CEGL, SV) will assess risk of bias for all the primary outcomes (i.e. growth, cognitive function, development outcomes) in the long term (more than six months after intervention), using the RoB 2 (Higgins 2022c). We are interested in the effect of the assignment.

For each of the domains, review authors will answer a set of signaling questions with a set of five possible responses (i.e. ‘yes’, ‘probably yes’, ‘probably no’, ‘no’, ‘no information’) and assign a risk of bias level (i.e. ‘low’, ‘some concerns’, or ‘high’). For cluster‐randomised trials, we will use the modified RoB 2 for cluster‐randomised trials. We will use the Microsoft Excel RoB 2 tools (www.riskofbias.info). Any discrepancies will be resolved through discussion and consultation with the senior author (JLF).

(1) Bias arising from the randomisation process

To evaluate potential bias arising from randomisation, we will describe in detail the randomisation procedures in each study: if the method of allocation sequence generation was random, adequately concealed, and whether the differences at baseline between intervention groups are compatible with chance.

We will evaluate the randomisation process as:

• low risk of bias (e.g. allocation sequence was random, adequately concealed, and differences among intervention groups at baseline are compatible with chance);

• some concerns (e.g. allocation sequence was random and adequately concealed, but baseline differences among interventions groups suggest a problem with the randomisation); or

• high risk of bias (e.g. allocation sequence was not concealed, or baseline differences suggest a problem with the randomisation).

(2) Bias due to deviations from intended interventions

We will describe methods used for blinding study participants, caregivers, or people delivering the intervention from knowledge of which intervention was received. We will evaluate any deviations from the intended intervention that arose due to the trial context and if appropriate analyses were used to estimate the effect of assignment to intervention.

We will evaluate the deviation from intended interventions as:

• low risk of bias (e.g. study participants, caregivers, or people delivering the intervention were unaware of intervention groups or were aware of them, but no deviations from intended intervention arose because of the trial context);

• some concerns (e.g. study participants, caregivers, or people delivering the intervention were aware of intervention groups and this resulted in a deviation from intended intervention that is not likely to affect the outcome); or

• high risk of bias (e.g. study participants, caregivers, or people delivering the intervention were aware of intervention groups and this resulted in deviation from intended intervention that is likely to affect the outcome).

(3) Bias due to missing outcome data

For each included study and outcome, we will examine completeness of data, attrition, and exclusions from analyses.

We will assess outcome data as:

• low risk of bias (e.g. outcome data were available for all, or nearly all, randomised participants);

• some concerns (e.g. outcome data were not available for all, or nearly all, randomised participants, and it is not likely that missingness of the outcome depended on its true value); or

• high risk of bias (e.g. outcome data were not available for all, or nearly all, randomised participants and it likely that missingness of the outcome depended on its true value).

(4) Bias in measurement of the outcome

For each included study and outcome, we will describe methods used for measurement of the outcome, whether the same method was used among intervention and control groups, the outcome assessor (i.e. participant, intervention provider, outcome assessor), and the methods that were used, if any, to blind the outcome assessor from knowledge of intervention assignment. For anthropometric outcomes, we will describe the accuracy and precision of the methods or equipment used, as well as the intra and inter observer differences in measurements.

We will assess the measurement of the outcome as:

• low risk of bias (e.g. method of outcome assessment was appropriate and did not differ between groups and outcome assessors were blinded from the intervention assignment);

• some concerns (e.g. method of outcome assessment was appropriate and did not differ between groups). It is unlikely that assessment of the outcome was influenced by knowledge of intervention received; or

• high risk of bias (e.g. method of outcome assessment was inappropriate or differed between groups).

(5) Bias in selection of the reported result

We will examine the pre‐specified analysis plan described in the trial registration and study protocols, and the reported results of included studies. We will assess if the analysis plan was followed, and if the numerical results reported could have been selected, on the basis of the results, from multiple eligible outcome measurements or analyses of the data.

We will assess the reported results as:

• low risk of bias (e.g. data were analysed according to pre‐specified analysis plan and results being assessed is not likely to have been selected, on the basis of the results, from multiple outcome measurements or data analyses);

• some concerns (e.g. data were not analysed according to pre‐specified analysis plan, but results being assessed is not likely to have been selected, on the basis of the results, from multiple outcome measurements or data analyses); or

• high risk of bias (e.g. data were not analysed according to pre‐specified analysis plan, and results being assessed are likely to have been selected, on the basis of the results, from multiple outcome measurements or data analyses).

If a study protocol, trial registration, or pre‐specified data analysis plan is not available, we will attempt to contact study authors to request this information. In the case of inadequate reporting of outcomes, we will contact study authors. If this information is not available, we will evaluate the reported results as some concerns of risk of bias, and report based on RoB 2 guidance (Sterne 2019).

Dichotomous data

For dichotomous outcomes, we will report data as risk ratios (RR) and 95% confidence intervals (CIs), based on their increased use in clinical and public health contexts and more intuitive interpretation compared to other measures (Higgins 2022a).

Continuous data

When the outcome measure is the same, we will report mean differences (MD) with their corresponding 95% CIs. If the scale used to assess the outcome differs between studies, we will report the standardised mean difference (SMD) and 95% CIs. We will ensure scales used measure the effect in the same direction, and will convert data into the required direction and format if necessary (Higgins 2022a).

Cluster‐randomised trials

We will include cluster‐randomised trials in the analyses along with individually randomised trials. In order to adjust the standard errors of cluster‐randomised trials, we will use an internal estimate of the intracluster correlation coefficient (ICC), if available, or an ICC from a similar trial or population (Higgins 2022b). If an ICC is not available for a cluster‐randomised trial, we will conduct sensitivity analyses to investigate the effects of higher compared to lower ICC values (i.e. the effects of any assumptions made such as the value of the ICC used for cluster‐randomised trials) on findings. We plan to combine the results from individually and cluster‐randomised trials if there is little heterogeneity between the study designs and if it is unlikely that the unit of randomisation alters the effect of the intervention.

Cross‐over trials

Cross‐over trials will not be included in this review.

Multiple intervention arms

For trials with multiple intervention arms, we will only summarise in detail and include in analyses intervention groups relevant to this review. For analysis, we will create a single pairwise comparison by combining groups, if appropriate, or selecting one pair of interventions. All intervention groups of multi‐intervention trials will be mentioned in the ‘Characteristics of included studies’ table.

Outcome assessment

If outcomes are reported at multiple time points in trials, we will consider the primary endpoint of the trial as the primary outcome for meta‐analyses. If multiple cognitive tests are assessed within the same cognitive domain, we will compute a standardised mean difference from the mean of the standardised scores of the cognitive tests.

Dealing with missing data

We will attempt to contact trial authors to obtain missing data (e.g. missing outcomes, summary data, participant, or study characteristics), and report correspondences in the Appendix. If missing data are provided, we will conduct meta‐analyses according to the intention‐to‐treat principles, retaining participants in the group to which they were originally randomised, regardless of whether the intervention was received. If study authors do not respond or are unable to supply the missing data upon request, we will analyse the available data only (Deeks 2022). Missing data will not be imputed.

We will report levels of attrition for included studies in the risk of bias tables, which will inform our risk of bias judgments. Sensitivity analyses will be conducted to examine the potential impact of studies with high levels of attrition in the overall results.

Assessment of heterogeneity

Clinical (participant characteristics, interventions, outcomes studied) and methodological (study design, outcome measurement tools, risk of bias) heterogeneity will be used to inform evaluation of heterogeneity.

To assess statistical heterogeneity, we will use forest plots and estimate the Chi², I², and Tau² statistics. The Chi² statistic assesses whether the differences in the observed results are compatible with chance. We will interpret the I² values according to Deeks 2022, where 30% to 60% may represent moderate heterogeneity, 50% to 90% may represent substantial heterogeneity, and 75% to 100% represents considerable heterogeneity. We will use Tau² as an estimate of the between‐study variation in random‐effects models (Deeks 2022). We will regard heterogeneity as substantial if I² is greater than 30% and either Tau² is greater than zero, or the P value of the Chi² test is less than 0.10.

If there is substantial heterogeneity, we will perform prespecified subgroup analyses to examine differences between these groups (Subgroup analysis and investigation of heterogeneity). For outcomes with substantial heterogeneity (based on clinical and methodological evaluation), we will not report a pooled estimate.

Assessment of reporting biases

We will search trial registries and other electronic databases to attempt to mitigate the risk of non‐reporting bias. Additionally, we will search for study protocols and trial registrations for each included study to evaluate potential discrepancies in methods or outcomes described. If there are 10 or more studies for a particular outcome in meta‐analyses, we will generate a funnel plot and assess its symmetry visually and conduct sensitivity analyses (i.e. comparing fixed‐effect and random‐effects estimates, and regression‐based method by Moreno 2009) to investigate asymmetry, if suggested by visual assessment. We acknowledge that there are many causes for funnel plot asymmetry, such as non‐reporting bias, methodological issues, and clinical heterogeneity (Page 2022).

Data synthesis

We will use Review Manager 2020 for statistical analyses. We will use fixed‐effect meta‐analysis for combining data for trials examining the same intervention and with similar methods and population characteristics. We will use the inverse‐variance method for meta‐analysis to minimise the imprecision of the pooled effect estimate (Deeks 2022).

If there is substantial clinical (e.g. different populations, different doses of interventions) or methodological (e.g. different study design or outcome measure tools) heterogeneity to suggests underlying treatment effects may differ between trials (Assessment of heterogeneity) we will use a random‐effects model with inverse‐variance (DerSimonian 1986). The summary estimate from random‐effects meta‐analyses will be considered the average of the range of possible treatment effects, and we will discuss the clinical implications of treatment effects differing between trials. If random‐effects meta‐analyses are used, we will present the results as the average treatment effect with 95% CIs and the Tau² and I² estimates.

If meta‐analyses are not conducted (e.g. due to incompletely reported outcome or effect estimates, different effect measures used across studies, or studies with high risk of bias), we will describe findings from individual trials, and consider synthesis and presentation of findings across studies using other methods, such as summarising effect estimates and combining P values (McKenzie 2022).

Subgroup analysis and investigation of heterogeneity

We will investigate heterogeneity using subgroup analyses and sensitivity analyses.

Where data are available, we plan to conduct the following subgroup analyses to examine the potential effect of participant and intervention characteristics on the response to vitamin B₁₂ supplementation.

1. Age groups: less than 2 years, 2 to less than 5 years, 5 to less than 10 years, and 10 to less than 12 years.

2. Sex: male, female, mixed, or not reported.

3. Baseline vitamin B₁₂ status: vitamin B₁₂ deficiency (e.g. less than 148 pmol/L), vitamin B₁₂ insufficiency (e.g. less than 221 pmol/L).

4. Regimen: daily supplementation versus intermittent supplementation versus other.

5. Dose of vitamin B₁₂: single versus multiple recommended daily allowance (Institute of Medicine 1998; Institute of Medicine 2006).

6. Form of vitamin B₁₂ supplementation: cyanocobalamin versus methylcobalamin versus other form.

7. Cognitive assessment tools by age groups above (i.e. less than 2 years, 2 to less than 5 years, 5 to less than 10 years, and 10 to less than 12 years).

These subgroups were developed a priori based on variation in vitamin B₁₂ requirements by age group, potential to benefit based on vitamin B₁₂ status at baseline, variation in selected outcomes (e.g. growth) by sex, and anticipated variation in effects of vitamin B₁₂ supplementation on status by frequency and duration of intervention.

We will only use the primary outcomes in any subgroup analysis. We will conduct a standard test for heterogeneity across subgroup results and report an interaction test I² value (Deeks 2022).

Sensitivity analysis

We plan to perform sensitivity analyses to explore the influence of the following factors (when applicable) on effect size.

1. Risk of bias of the included studies, by excluding studies assessed as having a high risk of bias (as determined by concealment of allocation, high attrition rates, or both).

2. Fixed‐effect or random‐effects analyses, to examine the potential impact of using a random‐effects model instead of a fixed‐effect approach in the effect estimate for outcomes with clinical heterogeneity.

3. ICC value used for cluster‐randomised trials, to examine the impact of higher compared to lower ICC values for cluster‐randomised trials on findings.

If quasi‐RCTs are identified and included in analyses, we plan to also consider sensitivity analyses without quasi‐RCTs.

Summary of findings and assessment of the certainty of the evidence

We will use the GRADE approach to evaluate the certainty of the evidence for the primary outcomes (Balshem 2011). The GRADE approach considers five domains (i.e. study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of the evidence. The certainty of the evidence is downgraded from ‘high certainty’ by one or two levels for serious or very serious limitations.

We will import data from Review Manager 2020 using the GRADEpro Guideline Development tool to create summary of findings tables, which will include a summary of the intervention effect and a measure of certainty for each of the primary outcomes in the long term (i.e. more than six months); and main comparisons (i.e. vitamin B₁₂ versus placebo; vitamin B₁₂ versus no intervention; and vitamin B₁₂ with other micronutrients, compared to the same formulation without vitamin B₁₂).

What's new

Date	Event	Description
15 November 2022	Amended	Correcting the order of authors on the by‐line.

History

Protocol first published: Issue 11, 2022

Appendices

Appendix 1. PubMed search strategy

("vitamin b 12"[MeSH Terms] OR "transcobalamins"[MeSH Terms] OR "vitamin b 12 deficiency"[MeSH Terms] OR ("B‐12"[Title/Abstract] OR "B12"[Title/Abstract] OR "B 12"[Title/Abstract] OR "cobalamin*"[Title/Abstract] OR "transcobalamin*"[Title/Abstract] OR "cyanocobalamin*"[Title/Abstract] OR "methylcobalamin*"[Title/Abstract] OR "hydroxycobalamin*"[Title/Abstract] OR "holotranscobalamin"[Title/Abstract] OR "cobamide*"[Title/Abstract] OR “eritron” [Title/Abstract])) AND ("infant"[MeSH Terms] OR "child"[MeSH Terms] OR ("infant*"[Title/Abstract] OR "newborn*"[Title/Abstract] OR "neonat*"[Title/Abstract] OR "baby"[Title/Abstract] OR "babies*"[Title/Abstract] OR "toddler*"[Title/Abstract] OR "preschool*"[Title/Abstract] OR "pre school*"[Title/Abstract] OR "schoolchild*"[Title/Abstract] OR "school age*"[Title/Abstract] OR "child*"[Title/Abstract] OR "boy"[Title/Abstract] OR "boys"[Title/Abstract] OR "girl*"[Title/Abstract] OR "preteen*"[Title/Abstract] OR “pre‐teen*” [Title/Abstract] OR "adolescen*"[Title/Abstract] OR "teen*"[Title/Abstract] OR "pubescent*"[Title/Abstract] OR "prepubescent*" [Title/Abstract] OR "youth*"[Title/Abstract] OR "young person*"[Title/Abstract] OR "young people*"[Title/Abstract] OR "first thousand days of life"[Title/Abstract] OR "underfive*"[Title/Abstract] OR "under five*"[Title/Abstract] OR "less than five"[Title/Abstract] OR "pre‐kindergarten"[Title/Abstract] OR "pre‐K"[Title/Abstract] OR "kindergarten*"[Title/Abstract] OR "elementary age*"[Title/Abstract] OR "primary school"[Title/Abstract])) AND ("randomized controlled trial"[Publication Type] OR "controlled clinical trial"[Publication Type] OR "randomized"[Title/Abstract] OR "placebo"[Title/Abstract] OR "drug therapy"[MeSH Subheading] OR "randomly"[Title/Abstract] OR "trial"[Title/Abstract] OR "groups"[Title/Abstract]) NOT ((animals[MeSH Terms]) NOT (humans[MeSH Terms]))

Contributions of authors

CEGL wrote the initial draft of the protocol. CEGL, SV, and XEPM wrote the Background section. CEGL developed the search strategy and Methods section with guidance from JLF. JLW and RK provided clinical expertise, and KSC and JLF provided expert guidance on statistical and epidemiological methods in the protocol. JLF, XEPM, SV, JLW, RK, and KSC revised the protocol. All authors reviewed and approved the final version of the protocol. JLF coordinated the protocol and is the guarantor.

Sources of support

Internal sources

• None, Other

  Not applicable

External sources

• None, Other

  Not applicable

Declarations of interest

Cristina E Güitrón Leal reports a Fellowship from Consejo Nacional de Ciencia y Tecnología, Mexico, for PhD tuition and living expenses. This fellowship is not related to this Cochrane Review or its subject. CGL declares no known conflicts of interest.

Ximena E Palma Molina declares no known conflicts of interest.

Dr Sudha Venkatramanan declares no known conflicts of interest.

Dr Jennifer L Williams is a PhD nurse epidemiologist and family nurse practitioner for the United States Public Health Service (stationed at the US Centers for Disease Control and Prevention (CDC)*. JLW declares no known conflicts of interest.

Dr Rebecca Kuriyan declares no known conflicts of interest.

Dr Krista S Crider is an employee of the US government at the CDC*. KSC declares no known conflicts of interest.

Dr Julia L Finkelstein is a Principal Investigator on research grants from the US CDC (population‐based biomarker survey among women of reproductive age, and a randomised trial of micronutrient interventions in women of reproductive age), the US Department of Agriculture (effects of micronutrient interventions on gastrointestinal microbiome in women of reproductive age), and the National Institutes of Health (on biomarkers of iron status in women of reproductive age); all were institutional research grants awarded to Cornell University. JLF declares no known conflicts of interest.

None of the authors are investigators on any trials likely to be included in this review.

*The content is solely the responsibility of the authors and does not necessarily represent the official positions, decisions, policy, or views of the US CDC.

Edited (no change to conclusions)