Study setting

The study site is located in Bhaktapur, 15 km east of the capital city Kathmandu. Bhaktapur city is an ancient city famous for its traditional temples and buildings. It is listed under Unesco world heritage and one of the main tourist attractions in Nepal. Bhaktapur is a homogenous community consisting mostly of practising mixed Hindus and Buddhists. It is a periurban agricultural-based community located 1400 m above sea level with a population predominated by the Newar ethnic group. In the Bhaktapur municipality, domestic migrant workers from diverse ethnic groups work in several carpet factories. The local climate is humid subtropical with a wet and hot season (monsoon) from May to August, and a dry and cool season from October to March. The annual average rainfall is 78.3 mm, and the temperature ranges from −2°C to 35°C. The majority of the residents in Bhaktapur consume foods grown in the community. Rice is the staple food. The eating pattern varies with season, workload in the field and availability of foods. A variety of local green leafy vegetables is widely consumed mainly in the winter and spring seasons.

Eligibility criteria

1. 20 to 40-year-old pregnant women from early pregnancy, not later than 15 weeks pregnant;

2. residing in Bhaktapur municipality and surrounding areas in Bhaktapur district;

3. availability of informed consent.

Exclusion criteria

1. taking dietary or multivitamin supplements containing vitamin B₁₂;

2. known cases of chronic disease under treatment such as tuberculosis, diabetes, hypertension, hypo or hyperthyroidism, pernicious anaemia, Crohn’s disease and current use of anticonvulsant drugs;

3. severe anaemia (haemoglobin concentration <7 g/dL); blood will be analysed for haemoglobin concentration immediately after collection at enrolment by using HemoCue;

4. suffering from any condition that requires treatment with vitamin B₁₂ such as pernicious anaemia and strict vegans.

Intervention

The supplements are produced by GC Rieber Compact (Bergen, Norway). The vitamin and placebo tablets are, except for the cyanocobalamin content, identical in composition, taste, smell and appearance. The compositions of the tablets follow the recipe of the commercial product ‘Seven OceanS Emergency Ration.’ Each daily dose (placebo or 50 µg vitamin B₁₂) also contains 31 kcal from the 6.95 g vehicle which mainly consists of dextrose and palm oil. First dose of supplementation will be given by research staff at the hospital, and mothers will then be instructed to properly store the supplements, and consume these every day preferably in the morning. During the weekly visits, our research staff will record consumption of supplement during the last 7 days including iron, folic acid and calcium, and also count the amount of remaining supplements. We will also record if there was vomiting after supplementation, and reasons why the supplements were not taken. During the antenatal visits or other visits at the hospital, the study gynaecologists will verify the compliance. Additionally, study supervisors will randomly select days (5% of total days) to make independent visits to participant’s mother's home in order to record compliance with supplementation.

Cointervention

All pregnant women will also be given folic acid (0.4 mg) for the first 2 months of pregnancy followed by iron (60 mg elemental iron) and calcium supplements (500 mg) until 45 days after delivery according to the WHO guidelines.30

Recruitment procedure, confirmation of pregnancy and assessment of gestational age

Prior and throughout the study period, we will identify pregnant mothers through a hospital and community-based surveillance system (figure 1). We have records of newly married couples in the study area that are updated quarterly. Information regarding pregnancies will be obtained monthly by household surveys or phone calls. Women with history of interrupted menstruation cycle or pregnancy detected by standard human chorionic gonadotropin kits, are screened in the hospital. In these women, pregnancy and gestational age is estimated by ultrasonography (USG). Follow-up USG is performed between 18 and 22 weeks for anomalies scanning.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Study flow chart—recruitment, supplementation and follow-up plan. BSID, Bayley Scales of Infant Development.

After screening for eligibility, a detailed information regarding the study will be provided focusing on the duration, collection of biological samples and the follow-up plan of the study. The consent forms will be filled up by study gynaecologists or supervisors, preferably in the presence of the husband. A copy of the consent form along with the information sheets will be provided to each participant. In the consent form, there is a separate provision for consent of biological samples where the participants may agree or refuse. In case of illiterate participants, we will obtain thumbprints after getting signature from an impartial witness who is not part of research team.

Allocation, randomisation and blinding

The randomised allocation list was generated in STATA (Stata, College Station, TX) using the script ‘randomize’ (net from http://folk.uib.no/mihtr/stata/). We randomised the women in a 1:1 ratio in blocks of 8. Each woman is assigned a packet of supplementation according to her ID number which is sequentially numbered. This packet is only labelled with general information about the study and the unique ID number. The list that links the ID number to the randomisation code is kept with the company that produces the intervention and the placebo, and with the scientist who generated it who is otherwise not involved in the study. We will ensure allocation concealment throughout the study as none of the investigators will have access to this list until completion of data collection. This is only after the database with the main outcomes (growth and neurodevelopment scores) has been cleaned and locked, and after the plan of analysis for the main outcomes is finalised and approved by all involved scientists.

Rationale for choosing 50 µg dose of vitamin B₁₂

Based on results from previous studies in this area, we expect the intake of vitamin B₁₂ among women in our study to be low, both due to a predominant plant-based diet and poor gut function that again may disturb vitamin B₁₂ absorption and increase dose requirements. Absorption of vitamin B₁₂ is strictly regulated, and the body is not able to absorb more than 2–4 µg/day. The vitamin B₁₂ supplementation study in pregnancy from South India24 and the study from Bangladesh26 used very high doses of 20–100× RDA with no reported adverse effects. There are other large RCTs where similarly high doses of vitamin B₁₂ have been given together with other micronutrients.31 32 These studies found that the micronutrient mixture that included high doses of vitamin B₁₂ reduced the risk for small for gestational age (SGA) with no adverse effects. In Europe and the USA, vegans and vegetarians are aware that there is a risk of B₁₂ deficiency that can result in poor neurodevelopment and other adverse outcomes.33 Many therefore take additional supplements with similar high doses of B₁₂ during pregnancy.

Timeline

The first participant in the study was enrolled in March 2017 and we plan to enrol for 2 years. End of follow-up of infants (for collecting primary outcomes) will be completed approximately 18 months after the last enrolment.

Laboratory procedures

From the mother, blood samples will be obtained at enrolment (baseline) and at the end of supplementation (6 months postpartum). In the infants, we will collect blood samples at 6 months by a trained lab technician with direct supervision of study paediatrician. Three millilitres of blood will be collected into vials containing EDTA as anticoagulant. The plasma will be centrifuged at approximately 700 g at room temperature for 10 min, separated and transferred into storage vials, and stored at −70° before analysis.

The blood samples will be analysed for the following:

1. Haemoglobin concentration by using HemoCue B immediately after blood collection (HemoCue, Vedbæk, Denmark).

2. Plasma vitamin B₁₂ and plasma folate concentrations will be estimated by microbiological assays using a chloramphenicol-resistant strain of Lactobacillus casei and colistin sulfate-resistant strain of Lactobacillus leichmannii, respectively.

3. The plasma will be stored and used for other biomarkers (nutrients, markers of inflammation, epigenetic markers and markers related to neurodevelopment).

Follow-up

We keep track of the enrolled women in a Global Positioning System (GPS)/mobile-based surveillance system. Enrolled women will be visited every week in the period of supplementation and the fieldworker will ask questions about their physical condition, compliance among both study groups and adverse events (AE). The family/women are asked to inform us when they go into labour, and we will ensure transportation to hospital where they will give birth. This will also ensure that we get information about the pregnancy outcome. We will not try to influence the choice of place to deliver; however, we will encourage the women to deliver in a health facility rather than at home. In the Kathmandu Valley, less than 5% of deliveries are at home.

After delivery, we will follow-up the child at home every month until 2 years of age. We will collect information on breast feeding and complementary feeding (frequency, whether other drinks or solids/semisolids have been given). We will also ensure that the child receives standard vaccines as per the expanded programme on immunization schedule. Growth will be measured and dietary recalls undertaken every 3 months. Neurodevelopment will be measured at different time points during infancy.

Outcomes

Quality control of field activities

The fieldwork will be monitored by skilled supervisors, medical doctors, gynaecologists and qualified psychologists. For all outcomes, the study staff have received training before initiation of the project, and the staff will be used based on their skill level. For a specified proportion (5%) of the primary outcomes, a study supervisor or scientist monitors the worker's performance. Some of these assessments are also done independently. For example, when a fieldworker has measured the weight and length of a child, a supervisor or scientist will undertake the same measurement and compare their findings with that of the fieldworker. For all main outcomes, we compare the reliability and precision of each examiner with the reliability of a gold standard. For example, for measuring length we have training sessions where the fieldworkers and a gold standard measure each child twice. In a typical training session, each worker measures 10 children twice. The intraindividual variability (mean squared difference) of a trainee will be compared with that of the gold standard. For length measurements, we will accept no more than 50% higher intraindividual variability of the examiner compared with the gold standard. In this exercise, we can also compare the mean deviations of absolute differences between the gold standard and the worker. The mean deviation is also compared with the mean variability of the measurements of the gold standard. The deviation is also compared on a qualitative scale to assess whether there are systematic higher or lower readings. In case of poor agreement, extra training will be carried out until acceptable agreement was achieved.

Quality control for the neurodevelopmental tools

All methods that will be used to measure neurodevelopment have been translated, adjusted and/or piloted in previous studies at the current study site. The comprehensive neurodevelopmental tools such as the Bayley-III and TIMP will be administered by the medical doctors and psychologists. Medical doctors and qualified psychologists will train and monitor fieldworkers to perform the neurodevelopmental assessments such as the ASQ-3, HOME assessment and the various questionnaires. Ahead of study start, we will arrange training sessions. The trainings will be supervised and monitored by experienced psychologists from the Regional Centre for Child and Adolescent Mental Health and Welfare, Uni Research Health, Bergen, Norway. Standardisation exercises will be conducted measuring the inter-rater reliability of the assessor compared with a gold standard until satisfactory level of agreement. Video recording will be done for all Bayley assessments and 10% of the videos will be scored independently for quality control. For all other instruments, 5% of the assessments will be double scored. Once a month inter-rater reliability will be analysed for quality control.

Safety considerations, safety monitoring and AE reporting

All AEs and serious adverse events (SAE) will be recorded by the study staff. AEs are graded from 1 to 5 according to their severity according to Common Terminology Criteria for Adverse Events.45 All immediate AEs following each dose of intervention will be documented for all women.

All AEs will be followed up until resolution or stabilisation as judged by gynaecologist or paediatrician (site investigator) and the principal investigator. All SAEs will be followed up until satisfactory resolution or until the treating physicians and the principal investigator deem the event to be chronic or the participant to be stable.

SAE, which includes critical or life-threatening illness or death, will be documented from the time of enrolment throughout the study period. The pregnant women will be visited every week. During each contact, the study staff will collect data to ascertain SAEs and illness requiring hospitalisation. All SAEs of death, development of signs of critical illness and severe illness requiring hospitalisation, and AEs as described previously will be reported to the Ethics Committees, the Data and Safety Monitoring Board (DSMB) and to the sponsors of the study within 24 hours of awareness of the event, followed by a final report within 10 days. SAE’s relatedness to the administration of vitamin B₁₂ or placebo will be judged by the investigator/designee, the Ethics Committee and the DSMB who will have access to all relevant investigations, clinical assessments and if require unblinding allocated interventions.

Data and Safety Monitoring Boards

The DSMB comprises a paediatrician, public health expert and a biostatistician. The DSMB is independent from the sponsor and has no competing interests. The DSMB members have prepared a charter and decided the study stopping rules, and will review the SAEs and AEs reported in the study periodically. They will examine all pregnancy complications, cases of severe illness, infant deaths and other SAEs to decide if the study should be continued, based on the predecided stopping rules. After one-third of the study participants have been enrolled and completed follow-up, the DSMBs will review the data and make recommendations concerning continuation, modification or termination of the study due to unexpectedly large beneficial effects or serious side effects and can suggest extension of the trial should the primary outcomes occur less often than anticipated. Any important protocol modification such as changes to eligibility criteria, outcome analyses will be thoroughly communicated with DSMB members and will obtain amendment clearance from the institutional review board (IRB).

Sample size calculations

We expect that as many as 15% of the enrolled women will be lost to follow-up because of abortions, stillbirths and migration (see table 1).

We have calculated the sample sizes based on standard formulas46 to compare two means which are used in the ‘power’ function in STATA (Stata). For an effect size of .25 SD, we would need to analyse 676 infants to achieve a statistical power of 90% (p values of 0.05); we will therefore target 800 women–infant dyads (676/(1–0.15)). In these calculations, we assumed equal SD in the placebo and vitamin B₁₂ groups. The power to detect differences between the study groups is larger than what is projected here as we will measure most of the outcomes more than once and also use these repeated measurements in the same analyses, adjusting for interdependence within a child using mixed effects models, generalised estimating equations or other appropriate methods. Various required total sample sizes according to meaningful differences and powers of 80% and 90% are depicted in figure 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Estimated required total sample sizes based on relevant effect sizes at 80% and 90% power.

These figures are based on national figures and our experiences from the field site. Infant morbidity and mortality are expected to be lower than national figures because of the proximity to the capital Kathmandu and because of our close follow-up.

Data management

All information will be collected in structured forms designed specifically for this project. A relational database has been designed. All forms are checked manually by the study supervisors before they are sent for computer entry. All data are entered twice by two different data operators within 1 week of data collection.

Data are stored in secure servers in the field office and in a data management centre (DMC) in Kathmandu. A substantial proportion of the forms is captured using iPads or similar tablets using the cloud-based data management system. We have used this electronic data capture system with success in many previous studies. Only computer data entry staff and supervisors will have access to the data. In the data system, the entered data are processed in several steps and the system will be designed to identify inconsistencies between different variables and different forms. The system can also detect entries that are out of range and notify when information is missing. This system also includes reminders about study activities, that is, the study staff will be notified when a birth or a scheduled vaccine is due. We are collecting GPS coordinates and phone numbers from several family members. By this, we can more easily reach the families and will probably reduce the numbers of missed home visits.

Data tracking, cleaning and quality checks

The DMC is also responsible for initial cleaning of the data. Interim tabulations and scatter plots for some variables will be made at regular intervals to identify data errors. Special checks are made on observations that are more than two or three SDs from the mean. There is a regular feedback of errors from the DMC to the study sites.

Record retention and archival

All the study documents including participant’s source data and documents will be archived by the study sites after the completion of the study, until the time the sponsor informs in writing to the study sites that they no longer need to maintain the study documents.

Plan of analysis

All analyses will initially be done on an intent-to-treat basis. All randomised participants will be included in the analyses if the relevant outcome variables have been collected. The main outcomes are continuous and expected to be normally distributed. We will use the Bayley-III scores at 6 and 12 months in separate analyses where B₁₂ supplementation is the main exposure. We will compare the mean Bayley-III scores (scores on the cognitive, language and motor subscale scores, with the language scale analysed both separately for receptive and expressive language and as a composite measure, and the motor scale analysed both separately for fine and gross motor development and as a composite measure) between the vitamin B₁₂ group and the placebo group. For growth, the main outcome will be measured when the child is 12 months. Linear growth, expressed as centimetres and height for age z-score (HAZ), and weight expressed as kilograms, weight for age (WAZ) and weight for height (WHZ) will be compared between the study groups. If baseline differences are observed, we will adjust these imbalances in multiple linear regression models, adjusted and unadjusted effect estimates will then be presented.

We will analyse the effect of vitamin B₁₂ separately in the following subgroups:

1. Vitamin B₁₂ status:

   • based on cobalamin (cut-off: 150 pmol/L)

   • based on low total homocysteine (cut-off: 10 µmol/L)

   • based on low methylmalonic acid (cut-off: 0.26 µmol/L)

2. Maternal body mass index (cut-off: 18.5)

3. Vegetarian (yes/no)

4. Birth weight (cut-off: 2500 g)

In these subgroup analyses, the differences in the effect between the levels of the subgroups will be measured by including two-way interaction terms in multiple regression models. Statistically significant effect modification will be when the p value of the interaction term is <0.05.

Per protocol analysis

In addition to a standard per protocol analysis, we will use instrumental variable analysis in an attempt to estimate the true effect of cobalamin had it been given to all women in the scheduled doses and intervals. The random allocation will be the instrument in these analyses. For per protocol analysis, women who receive less than 50% of the projected doses during pregnancy will not be included in the analyses, well acknowledging that the ensuing effect estimates may be biased and will certainly represent an effect higher than what can be achieved even in our well-resourced study setting.