Background: Investigators at Bandim Health Project (BHP, www.bandim.org) in Guinea-Bissau have shown in several randomized trials that the Bacille-Calmette-Guérin (BCG) vaccine against tuberculosis (TB) is associated with reduced mortality in the first months of life.

BCG is a live attenuated vaccine, which means that it consists of active tuberculosis bacteria that are not capable of infecting a human with TB. BCG has been grown and maintained at many different laboratories all over the world using slightly different laboratory techniques. Due to the accumulation of genetic mutations in the different BCG strains, many variants of the vaccine exists today. These have different properties when it comes to immune response, side effects, protection against TB and scar formation. The BCG scar status after vaccination is a good marker for the non-specific effects of the vaccine; among BCG-vaccinated infants, those with a BCG scar have improved survival. The investigators hypothesize that the different types of BCG vary in terms of the strength of the non-specific effects and thus the impact on overall morbidity and mortality.

In the trial, the investigators will compare the two most widely used BCG strains in the world, BCG-Russia and BCG-Japan, with respect to their non-specific effects on morbidity and mortality. As an addition, the investigators will study the effect of maternal BCG vaccination on the subsequent effect of BCG-vaccination in the offspring, since there are indications that the maternal BCG scar status primes for a stronger non-specific response in the offspring.

In the trial, the investigators will compare the two currently most widely used BCG strains in the world, BCG-Russia and BCG-Japan, with respect to their non-specific effects on morbidity and mortality.

The investigators propose a 3-year randomized controlled trial (RCT) at the maternity ward of the main hospital Simão Mendes (HNSM) in urban Bissau, Guinea-Bissau, comparing 1:1 BCG-Japan vs. BCG-Russia in 15,600 infants with respect to morbidity and mortality. Additionally, the investigators will use this large study to test the role of the mother's BCG scar status for the overall health effect of BCG in her child.

Objective

The investigators aim to investigate the following hypotheses:

1. Compared with BCG-Russia, vaccination with BCG-Japan is associated with fewer admissions and deaths.
2. The effect of BCG in the child is more beneficial if the mother has a BCG scar.

METHODS

Inclusion:

Mothers/guardians to infants eligible for the study (participants) will receive an oral study explanation in Portuguese Creole and a written explanation in Portuguese. Provided oral consent, the mother/guardian signs a written consent form; if the mother or guardian is illiterate, a fingerprint can be provided to confirm participation. Telephone contact information for the mother, the father and up to two family members or persons living in the same house are recorded at inclusion. The BCG vaccine will be provided at discharge for all infants at the ward. A sticker with a unique study number is placed on the vaccination card to indicate eligibility to the study.

Randomization: Upon informed consent, the mother selects, from a stack of envelopes, a closed envelope that contains a sealed randomization lot of either Japan or Russia. Randomization lot envelopes are prepared by a senior staff supervisor in separate bundles of 40 for each sex.

Follow-up:

1. All enrolled participating infants: Family telephone interview at 6 weeks and 6 months.

   Families to all participants are telephoned 6 weeks and 6 months after birth to register dates and outcomes of admissions and whether the participant had died. If the participant died, the investigators will ensure that the family of the participating infant (the mother/guardian) is asked about symptoms and whether the death occurred at home or at a hospital. Information on the 6 week vaccinations, infant BCG scar status and adverse events is also collected. At 6 weeks, information on medical consultations is also collected. Priority will be given to interview the mother of the participating child when possible.

2. Special cohort of BHP study area infants: home visits at 2 and 6 months of age. With the proposed sample size of 15,600 and approximately 17% expected to reside in the BHP Health Demographic Surveillance System (HDSS) study area, the investigators will include approximately 2,650 children from the HDSS. These children will be followed within the routine Health Demographic Surveillance System, and the participants residing within the HDSS will additionally receive extra home visits at 2 and 6 months of age, where additional information on mortality, morbidity, BCG scar, maternal and paternal BCG scar and adverse events will be collected.

3. All participating infants: Registration of admissions and consultations at the HNSM pediatric ward.

   Admissions, diagnoses and outcomes at the pediatric ward are documented by a BHP supervisor all days of the week. For approximately 60% of all infant admissions occurring at the ward, data from the child's health card is documented during the admission, thus registering the admission with the child's unique study identification number (ID) and other information transferred from the card. The BCG scar status will be registered and the size will be measured upon admission to the ward. To identify study admissions not registered with the unique study ID, a standardized systematic database linkage and search algorithm is applied. Data on consultations at HDSS health centers is collected as part of the routine BHP surveillance.

4. All participating infants: Mortality data. Data on mortality will be collected from all available information sources (admission at HNSM, telephone follow-up, HDSS data) and any discrepancies in death information will be resolved by an extra telephone interview, when applicable. For infant deaths within participating families residing in the HDSS study area, a standard verbal autopsy is performed approx. 3 months after the death.

Adverse events: To register adverse events from BCG vaccination in the form of suppurative lymphadenitis, mothers are given a detailed explanation of suppurative lymphadenitis and they are encouraged at inclusion and at subsequent follow-up visits to bring their child to a HDSS health center for consultation if such a condition should arise. Furthermore, participating mothers are asked on the phone whether their child has or have had a swollen lymph node in the left armpit and if yes, whether there has been pus secretion. At HDSS study area home visits, participating mothers are equally asked and the size of the axillary lymph node is examined. A size <16 mm as judged by the locally trained field assistants will be considered normal.

Sample size:

6-7,000 infants are born each year at the maternity ward of the HNSM. With an expected monthly inclusion rate of 450 infants during a 3-year study period, the investigators expect to be able to include at least 15,600 children, i.e. 7,800 in each BCG strain group.

Primary outcome: The investigators expect that around 4.5% will be hospitalized at HNSM within the first 6 weeks of life. Thus, to detect if one of the strains is associated with a 20% lower admission rate with 80% power, an alpha of 0.05 and 10% loss to follow-up, 631 events would be necessary, which is obtained with a sample size of 15,600 infants.

Secondary outcomes: The investigators expect that at least 1.5% will die within the first 6 months of life. With a sample size of 15,600 infants, a 6 month mortality of 1.5% and 15% loss to follow-up, the investigators will be able to detect if one of the strains is associated with a 33% lower mortality rate with 80% power and an alpha of 0.05 (based on 196 events).

In a previous study with <7,000 study participants done by the same research group, also studying the importance of maternal BCG scar, the investigators found a significant beneficial effect of BCG in mothers who had a scar and a significant interaction between maternal and child BCG scar; hence, though it is the effect of BCG vaccination to the child and not BCG scar in the child that is evaluated in the present study, the sample size should be sufficient to show clinically relevant differences in BCG vaccination effects in the child, depending on whether the mother has a BCG scar or not. In addition, the investigators tentatively collect data on paternal BCG scar at inclusion and zone follow-up visits, if the father is present.

Analyses:

The hospitalization and mortality data will be analyzed as intention-to-treat in Cox regression models with age as the underlying time variable, i.e. with complete adjustment for age. In case an Oral Polio Vaccine (OPV) or vitamin A supplementation campaign or similar campaigns with potential immune stimulatory effects occurs during the study period, the main comparison of the two strains will be conducted by censoring all children on the first day of the campaign, to exclude any interaction of the campaign with the BCG strains.

As a sensitivity analysis, an analysis where same-day deaths and admissions (events occurring on the day of BCG vaccination) are omitted will be conducted. In-hospital case-fatality rates will be compared between strains in regression models.

All analyses will be done overall and stratified by maternal BCG scar. Furthermore, the investigators will study the independent effect of maternal BCG scar on trial outcomes. The analysis will adjust for type of BCG vaccine that the child received and for possible determinants of maternal BCG scar and morbidity/mortality, e.g. age, maternal mid-upper arm circumference and ethnic group.

Trial vaccines:

BCG will be acquired from the Japan BCG laboratory (BCG-Japan) and the Serum Institute of India (BCG-Russia).

Possible difficulties in procuring BCG vaccines: For a 3-year trial utilizing two specific BCG vaccine strains, a backup for the proposed study vaccine strains is necessary as a strain could become inaccessible due to production halts, worldwide shortages or logistic difficulties in delivering the vaccines to Bissau. In such cases, the investigators proposes to substitute study vaccines as follows:

1. BCG-Russia substituted with either of the following, both derived from BCG-Russia:

   • BCG-Russia (Russian BCG-I strain, Microgen, Russia)
   • BCG-Bulgaria (BCG-SL 222 Sofia strain, InterVax Ltd., Canada/BulBIO LTD., Bulgaria).

2. BCG-Japan substituted with either of the following, both derived from BCG-Japan:

   • BCG-Taiwan (Tokyo 172 strain, Taiwan Center for Disease Control)
   • BCG-Thailand (Tokyo 172 strain, Queen Saovabha Memorial Institute).

Data analysis if BCG strains have to be substituted:

Should it be necessary to substitute study BCG vaccines, study outcomes will be presented both by pooling genetically similar strains (BCG-Russia+BCG-Bulgaria vs. BCG-Japan+BCG-Taiwan/BCG-Thailand) and as a separate sub analysis, pending approval and recommendations by the Data Safety Monitoring Board (DSMB).

Ethical considerations:

The proposed study will randomize children to two BCG strains that are widely used in Guinea-Bissau and the rest of the world. Previous studies have shown that providing BCG vaccination at discharge is safe and beneficial. Oral and written informed consent will be obtained. The BHP offers free-of-charge medical consultations and some essential medications at three health centers in the HDSS to all infants assessed for eligibility, independent of the choice to participate or not. Participating mothers are encouraged to bring their child for consultation at one of the health centers within the HDSS if their child develops suppurative lymphadenitis.

The study has been approved by the Ethical Committee in Guinea-Bissau and the Central Ethical Committee in Denmark has given its consultative approval.

Inclusion criteria:

Newborn, unvaccinated infants identified at the maternity ward and ready to be discharged

Exclusion criteria:

Infants above 6 weeks (>42 days) of age; Infants with gross malformation; Infants that are acutely ill

Non-identifiable individual data can be shared on the basis of a data sharing proposal sent to cbenn@health.sdu.dk

When follow-up has been completed and the dataset have been closed (e.g. by november or december 2020)

Non-identifiable individual data can be shared on the basis of a data sharing proposal sent to cbenn@health.sdu.dk