Trial to Compare Two Strains of BCG

This two-year outcome assessor-blinded RCT will be conducted at the maternity ward of hospital Simão Mendes (HNSM) in urban Bissau, Guinea-Bissau to compare BCG-Japan versus BCG-Russia 1:1 in 15,000 infants with respect to mortality, morbidity and case-fatality rate during hospital admission.

The trial will also examine the association between BCG strains and BCG skin reaction characteristics by six weeks (data collected by telephone) and at two and six months (data collected at home-visits to a subgroup of the cohort).

HYPOTHESES

The aim is to investigate the following hypotheses:

• Compared with BCG-Russia, receiving BCG-Japan is associated with

• a 16% reduction in all-cause deaths and
• a 10% lower case-fatality rate for hospitalized infants.
• BCG-vaccinated children have lower mortality if the mother has a BCG scar, when compared to if the mother does not have a BCG scar.

METHODS Setting: The RCT will be carried out by the Bandim Health Project (BHP) in a close collaboration with the HNSM Maternity Ward. BHP maintains a Health and Demographic Surveillance System (HDSS) site in Guinea-Bissau, covering approx. 100,000 individuals in six suburbs of the capital Bissau. A dedicated BHP team registers all births and vaccinations at the Maternity Ward, where BHP has conducted a series of RCTs since 2002, with the aim of improving early-life health outcomes.

Inclusion: Neonates born at the HNSM Maternity Ward and neonates referred to the ward for vaccination are eligible for participation in the study.

Mothers/guardians to infants eligible for the study will receive an oral study explanation in Portuguese Creole and a written explanation in Portuguese. Provided that oral consent is obtained, the mother/guardian signs a written consent form; if the mother or guardian is illiterate, a fingerprint can be provided to confirm participation. The family can request that their child leaves the trial at any time. Infants that are not eligible for participation or whose mother/guardian declines participation will be registered and vaccinated by our team (standard practice). Information on maternal and paternal BCG scar status, scar size, mid-upper-arm circumference and socioeconomic factors will be collected during the inclusion procedure.

Telephone contact information for the mother, the father and family members and/or persons living in the same house are recorded at inclusion. BCG will be provided at discharge for all infants at the ward.

Randomization:

Following informed consent, the mother selects, from a stack of envelopes, a closed envelope that contains a sealed randomization lot indicating allocation to either BCG-Japan or BCG-Russia. The mother, inclusion assistant and vaccinator will thus not be blinded to the intervention allocation.

Vaccination: The infant is vaccinated intradermally with 0.05 ml of the allocated BCG strain in the left upper deltoid, followed by vaccination with OPV. OPV is provided via the National Vaccination Program; if OPV is in shortage, only BCG vaccination will be provided.

Follow-up:

All assistants assessing outcomes during the follow-up procedures outlined below will be blinded to the randomization allocation.

Follow-up takes place through three mechanisms:

• All enrolled infants with a telephone number recorded at inclusion: Family telephone interview at 6 weeks and 6 months to register dates and outcomes of consultations, hospital admissions and whether the child died. If the child died, the mother/guardian is briefly asked about symptoms and whether the death occurred at home or at a hospital. Information on adherence to the 6-week vaccination schedule, infant BCG reaction status and adverse events is also collected. If the infant has not yet received the 6-week vaccines, the mother/guardian will be reminded that it is time for the infant to be vaccinated.
• Cohort of BHP HDSS infants: home visits at 2 and 6 months of age. With the proposed sample size of 15,000 infants and an estimated 15.5% residing in the HDSS study area, the trial will enroll approximately 2,300 children from the HDSS. These children will be followed by our routine surveillance system and receive two additional home visits at 2- and 6 months of age. At the visits, data is collected on mortality, morbidity, BCG scar status and size, adverse events and maternal and paternal BCG scar status (if not collected at inclusion).
• All enrolled infants: Registration of admissions and consultations at the HNSM pediatric ward.

Admissions, diagnoses and outcomes at the pediatric ward are documented by a BHP team on all days of the year. Parental names and telephone numbers are registered for all admissions.

Sample size:

Primary outcome: Based on BCGSTRAIN I trial data and BCGSTRAIN II trial data (unpublished), an overall mortality rate of 1.1% by 6 weeks of age is anticipated. Given that two large-scale RCTs testing BCG-Japan vs. BCG-Russia have been conducted in Guinea-Bissau, the sample size needed to demonstrate a significant difference in all-cause mortality between BCG-Japan and BCG-Russia has been calculated based on the conditional power of a meta-analysis involving BCGSTRAIN I, II and a third RCT, as detailed by Roloff et al. In order to detect a 16% reduction in all-cause mortality associated with BCG-Japan in the meta-analysis of the three RCTs, with an expected heterogeneity of 0.002 between the trials, a conditional power of 0.80 and an alpha of 0.05, an additional 148 events in the third trial is necessary. With an expected mortality rate of 1.1%, this corresponds to a sample size of 15,000 inclusions when considering an expected loss to follow-up of approx. 10%.

With an expected monthly inclusion rate of 600 infants based on previous experiences, it is expected that at least 15,000 children, i.e. 7,500 in each BCG strain group, can be included during an expected timeframe of approx. 2 years for inclusion procedures and an additional 6 months for follow-up procedures.

Analyses:

The mortality and morbidity data will be analyzed as intention-to-treat in Cox regression models with age as the underlying time variable. In-hospital case-fatality rates between BCG strains will be compared using Fischer's exact test (2-sided). In case an 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 and of the effects of maternal BCG scarring on infant outcomes 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 sensitivity analyses, an analysis where same-day deaths and admissions (events occurring on the day of BCG vaccination) are omitted will be conducted. An analysis of main outcomes excluding neonates that were admitted to intensive care before inclusion will also be conducted.

All analyses will be conducted overall and stratified by maternal BCG scar status and sex.

Trial vaccines:

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

Ethical considerations:

The proposed study will randomize children to BCG strains that are distributed by UNICEF and widely used in Guinea-Bissau and the rest of the world. Our previous studies have shown that providing BCG vaccination at discharge is safe and beneficial. Oral and written informed consent will be obtained in all cases. The study protocol was approved by the Ethical Committee in Guinea-Bissau and the Central Ethical Committee in Denmark gave its consultative approval. A local clinical monitor will be appointed to oversee the study.

Public health importance:

Global annual infant BCG vaccinations exceed 120 million. Current BCG vaccine strains are heterogeneous due to accumulation of genetic diversity and non-standardized production techniques. Yet, there is a substantial lack of data comparing the various BCG strains both in terms of specific and non-specific effects.

By providing information regarding BCG strains and data on the importance of maternal immune priming with BCG, the proposed study is likely to be an important contribution to future BCG policies. Substantial improvements in protection from TB and reductions in early-life morbidity and mortality will likely be achieved at low cost, if more information on the efficacy of the different strains of BCG and the importance of maternal immune priming is made available.

Inclusion criteria

• Healthy infants present at the HNSM maternity ward on the day of discharge or procuring vaccination at the ward.

Exclusion criteria

• Infants older than 6 weeks (>42 days) of age
• Infants that have already received either BCG or oral polio vaccine (OPV)
• Infants that are moribund due to gross malformation or acute illness (about to die or be transferred to the pediatric ward according to the local health physician assessment; the latter children may be recruited when they are discharged from the pediatric ward or the neonatal intensive care unit).