Nankabirwa et al.

Trials (2017) 18:152

DOI 10.1186/s13063-017-1881-z

STUDY PROTOCOL Open Access

Early versus late BCG vaccination in HIV-1-

exposed infants in Uganda: study protocol
for a randomized controlled trial
Victoria Nankabirwa1,2* , James K. Tumwine3, Olive Namugga2, Thorkild Tylleskär2, Grace Ndeezi3,
Bjarne Robberstad2, Mihai G. Netea4 and Halvor Sommerfelt2,5

Abstract
Background: Bacillus Calmette-Guérin (BCG) vaccination may have nonspecific effects, i.e., effects on childhood
morbidity and mortality that go beyond its effect on the risk of childhood tuberculosis (TB). Though the available
scientific literature is mostly from observational studies, and is fraught with controversy, BCG vaccination at birth
may protect infants in high-mortality populations against serious infections other than TB. Yet, other studies
indicate that giving BCG later in infancy may modify immune responses to non-TB antigens and potentially
enhance immunity, potentially also against tuberculosis (TB). It is unclear whether BCG vaccination very early in life
offers adequate protection against TB and other infections among HIV-1-exposed children because even those who
remain uninfected with HIV-1 show signs of impaired immunocompetence early in infancy. This study will compare
BCG vaccination at birth with BCG vaccination at 14 weeks of age in HIV-1-exposed infants.
Methods: This is an individually randomized controlled trial in 2200 HIV-1-exposed infants. The intervention is BCG
vaccination within 24 h of birth while the comparator is BCG given at 14 weeks of age. The study co-primary
outcomes are severe illness in the first 14 weeks of life, and production of tumor necrosis factor, interleukin (IL)-1β,
IL-6 and interferon-γ in response to mycobacterial and nonmycobacterial antigens. The study is being conducted in
three health centers in Uganda.
Discussion: A well-timed BCG vaccination could have important nonspecific effects in HIV-1-exposed infants. This
trial could inform the development of appropriate timing of BCG vaccination for HIV-1-exposed infants.
Trial registration: ClinicalTrials.gov, identifier: NCT02606526. Registered on 12 November 2015.
Keywords: BCG, Vaccination, Infants HIV-1-exposed, Nonspecific effects, Trial

Background tuberculosis (TB) [7–11]. A case control study among

BCG vaccination and nonspecific effects infants in Guinea-Bissau showed that infants who had
Several observational studies suggest that routine child- received the BCG vaccine were at least 60% less likely to
hood vaccines may have effects on childhood morbidity have acute lower respiratory tract infections (ALRI) and
and mortality that are separate from their effects on the respiratory syncytial virus infections [12] than those who
incidence of the diseases that they target [1–6]. Import- had not been vaccinated. A small Malawian study indi-
antly, the Bacillus Calmette-Guérin (BCG) vaccine may cated that the likelihood of septicemia among infants
have effects on morbidity and mortality that cannot be with a BCG scar was lower than that among infants
attributed to the vaccine’s protective effect against without a scar [13]. A Senegalese prospective study ana-
lyzing data from two birth cohorts (a total of 11,369
* Correspondence: [email protected]
1
children under 2 years of age) found an association
Department of Epidemiology and Biostatics, School of Public Health,
College of Health Sciences, Makerere University, Kampala, Uganda
between a combination of BCG/DTP administered
2
Centre for Intervention Science in Maternal and Child Health, Centre for simultaneously and reduced mortality [1]. A secondary
International Health, University of Bergenhttp://www.cismac.org analysis of data from a randomized controlled trial
Full list of author information is available at the end of the article

© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Nankabirwa et al. Trials (2017) 18:152 Page 2 of 17

among low-birth-weight preterm infants indicated that currently no other, long-term, ongoing randomized trials
there may be an association between early BCG vaccin- with clinical endpoints similar to ours.
ation and lower perinatal mortality, but not lower infant
mortality [9]. In a birth cohort study in Uganda, we pre- Immunological reprogramming induced by BCG
viously showed that children between 1 month and In order to strengthen the medical literature, one of the
5 years of age who had received BCG had a substantially most important aims of the study is to identify any im-
lower mortality than those who had not [14]. In fact, this munological reprogramming induced by BCG, which
association was even stronger in the neonates but we has been recently suggested to be responsible for NSEs
decided not to display this potentially important finding [30]. The immunological responses thought to account
because of the possibility that some of the association for the NSEs of BCG may, in broad terms, be considered
could have been due to reverse causality, as vaccinators to involve two processes: trained innate immunity and
could to some extent have refrained from giving BCG to heterologous immunity. Trained immunity is a novel
sick babies. Other observational studies among older concept that describes the functional reprogramming of
children, adolescents and adults have shown mixed innate immune cells (such as monocytes, macrophages
results for the effects of the BCG vaccine against atopy or natural killer (NK) cells) after an infection or vaccin-
[15–18] and helminthic infections [19–21]. Nonspecific ation [31]. The molecular mechanism mediating trained
effects of vaccines, are now receiving increased atten- immunity is represented by epigenetic modifications due
tion, all the more so when the incidence of the TB is de- to changes in histone methylation and acetylation, lead-
creasing worldwide [22]. In fact, the World Health ing to a looser chromatin structure and increased gene
Organization (WHO) recently established a Strategic transcription [32]. Similar epigenetic reprogramming has
Advisory Group of Experts working group on NSEs of been described in monocytes of individuals vaccinated
vaccines [23]. with BCG [33]. In addition to trained immunity, a sec-
ond important immunological mechanism that is likely
Methodological challenges in existing studies to be involved in the NSEs is heterologous immunity:
The medical literature on the NSEs of vaccines is the capacity of memory T-lymphocytes to respond in an
fraught with controversy [7]. Much of this controversy interleukin (IL)-12/IL-18-dependent manner with in-
stems from interpretational challenges of the observa- creased production of cytokines upon stimulation with
tional studies carried out in high-mortality areas, and in different stimuli [34].
heavily burdened health care systems [24, 25]. It has
been argued that children with a lower morbidity and Timing of BCG vaccination among HIV-1-exposed children
mortality are more likely to be vaccinated than higher- Based on studies showing a nearly 1000-fold increased risk
risk children, resulting in a spurious nonspecific protect- of disseminated BCG infection among HIV-1-infected in-
ive effect of vaccines in observational studies [24, 25]. fants, recent WHO guidelines consider HIV-1 infection a
There is also the possibility of survival bias if death (a contraindication to BCG vaccination [22, 35, 36]. How-
common study outcome) is associated with missing ever, the practical implementation of this guideline has
vaccination cards. This has been observed in some retro- been difficult as HIV-1 diagnosis at birth (when the vac-
spective studies in Africa where child deaths were cine is to be given) is not available in many low- and
quickly followed by destruction or loss of vaccination middle-income countries (LMICs) [35, 37]. As a result, all
cards [1, 24, 26, 27]. These selection challenges can be HIV-1-exposed infants, including those yet to be diag-
alleviated by well-designed and well-executed random- nosed as HIV-1 infected, continue to receive BCG at birth
ized controlled trials. However, as most childhood in several LMICs [35, 37]. To circumvent this, an alterna-
vaccines are part of routine national and international tive approach has been proposed in which the vaccine is
child health programmes, it has been considered uneth- delayed until these infants are diagnosed as not having
ical to randomize children to not receive BCG. For this HIV-1 infection [35, 37].
reason, the medical literature on NSEs has continued to On the one hand, if vaccinating HIV-1-exposed babies
be overshadowed by controversy and continues to be with BCG at birth protects them against serious infec-
dependent on initial studies carried out several decades tions other than TB, i.e., through NSEs [7, 8, 11], delay-
ago when morbidity and mortality patterns, as well as ing vaccination could result in increased morbidity and
the vaccines themselves, were different from what they even mortality. In addition, delaying BCG vaccination to
are now [24, 25]. Two recent systematic reviews con- a later time when a definitive diagnosis is made could
cluded that the currently available evidence, which is reduce its coverage [37]. The Ugandan Ministry of
mainly from observational studies, was of low quality Health has decided that all infants, including those with
and reiterated the need for randomized controlled trials HIV-1 infection, but without symptomatic HIV-1
[28, 29]. To the best of our knowledge, there are disease, are to receive BCG.
Nankabirwa et al. Trials (2017) 18:152 Page 3 of 17

On the other hand, recent studies put into question reported robust T-cellular responses to Bordetella per-
the age at which the BCG vaccine is best given to HIV- tussis and tetanus toxoid in infants in whom BCG
1-exposed and uninfected children [35, 38–41]. It is vaccination was delayed to 8 weeks of age [49]. It is
presently unclear whether BCG vaccination very early in unknown, whether these effects of delayed BCG
life provides them with adequate protection against TB administration translate into enhanced protection
and other infections. HIV-1 infection during pregnancy against TB or, for that matter, against other serious
may transiently modulate their responses to the BCG infections.
vaccine [38]. It appears that they have an impaired The most appropriate timing for BCG vaccination that
response to purified protein derivative (PPD) and heat- maximizes both specific and possible NSEs, particularly
killed BCG in comparison to unexposed children [38]. in HIV-1-exposed children, is thus presently unknown.
These effects seem to be greatest among younger HIV- This uncertainty compels us to study the NSEs of the
1-exposed uninfected children. A recent study in Brazil BCG vaccine with a design that overcomes the methodo-
found the cellular immune response to BCG among logical challenges of observational studies.
younger HIV-1-exposed uninfected children to be im-
paired in comparison to their unexposed counterparts Purpose and rationale of this study
[35]. BCG-specific T-cell proliferation and interferon A rapid increase in the number of HIV-1-positive preg-
(IFN)-γ concentrations were lower in the younger than nant women receiving antiretroviral treatment (ART) to
in the older HIV-1-exposed uninfected infants and than prevent mother-to-child transmission (MTCT) of HIV-1
in the unexposed infants. This signified a delay in im- in the last 10 years [50], coupled with stagnating or in-
mune system maturation of HIV-1-exposed infants. creasing adult HIV-1 prevalence in LMICs, has resulted
Moreover, a Gambian study showed that children vacci- in an increase in the number of HIV-1-exposed unin-
nated at a median age of 14 days were less likely to de- fected children [51–53]. While data on HIV-1-exposed
velop a BCG scar if they were HIV-1-exposed uninfected uninfected children is scanty, available evidence suggests
than if they were HIV-1-unexposed and uninfected [39]. that these children have a higher morbidity and mortal-
A Rwandese study did not replicate this finding but ity in comparison to their unexposed counterparts [54,
found that HIV-1-exposed uninfected children vacci- 55]. In comparison to unexposed children, HIV-1-
nated at birth were substantially less likely than HIV-1- exposed uninfected children are more likely to contract
unexposed children to react to a tuberculin skin test at the common childhood infectious diseases in severe
6 months of age [40]. Other studies have linked negative forms [56–58], more likely to be hospitalized and more
tuberculin skin tests and absence of BCG scars to high likely to be resistant to treatment [59–62]. They are also
morbidity and mortality in children [41]. These BCG- vulnerable to diseases previously reported among im-
specific immune abnormalities are consistent with a munocompromised children, such as cytomegalovirus
wider range of other anomalies reported among very colitis [63], hemorrhagic chicken pox, Pneumocystis jiro-
young HIV-1-exposed uninfected children and are veci pneumonia and others [64].
thought to result from in-utero exposure to the HIV-1 Bearing the above considerations in mind, we believe
infection and/or antiretroviral drugs. These include: a re- that there is clinical equipoise between early and de-
duced CD4/CD8 ratio, reduced CD4+ and CD8+ naive ferred BCG vaccination in HIV-1-exposed, and probably
T-cell percentages, an increased percentage of activated even in HIV-1-unexposed infants. Moreover, the long-
CD8+ T-cells, augmented percentages of CD3(+)/4(−)/ term cost-implications and cost-effectiveness of the
8 (−) (DN) and DN/25(−)/44 (+) [42–44], altered cell- alternatives have never been investigated.
mediated immunity and T-cell maturation [42], reduced It is possible that well-timed BCG vaccination could
IL-12 production that persists until 6 months of age that have important NSEs in HIV-1-exposed children. We
could possibly be immunosuppressive [45], and several therefore propose a parallel-group superiority trial that
cytokine and other immune abnormalities [42, 43, 45– could inform the development of programmatically ap-
47]. Importantly, many of these anomalies appear to be propriate timing of BCG vaccination for HIV-1-exposed
transient, disappearing as children grow older. infants by measuring the safety, potential benefits or dis-
A small trial in South Africa in which newborn advantages, costs and cost-effectiveness of early versus
HIV-1-unexposed babies were randomized to receive late BCG.
BCG at birth or at 10 weeks of age, showed that
delaying the vaccination resulted in increased num- Methods/design
bers of BCG-specific CD4 T-cells, most importantly Hypotheses
polyfunctional T-cells co-expressing IFN-γ, tumor
necrosis factor (TNF)-α and IL-2, at 1 year of age 1. Compared to deferring BCG vaccination until
[48]. A recent study among HIV-1-exposed children 14 weeks of age, BCG administered within the first
Nankabirwa et al. Trials (2017) 18:152 Page 4 of 17

24 h of birth leads to at least a 30% relative risk To increase the diagnostic specificity, we will also try
reduction of severe illness1 among HIV-1-exposed to capture “clinical sepsis2,” where an infant with severe
infants illness and a negative blood culture, or from whom a
2. Delayed BCG results in at least a 30% higher blood culture could not be taken, has two or more of
production of TNF, IL-1β, IL-6, IL-10, IL-17, IL-22 the following findings: total lymphocyte count (TLC)
and IFN-γ in response to mycobacterial (from Myco- <5000 cells/mm3, absolute neutrophil count <1,500
bacterium tuberculosis and PPD) and non- cells/mm3, band to total polymorph ratio >0.2 and both
mycobacterial (from Escherichia coli, Candida albi- of two C-reactive protein (CRP) levels >1 mg/dl in two
cans and Staphylococcus aureus) antigens compared specimens taken 24 to 48 h apart. Because we are yet to
to when BCG is administered at birth in HIV-1- know in what proportion we can manage to collect the
exposed infants required blood specimens, the comparison between the
risk of “clinical sepsis” and/or “confirmed sepsis3” be-
tween the two trial arms will form part of our secondary
Primary objectives
analyses. Although it is unclear whether BCG scarring is
To compare
a marker of immune responsiveness in general and it
may be that such scarring is unrelated to protection
1. The risk of severe illness in the first 14 weeks of life,
against TB, it is being captured as a marker of “vaccine
and
take.” We will also explore whether children with a scar
2. The production of TNF, IL-1β, IL-6, IL-10, IL-17,
are less or more prone to severe illness during infancy.
IL-22 and IFN-γ in response to mycobacterial (from
The study will also examine the effect of the timing of
M. tuberculosis and PPD) and non-mycobacterial
BCG administration on infant growth.
(from E. coli, C. albicans and S. aureus) antigens at
birth (before BCG vaccination) and at 1, 14, 15 and
Intervention and co-interventions
28 weeks of age among HIV-1-exposed infants ad-
The study intervention is an intradermal administration
ministered BCG at birth with those administered a
to HIV-1-exposed babies of 0.05 ml Tubervac® BCG vac-
delayed BCG
cine from the Serum Institute of India, the vaccine cur-
rently provided by the Ugandan vaccination programme.
Secondary objectives After randomization within 24 h of birth, infants in one
To compare: trial arm will receive the vaccine immediately (BCG at
birth) while infants in the other arm will receive the
1. The risk of severe illness from 48 h after vaccine at 14 weeks of age (delayed BCG). Our study
randomization until 14 weeks of life among HIV-1- nurses, already experts in administering all childhood
exposed infants administered BCG at birth with vaccines, including Tubervac®, will ensure that the
those receiving delayed BCG vaccine is administered as scheduled.
2. HIV-1-exposed infants who received BCG at birth
with those administered a delayed BCG with respect Key outcome measures
to the occurrence of diarrhea, pneumonia and/or Primary outcome measures
severe illness from 14 weeks to 1 year of age
3. HIV-1-exposed infants who received BCG at birth 1. Severe illness in the first 14 weeks of life
with those administered a delayed BCG with respect 2. TNF, IL-1β, IL-6 and IFN-γ in response to mycobac-
to the occurrence of severe illness during infancy terial (M. tuberculosis and PPD) and non-
4. Occurrence of diarrhea, pneumonia and/or severe mycobacterial pathogens (E. coli, C. albicans and S.
illness during infancy in HIV-1-exposed infants vac- aureus) at 1, 14, 15 and 28 weeks of age
cinated with BCG at birth versus those who receive
delayed BCG Secondary outcome measures
5. The mortality from birth to 14 weeks of age in HIV-
1-exposed infants vaccinated with BCG at birth ver- 1. Severe illness from 48 h after randomization to
sus in those who are yet to receive BCG 14 weeks of life
6. Infant mortality between HIV-1-exposed babies vac- 2. Severe illness and/or diarrhea and/or pneumonia
cinated with BCG at birth versus in those who re- between 14 and 52 weeks and between 0 and
ceive delayed BCG 52 weeks of life
7. Through life-cycle modeling, long-term aggregate 3. Adverse events up to 52 weeks of life
health benefits and cost-implications of the delivery 4. Infant death
strategies, and estimate their cost-effectiveness 5. Growth up to 52 weeks of life
Nankabirwa et al. Trials (2017) 18:152 Page 5 of 17

6. BCG scar 12 weeks post vaccination

Study setting
The study is being conducted in Kawaala Health Center
III, Kitebi Health Center III in Kampala district and in
Mukono Health Center IV, which is located in Mukono
district, 27 km east of Kampala. These centers were
chosen based on their proximity to laboratories essential
for specimen processing and their high number of HIV-
1-infected women attending their antenatal and birth
clinics. These three government facilities together con-
duct at least 1200 deliveries per month.

Selection of participants, inclusion and exclusion

criteria
HIV-1-infected pregnant women between 28 and
40 weeks of gestation attending antenatal care (ANC)
and receiving the national elimination of mother to child
(eMTCT) programme’s ART for life (Option B+) strategy
are approached by the study team (Fig. 1). ANC clinics
in the national programme offer routine opt-out HIV
counseling, testing and ART according to the national
guidelines and WHO’s Option B+ recommendation.
Pregnant HIV-1-infected women attending our three
study clinics are informed about the BCG study, its pur-
pose, benefits, risks and their potential eligibility. The
study team will encourage the women to deliver at the
health units and to follow the national guidelines in
order to mitigate MTCT of HIV during pregnancy,
childbirth and the breastfeeding period.
For the pregnant women, these guidelines include:

1. Daily, orally administered cotrimoxazole

Fig. 1 Bacille Calmette-Guérin (BCG) trial participant screening,
2. Initiation of triple therapy (tenofovir, lamivudine and
enrollment and follow-up
efavirenz) starting during pregnancy and continuing
for life
3. Appropriately modified obstetric practices Inclusion criteria
4. Counselling and support for optimal infant feeding A baby born at a participating study clinic is included if
they:
For the HIV-1-exposed infants, these guidelines
include: 1. Have a mother with a positive HIV-1 test (ELISA or
rapid test)
1. Nevirapine syrup for the first 6 weeks regardless of 2. Are receiving periexposure prophylaxis as part of
infant feeding method the standard/national Option B+ guidelines in
2. Daily cotrimoxazole starting at 6 weeks of age Uganda
3. Have a mother who is of legal age for participation
For the HIV-1-infected infants, these guidelines in clinical research studies in Uganda or is an
include: emancipated minor
4. Have a mother/caregiver who resides within the
1. Initiation of ART with zidovudine, lamivudine and study area, is not intending to move out of the area
nevirapine or efavirenz as the preferred regimen. in the next 4 months and is likely to be traceable for
2. Daily cotrimoxazole starting at 6 weeks up to 12 months
Nankabirwa et al. Trials (2017) 18:152 Page 6 of 17

5. Have a mother/caregiver who gives informed Randomization

consent to trial participation Infants of HIV-1-infected mothers are randomized
6. Have a mother who has received ART for at least within 24 h of birth using pre-prepared randomization
4 weeks lists. Randomization is stratified by clinic using per-
muted blocks of varying size (4, 6 or 8). Eligible infants
Exclusion criteria are allocated in a 1:1 ratio to one of the two following
A newborn child is excluded if they have: arms: BCG vaccination within 24 h of birth or BCG
vaccination at week 14. All computer-generated
1. Serious congenital malformation(s) randomization lists were prepared for each clinic by an
2. Severe illness requiring hospitalization independent scientist, and are kept safely off site. Con-
3. A birth weight <2.0 kg cealment until the mother-infant pair has been deemed
4. A mother participating in another research study on eligible for final inclusion is ensured using an application
the day of enrollment or a mother who will running on a dedicated Android cell phone at each
participate in another research study within the next clinic.
3 months
5. A mother or other household member with Blinding/masking
symptoms and signs of TB on the day of enrollment Although scientifically advantageous, giving a placebo
6. A severely ill mother with a condition or conditions injection in the delayed BCG arm may be practically and
requiring hospitalization ethically questionable, and the study is an open-label
trial. Rigorously trained [67] research assistants who rec-
ord outcomes are blinded as much as possible to which
Trial size estimates for the primary hypotheses trial arm the children belong. Further, the vaccine is ad-
Hypothesis A ministered by routine health workers at the health cen-
The outcome in this hypothesis is severe illness in the ters that regularly administer the BCG vaccine as well as
first 14 weeks of life. Based on an estimated 10% risk of other childhood vaccines.
severe illness in the first 14 weeks of life [65], 90% power
and a risk ratio of 0.7, the required sample size is 1914 Implementation plan, staff recruitment and
(Table 1). training
Taking into account 10% attrition and that up to 5% of Field organization
the infants may have contracted HIV-1 infection when Recruitment at each of the three health facilities is done
enrolled, a total of 2200 babies are required for this trial, by three full-time research assistants (nurses) and a part-
1100 children in each study arm. time research assistant. The coordinator (ON) is respon-
sible for the day-to-day running of the study, and will
Hypothesis B report to the principle investigator (PI; VN) and co-PI
The outcome for this hypothesis is the production of (HS) who will ensure the overall scientific integrity of
relevant cytokines in response to mycobacterial and the study. The study will also employ a data manager
non-mycobacterial antigens. Based on previous studies that will oversee data quality. All study staff have
[33, 66] we expect that comparisons involving 50 chil- received intensive training on study procedures and
dren in each arm will be sufficient to detect biologically refresher training will be conducted during the trial.
relevant differences.
Screening of pregnant women
Table 1 Assumptions used for sample size calculation for the All HIV-1-positive pregnant women between 28 and
first 14 weeks of life 40 weeks of gestation attending antenatal care at the
Assumption Estimated level three clinics are informed about the study. Specifically,
Two-sided significance level 0.05 they are informed that if they choose to participate in
Ratio of sample size early/late BCG vaccination 1 the study, their infants will be randomized to receive the
BCG vaccine within 24 h of birth or at 14 weeks of age.
Percent of vaccinated at 14 weeks of age with 15
severe illness Consenting women are offered a CD4 cell count, a full
Percent of vaccinated at birth with severe illness 10
blood count including hemoglobin estimation and a clin-
ical assessment for HIV-1 disease staging and investiga-
Risk ratio: 0.7
tions for opportunistic infections, including TB.
Risk difference 5 Pregnant women who are eligible following the screen-
Total sample size (i.e., in both trial arms) 1914 ing and who consent to participate are recruited into the
BCG Bacillus Calmette-Guérin study at this first meeting. During recruitment, data on
Nankabirwa et al. Trials (2017) 18:152 Page 7 of 17

baseline characteristics and other social-demographic received tetanus toxoid during pregnancy as per national
variables is collected. recommendations.

Postnatal screening Follow-up and subject retention

Infants are assessed for other inclusion and exclusion Follow-up visits at the study clinic are scheduled on day
criteria, and if eligible, randomized into the study within 7 and at weeks 2, 6, 10, 14, 15, 26, 28, 48 and 52 after
24 h of birth. At birth, blood is drawn to perform a de- birth (Figs. 2 and 3). At each of these visits, data is col-
oxyribonucleic acid polymerase chain reaction (DNA- lected on the primary and secondary study outcomes,
PCR) assay for diagnosing any HIV-1 infection within adverse events (localized abscess formation, suppurative
6 weeks of birth. The DNA-PCR assays are performed in lymphadenitis and disseminated BCG infection), anthro-
batches (on samples collected over a 1-month period; pometry and on potential confounding variables (see
Fig. 2) using standard procedures. In addition, cord above). Mothers are given a cell phone to facilitate con-
blood is collected for the child’s full blood count at base- tact with study staff in case of an illness or other import-
line. A sample of this cord blood is stored for future ant child events.
investigations. Home visits will be made for participants missing
clinic visits while those coming to their scheduled clinic
Collection of data on baseline features and visits will have the transport fees reimbursed to reduce
potential confounders losses to follow-up. Data on illness, hospital visits and
Data on relevant baseline characteristics is collected dur- hospitalization is collected from interviews with the
ing pregnancy and at birth. This will include indicators mother, with more detailed information obtained from
of socio-demographic characteristics (maternal educa- the hospital records. Illness and hospitalization data is
tion and age, food insecurity, income, employment, etc.), based on recall between a previous visit and a current
health (illness history, parity, disclosure of HIV status to visit.
a partner, clinical examination, hemoglobin, Body Mass Upon enrollment and on subsequent visits, mothers
Index (BMI), viral load, CD4 cell count), premature rup- and other caregivers are informed about symptoms and
ture of membranes, postpartum complications, adher- signs of severe illness, and are encouraged to contact the
ence to ART and cotrimoxazole prophylaxis, sex of the study clinic in case the baby develops any of these symp-
child, infant feeding practices, and whether the mother toms. Each mother is provided with a durable note book,

Fig. 2 Time schedule for study procedures

Nankabirwa et al. Trials (2017) 18:152 Page 8 of 17

Fig. 3 Flow chart of study participants

in which she is instructed and trained to note down all physician or is admitted to a hospital. A child who
relevant events in her child. The front page of the book comes to a clinic with such symptoms is examined by a
will have clear reminders that she needs to contact the study nurse/midwife. If the child does not attend after
study team in case the child visits a health clinic, a the mother/caregiver has informed the study team of
Nankabirwa et al. Trials (2017) 18:152 Page 9 of 17

such signs or for a scheduled visit, the study nurse/mid- according to the Ugandan national guidelines. At week
wife will make a visit to the child’s house. If the nurse/ 14 and week 52, blood will be drawn by venipuncture
midwife classifies the condition as possible severe illness, for full/complete blood counts. TLC, differential count
a study physician will be called to examine the child. If and CRP will be captured as part of the sepsis screen in
the physician classifies the child as having severe illness, babies with clinical severe illness.
they will collect a blood specimen for blood culture
(using BACTEC) and a sepsis screen (CRP, TLC, differ- Functional immunology assessment
ential count, and band cell:neutrophil count ratio) as We will investigate both the induction of trained im-
soon as possible. A repeat specimen will be collected munity and heterologous immunity. Blood cells will be
again for CRP between 24 and 48 h. The study team will stimulated for 48 h with a lysate of BCG or M. tubercu-
do its utmost to obtain specimens for such a sepsis losis, and IFN-γ production will be measured by ELISA
screen from as many infants with signs of severe illness as a measure of classical adaptive immunity. Using the
as possible. However, we have yet to learn what percent- cell culture medium (RPMI) as a control stimulus,
age of them in this way can actually be screened for trained innate immunity will be assessed by stimulation
“clinical sepsis.” When 200 enrolled babies have been of blood for 24 h by unrelated nonspecific stimuli: lipo-
followed until the age of 6 months, the percentage of polysaccharide (LPS), heat-killed E. coli, C. albicans or S.
infants registered as having had severe illness who are aureus at 1, 15 and 28 weeks of age. Production of the
screened for clinical sepsis will be calculated. If this monocyte-derived cytokines TNF, IL-1β and IL-6 will be
percentage is less than 80%, measures will be taken to assessed by ELISA. Genome-wide gene transcription will
increase it above this value, if these measures are unsuc- be assessed in the unstimulated and stimulated whole
cessful, we will need to abandon the attempt to identify blood by ribonucleic acid (RNA) sequencing. Heterol-
clinical sepsis with the sepsis screen. Length and weight ogous immunity will be assessed by measurement of
measurements are taken twice at each visit, based on the both Th1 (IFN-γ) and Th17 (IL-17, IL-22) cytokines in
WHO guidelines, and z-scores calculated using WHO blood stimulated for 96 h with the nonspecific stimuli
child growth standards [68]. described above. The anti-inflammatory cytokine IL-10
will also be assessed at the 48-h stimulation time point.
Data collection and management Protein and messenger ribonucleic acid (mRNA) assess-
All trial procedures have been standardized. Training ex- ment will be our main immunological readouts.
ercises were undertaken and refresher sessions are
scheduled during the trial. Using Open Data Kit (ODK, Quality assurance and quality control
https://opendatakit.org), we will use pre-coded elec- We will use the WHO Good Laboratory Practice Guide-
tronic Case Report Forms (eCRFs)/questionnaires with lines [69] as our reference when establishing quality con-
range and consistency checks. Entered data will be trol laboratory procedures. HIV-1 serology, CD4 counts,
checked for consistency and any errors corrected hematology/biochemistry and immunology assays are
Source documents and files will not be destroyed with- submitted to a stringent quality assessment programme.
out specific written permission from the PI. Only the PI, On-going training and monitoring will occur during the
Co-PI, study coordinator and authorized study personnel study.
will have access to the CRFs and supporting documents
that will be kept on password-secured computers to en- Handling losses, withdrawals and protocol
sure participant confidentiality. To ensure correct oper- deviations
ation according to standard operating procedures (SOPs) Protocol deviations
all system users have been trained in their use and evalu- If a protocol violation, such as an inadequate informed
ated on a regular basis. consent, inappropriate randomization or concealment or
wrongful enrollment of an under-age mother occurs, the
Clinical laboratory tests PI will inform the Independent Data Monitoring Com-
Heel-prick blood samples are obtained from the children mittee (IDMC), which consists of a pediatrician, a statis-
for HIV-1 diagnosis. Dried blood spots (DBS) using a tician and a pediatrician/immunologist), the Project
commercial PCR kit are used for HIV-1 diagnosis. This Management Team (PMT, which consists of some of the
testing is supported by standardized protocols that have authors of this paper) and the local Ethical Committee
been successfully validated in Uganda. Flow cytometry is as soon as possible and no later than five working days
used for CD4 count measurement while standardized after the event. Detailed documentation mentioning the
automated procedures and techniques are used for dates and reasons of these protocol deviations will be
hematological and biochemical plasma assays. ELISA kept by the PI. If continuing the child in the trial puts
and/or rapid tests are used for maternal HIV-1 diagnosis the health of the mother or her child at risk, the child
Nankabirwa et al. Trials (2017) 18:152 Page 10 of 17

will no longer be followed up, but their data will be in- modified Poisson regression model with robust variance
cluded in time-to-event analyses up to the time that they [71, 72].
were removed from the trial. Also, autoregressive correlation approaches will be
used to account for multiple and possibly correlated ob-
servations within the same study participant. These
Plan of analysis
methods take the correlation structures into account
The data will be analyzed using Stata version 14 or later
and thereby the fact that measurements taken closer in
(StataCorp LP, TX, USA) and SAS version 9.2 or later.
time for an individual are likely to be more correlated
Continuous variables with right-skewed distributions will
than two measurements taken farther apart for that
be log-transformed. Means with standard deviations will
same individual. Efficacy will be calculated as 100 × (1 −
be used to summarize symmetrically distributed con-
RR), 100 × (1 − IRR) or 100 × (1 − HR). The prevalence of
tinuous variables while medians with interquartile ranges
BCG scarring will be compared between the two study
will be used for non-normally distributed continuous
arms using the abovementioned approaches, as appro-
variables, and percentages for categorical variables.
priate. Linear regression and or t tests will be used to
Between-group comparisons for continuous variables
compare the TNF and IFN-γ responses between the trial
that are symmetrically distributed will be made using t
arms. We will consider adjusting for potential con-
tests or, if adjustment for other variables are required,
founders if: (1) there are baseline imbalances between
linear regression, while Wilcoxon rank sum tests will be
trial arms, (2) the variables for which there are such
used to compare continuous variables where even their
baseline differences are strongly associated with the
log-transformed values remains nonsymmetrically dis-
study event or (3) they cause a ≥5% difference in the ef-
tributed. Group comparisons for categorical variables
fect measures when they are added to the main model.
will use chi-square tests and regression with generalized
Full case analyses will be default, but appropriate impu-
linear models of the binomial family with a log link
tations will be considered for missing data.
(“relative risk regression”) or logistic regression. We will
In addition to the intention-to-treat analysis, where
use two-sided statistical tests and 95% confidence inter-
outcomes will be compared according to the random al-
vals for descriptive results, efficacy estimates and safety
location, instrumental variable analyses will be con-
estimates. All relevant data; from both scheduled and
ducted in an attempt to estimate biological/causal effects
unscheduled visits will be included in the analysis.
of the actual receipt of the vaccine. In these analyses,
random allocation will be the instrument. To enable
Primary analysis such analyses, actual receipt as well as the age of BCG
The main study outcomes are severe illness during the vaccination will be captured in all participants. We will
first 14 weeks of life and immunological parameters. Al- also perform per-protocol and as-treated analyses.
though the sample size estimations are based on speci- The primary analyses will include all HIV-1-exposed
fied relative risks, i.e., the occurrence of one or more infants, adjusting for any imbalance in HIV-1 status be-
illness episodes per child, our final analysis will also use tween the two trial arms, as appropriate. However, we
incidence density and use Poisson regression or, in case will also perform an analysis of the large subgroup of ba-
of overdispersion, negative binomial regression analyses, bies who remain uninfected with HIV-1.
to estimate incidence rate ratios (IRR), enabling us to
capture more than one illness event in each child. Fur- Secondary analysis
ther, children who are lost to follow-up will be included
in time-to-event analyses, censoring children during 1. Risk of severe illness from 48 h after randomization
periods when data could not be recorded. All random- until the first 14 weeks of life, after the 14 weeks of
ized children will be included in an intention-to-treat infancy and during the entire infancy period will be
analysis. analyzed as described above for the primary analysis
To take into account multiple study events occurring 2. As a hypothesis-generating effort to understand the
in the same child, we will in the regression models latency period after which BCG may induce NSEs,
utilize generalized estimating equations (GEE), as appro- we will calculate the protection against severe illness
priate. The effect of the intervention on the occurrence during the first 14 weeks of life on a sliding scale
of severe illness will be estimated using both relative starting from randomization until seven completed
(relative risks (RR), IRR and hazard ratios (HR)) and ab- days post randomization
solute (risk differences and incidence rate differences 3. Risk of child death in the first 14 weeks as well as in
(IRD)) measures of effect. In the event that the “relative the remaining 38 weeks of infancy: time-to-event
risk regression” model fails to converge, risk ratios will analysis will be used to estimate time to severe ill-
be obtained from logistic regression [70] or from a ness or death. Kaplan-Meier methods and log-rank
Nankabirwa et al. Trials (2017) 18:152 Page 11 of 17

tests will be used for descriptive analyses. The results of the delivery strategies, and to compare their cost-
from these analyses will be presented as Kaplan- effectiveness. The decision model will have two arms,
Meier probabilities of the endpoints by a specified one for early, the other for delayed BCG administration
time for every 1000 children. Cox proportional haz- (Fig. 4). A Markov life-cycle will be attached to each de-
ards regression models will be used to estimate the livery arm, in which infants are followed in weekly cycles
effect of the intervention on child death. For this from 0 to 14 weeks, monthly cycles between 14 and
outcome, data will be censored when a child is lost 52 weeks, and thereafter yearly cycles. The model will
to follow-up track one annual Ugandan cohort of HIV-1-exposed in-
4. Risk of clinical sepsis and of “confirmed sepsis” in fants until they are 5 years old.
the first 14 as well as in the remaining 38 weeks of The main focus of the model will be severe illness dur-
infancy ing the first year of life, and will rely on prospectively
5. Safety will be analyzed according to type, frequency collected trial data (Table 2). For each cycle (weekly,
and severity of adverse events (AEs) that occur in monthly or annual) children may experience disease, or
children during the trial they may remain healthy. Children aged below 1 year are
6. Subgroup analysis for potential effect measure assumed to be at risk of experiencing a disease other
modification will be on the strata defined by than TB which may persist or from which they recover
maternal CD4 counts (<350 or ≥350 cells/μL), low before the next period. In all health states, individuals
birth weight (<2500 or ≥2500 g) and the babies’ HIV may die from that illness or from other causes. At birth,
status, sex of the child, in addition to other baseline all children are assumed to be alive and well.
characteristics which we have reasons to believe may The model represents a simplification of clinical real-
interact with delayed BCG administration. Such ity, and a simplifying assumption is that BCG-induced
subgroup/interaction analyses will be defined based protection against severe illnesses occurs only during
on the available scientific literature before the first 5 years of life. This reflects the availability of
embarking on the analyses, will not be driven by primary data and follow-up from the controlled trial.
study data, and will be described in a detailed Modeling of the health outcomes may be divided into
analysis plan three phases: (a) the observation period, (b) the period
7. Survival analyses using nonparametric and of assumed but gradually waning impact of the vaccine
parametric methods to extrapolate over 5 years the (less than 5 years) and (c) the post-vaccine period, where
incidence of NSEs and death for use in economic no further vaccination effects are assumed. Extrapolation
evaluation is required to capture probable health benefits during
the second of these phases, but the actual functional
Plan of analysis for economic evaluation shape and other assumptions about continued treatment
A Markov life-cycle decision model will be developed to benefits need to be informed by data through survival
model long-term aggregate health and cost implications analyses, and cannot be decided ex-ante. Transition

Fig. 4 The decision model

Nankabirwa et al. Trials (2017) 18:152 Page 12 of 17

Table 2 Economic evaluation data requirements and sources regardless of seriousness, severity, or presumed relation-
Parameter Source ship to study therapy, will be recorded using medical ter-
Epidemiology minology in the source document and on the AE page.
Age-specific disease incidence Primary data (for each condition)
Whenever possible, diagnoses will be given when signs
and symptoms are due to a common aetiology. Investi-
Background mortality Secondary data (Ugandan life table)
gators will record their opinion concerning the relation-
Effectiveness ship of the AE to BCG vaccination on the AE page. AE
Disease-specific efficacy Primary data reporting
Uptake of intervention Secondary data: effective coverage When the investigator, or trained physician, becomes
of vaccination aware that a serious AE (SAE) has occurred, the appro-
Aggregate health priate reporting form will be completed, and a copy
Years of life lost Secondary data (Ugandan life table) emailed to the local Institutional Review Board.
Disease weights Secondary data (Burden of Disease
Study) Management of SAEs
Costs In case of an SAE, the mother will be encouraged to im-
Intervention costs Primary data (prospectively costed)
mediately use the dedicated cell phone to contact the re-
search unit and or to bring the infant immediately to the
Treatment costs Primary data (retrospectively costed)
research unit. Infants will be seen by one of the research
midwives/clinicians and appropriate medical or surgical
probabilities are assumed to depend on vaccination interventions will be provided. All SAEs will be followed
strategy, and will be informed by the trial. More specific- up until resolution or until a stable clinical endpoint is
ally, transition probabilities, i.e., probabilities of moving reached. Insurance coverage is by BiomedicInsure
between the health states, will be calculated based on (http://www.biomedic-insure.com), and if any participant
hazard functions from the survival analyses (above). is harmed as a result of the BCG vaccine (within 2 years
After the extrapolation period, children are not assumed of the vaccination) they will be compensated
to be protected by BCG, but are assumed to experience
background mortality and morbidity for their remaining Monitoring
life time. Long-term health effects will be aggregated Independent Data Monitoring Committee (IDMC)
using disability-adjusted life years (DALYs) as the instru- A group of independent scientists with expertise in
ment, while incremental cost-effectiveness will be pediatrics, immunology, TB and statistics forms an
expressed in terms of US dollars per DALY averted. IDMC. After meeting shortly before trial start, it will
Costs will be collected from the perspective of the na- periodically review and assess available study data for
tional health care system, and include those related to safety, conduct and efficacy. The board will advise the
providing the BCG vaccine as well as those averted due project management on study continuation, modification
to the prevention of common illnesses in the short and or termination based on its reviews and pre-established
long term. stopping rules.
Interim analysis: an interim analysis for safety taking
AE reporting/clinical and safety monitoring into account the DAMOCLES group recommendations
An adverse event (AE) is defined as any harmful mani- [73, 74] will be performed by the IDMC when approxi-
festation occurring in a trial participant, whether this mately half of all the expected events have been
manifestation is related or not to the study BCG vaccine. recorded.
Potential AEs include localized abscess formation at the
injection site, suppurative lymphadenitis and dissemi- Auditing
nated BCG infection. The study will be monitored once a year by scientists
not involved in the day-to-day undertaking of the trial.
AE monitoring, recording and reporting
Potential AEs will be carefully monitored throughout the Protocol amendments
trial with specific questionnaires. The mothers will be Important protocol modifications (such as those result-
invited and encouraged to consult the study clinic in ing from changes to eligibility criteria, outcomes and
case of any disease or symptoms that arise between planned analyses) will be communicated to, and dis-
visits. AEs will be investigated at each follow-up visit. cussed among, the PMT members, discussed with the
Each AE will be reported spontaneously or in response IDMC and, when appropriate, the Ethics Committees.
to general, nondirected discussion with the attending Such modifications will also be reflected in amendments
midwife/researcher or physician/researcher. All AEs, to the description in ClinicalTrials.gov (NCT02606526).
Nankabirwa et al. Trials (2017) 18:152 Page 13 of 17

Dissemination of study findings 3. Higher incidence of severe illness during the first
Our study findings will be communicated in scientific 14 weeks of life among the children who receive
conferences and published in peer-reviewed scientific delayed BCG
journals. We will also prepare contextualized evidence
briefs for policy to relevant stakeholders including Ugan- To address the first concern, the study will exclude
dan authorities (Ministry of Health, National Drug Au- children with household members who have signs or
thority, Uganda National Academy of Sciences) and symptoms of TB or who have a diagnosis of TB. The risk
development partners supporting the Uganda National of TB infection among trial babies will, therefore, be
Expanded Programme on Immunization (UNEPI), WHO negligible [76]. In addition, the study will continue to ac-
(Vaccines and Biologicals) and UNICEF. We will, also in tively screen and refer suspected cases to the national
collaboration with CISMAC, prepare plain language TB clinics until the participating infants have received
summaries and press releases for consumer groups and BCG (at 14 weeks of age). The study will support diag-
mass media, respectively. Finally, we will engage the nosis and initiation of treatment at the TB clinics, of any
clinic staff and study participants and their communities of the participants’ household members with signs and
to inform them of our findings. symptoms of TB from randomization until 14 weeks of
age. Moreover, the study team will collaborate closely
with the families to identify, treat and actively follow up
Ethical considerations the babies should there be a history of exposure to active
According to WHO guidelines, BCG vaccination is con- TB or should they develop any symptoms indicative of
traindicated in children with known HIV-1 infection [22, TB. Should our study show that delayed vaccination is
35, 36] because of their increased risk of disseminated beneficial in terms of protecting HIV-1-exposed babies
BCG disease. But implementation of this guideline in against severe illness and in the long run it leads to a
many LMICs, including Uganda, is challenging. Accord- policy shift, with HIV-1-exposed children from house-
ing to the expanded programme of immunization, the holds without TB being vaccinated late, our findings will
vaccine is to be given as soon as possible after birth. inform subsequent programmes and would thereby con-
Diagnosing HIV-1 infection in infants requires a PCR tribute to improved infant health.
which is not accessible to Ugandan children at birth. To mitigate the second concern, an HIV-1 diagnosis
Therefore, Ugandan HIV-1-exposed newborns receive using DNA PCR is obtained at 6 weeks of age. All children
BCG at birth. On the one hand, receiving the vaccine at who are allocated to receive BCG at birth and are found to
birth could be of potential benefit through morbidity be infected with HIV-1 will be referred to a pediatrician for
and mortality reduction in the initial weeks of life. On assessment, while the HIV-1-infected participants allocated
the other hand, recent medical literature indicates that to receive BCG later will be referred to government clinics
HIV-1-exposed uninfected infants [35, 38–41], and for a decision to vaccinate or not. In a sample of 1100 chil-
maybe even HIV-1-unexposed infants [48], who receive dren, the probability of disseminated BCG is extremely low.
BCG somewhat later may yield more vigorous immune This is because in the presence of the current Ugandan
responses than those who receive it at birth. It is con- peripartum ART prophylaxis the risk of HIV-1 infection
ceivable that such a delay translates into better protec- assessed at 6 weeks of age is less than 3.5% [77]. Therefore,
tion against TB. Delaying vaccination may also carry the among the 1100 babies who receive BCG at birth, fewer
advantage of increasing the likelihood that HIV-1- than 40 children are likely to acquire HIV-1. Among chil-
exposed infants may be reliably diagnosed with respect dren with immunodeficiency, it is estimated that the risk of
to HIV-1 infection, both for logistic reasons but also be- disseminated BCG infection following vaccination is less
cause there is a genuine risk of HIV-1 transmission via than 1.6 per 1,000,000 vaccinated children [78]. Therefore,
breast milk during the first months of life [75]. This trial the likelihood of any child contracting disseminated BCG
will compare BCG take and risk of severe illness (includ- in our study is less than 0.015% if the government clinics
ing death) among HIV-1-exposed children receiving the decide to vaccinate all the HIV-1-infected infants at 14 years
vaccine shortly after birth to such a risk in infants re- of age, and 0.007% if they decide not to vaccinate any of
ceiving BCG at 14 weeks of age. them. So, this risk will be equal to, or lower than, that out-
The main ethical concerns specific to study are: side of the study, given that Ugandan HIV-1-exposed in-
fants receive BCG shortly after birth. In the study, we shall
1. Acquisition of TB before BCG has engendered a reduce this risk further with prompt standard TB prophy-
protective immune response (delayed BCG trial laxis using isoniazid as soon as an HIV diagnosis is made.
arm) With regard to the third concern, appropriate diagnos-
2. Severe BCG disease among HIV-1-positive infants tic procedures and treatment will be provided according
(early BCG trial arm) to national guidelines to all severely ill children.
Nankabirwa et al. Trials (2017) 18:152 Page 14 of 17

Ethics permission to conduct the study has been ob- among these infants, who will constitute an increasing
tained from the School of Medicine, Research and Ethics part of sub-Saharan African child cohorts in the years to
Committee (Makerere University) as well as from the come.
Regional Committees for Medical and Health Research
Ethics in Norway (REK) [79]. The trial has also been ap- Trial status
proved by the National Council of Science and Technol- Recruiting since July 2016.
ogy and the National Drug Authority in Uganda. The
study is performed in accordance with International Endnotes
Conference on Harmonization (ICH) guidelines for 1
Among children aged below 2 months of age, severe
Good Clinical Practice. Written individual informed illness (other than TB) will be defined as illness that is
consent in local language is obtained from each of the associated with any of the following danger signs ob-
participating mothers by trained study staff. The consent served or verified by a study clinician: inability to feed
process will explain the nature of the study, the risks or vomiting of everything, lethargy or unconsciousness,
and benefits of participating in the study, the interven- severe lower chest in-drawing, axillary temperature of
tion and that intervention allocation is by a random ≥37.5 °C or <35.5 °C, grunting, cyanosis, convulsions or
process. In situations requiring translation or in cases a history of convulsions, and/or results in hospitalization
where the mother is unable to read and write, the con- and/or results in death.Among children aged 2 months
sent process will take place in the presence of an inde- of age or older, severe illness (other than TB) will be de-
pendent third person, who will act as a witness and also fined as illness that: is associated with at least one of the
co-sign the Consent Form. Additional consent is ob- following danger signs observed or verified by a study
tained from study participants for the collection and clinician: inability to drink or breastfeed, lethargy or un-
storage of blood specimens for ancillary studies. Confi- consciousness, vomiting of all feeds, convulsions or a
dentiality of information and the right of the participant history of convulsions, and/or results in hospitalization
to withdraw from the study at any time during the study and/or results in death. Hospitalization and death result-
is explained to the mothers. All study staff are trained ing from violent injury or burns will not contribute to
on participant confidentiality and autonomy. the severe illness definition.
2
In infants from whom a blood culture was not per-
Discussion formed or in whom a blood culture was negative, clinical
This trial compares the effect BCG vaccination at birth sepsis is signaled by one of our study physicians (VN,
with BCG vaccination at 14 weeks of age in HIV-1- ON, JKT or GN) having initiated treatment for sepsis or
exposed babies on (1) severe illness in the first 14 weeks confirmed that such treatment was appropriate and/or
of life, (2) TNF, IL-1β, IL-6, IL-17, IL-22 and IFN-γ in by a positive sepsis screen. A positive sepsis screen is the
response to mycobacterial and non-mycobacterial patho- presence of any two of the following: total leukocyte
gens and (3) severe illness in 14–52 weeks of life and count <5000/mm3; absolute neutrophil count <1500/
throughout infancy. mm3; band cell:neutrophil count ratio >0.2; both of two
The study circumvents the methodological challenges C-reactive protein serum levels in specimens taken 24 h
of earlier observational studies that reported an associ- to 48 h apart >1 mg/dl.
ation between BCG vaccination and NSEs. The exclu- 3
Confirmed sepsis is signaled by a blood culture that is
sion criteria combined with the random allocation and positive for bacteria known to cause infant sepsis.
large sample size will substantially reduce the likelihood
that children with higher morbidity end up in either Additional file
group. There will be no survival bias linked to missing
vaccination cards as BCG is administered before out- Additional file 1: SPIRIT 2013 Checklist: recommended items to address
come measurement. This study will be further strength- in a clinical trial protocol and related documents. (DOC 122 kb)
ened by an exploration of immunological mechanisms
for the vaccine’s hypothesized NSEs. The study will ex- Abbreviations
AE: Adverse event; BCG: Bacille Calmette-Guérin; CD 4+: Positive to cluster of
plore the trained innate immunity and heterologous im- differentiation 4; CD 8+: Positive to cluster of differentiation 8;
munity processes that underlie the immunological CISMAC: Centre for Intervention Science in Maternal and Child Health;
responses thought to account for the NSEs of BCG. CRF: Case Report Form; DNA: Deoxyribonucleic acid; DPT: Diphtheria
Pertussis Tetanus; eCRF: electronic Case Report Form; ELISA: Enzyme-linked
This trial, comparing NSEs when the vaccine is given immunosorbent assay; eMTCT: Elimination of mother-to-child transmission of
at birth and when it is delayed, could inform the devel- HIV-1; GCP: Good Clinical Practice; HIV-1: Human immunodeficiency virus-1;
opment of programmatically appropriate timing of BCG IDMC: Independent Data Monitoring Committee; IFN-γ: Interferon gamma; IL-
x: Interleukin (x refers to number); LMIC: Low- and middle-income country;
vaccination for HIV-1-exposed infants. This, in turn MTCT: Mother-to-child transmission of HIV-1; NDA: National Drug Authority;
could importantly impact morbidity and mortality NK: Natural killer; PCR: Polymerase chain reaction; PPD: Purified protein
Nankabirwa et al. Trials (2017) 18:152 Page 15 of 17

derivative; SAE: Severe adverse events; TB: Tuberculosis; TNF-α: Tumor Publisher’s Note
necrosis factor alpha; WHO: World Health Organization Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
Acknowledgements
Author details
Makerere University: we thank Samuel Daniel Kagongwe, Stephen Kabanda, 1
Department of Epidemiology and Biostatics, School of Public Health,
Samuel Kirabira, Andrew Mark Vivian Kirabo, David Mukunya, Ekwaro Obuku
College of Health Sciences, Makerere University, Kampala, Uganda. 2Centre
and Josephine Tumuhamye.
for Intervention Science in Maternal and Child Health, Centre for
We are also very grateful to Dr. Anthony Konde and all the staff at Mukono
International Health, University of Bergenhttp://www.cismac.org.
Health Center IV, Kawaala Health Center III and Kitebi Health Center III for 3
Department of Paediatrics and Child Health, School of Medicine, College of
their support and willingness to participate in the study and Isaac
Health Sciences, Makerere University, Kampala, Uganda. 4Department of
Ssewanyana at the Central Public Health Laboratory in Kampala.
Internal Medicine and Radboud Center for Infectious Diseases, Radboud
University of Bergen and CISMAC: we are very thankful to Maharaj K. Bhan,
University Medical Center, Nijmegen, Netherlands. 5Global Women and
Rajiv Bahl, José C. Martines, Marte Emilie Sandvik Haaland and Solfrid Vikøren
Children’s Health, Norwegian Institute of Public Health, Oslo, Norway.
for their valuable contributions.
Serum Institute of India: we gratefully acknowledge Serum Institute of India
Received: 21 September 2016 Accepted: 7 March 2017
for contributing the Tubervac® vaccine for the trial.

Funding
This study is funded by the Research Council of Norway through its Centres References
of Excellence scheme and the University of Bergen (UiB), Norway to the 1. Elguero E, Simondon KB, Vaugelade J, Marra A, Simondon F. Non-specific
Centre for Intervention Science in Maternal and Child Health (CISMAC; effects of vaccination on child survival? A prospective study in Senegal.
project reference number 223269). CISMAC’s independent review process Trop Med Int Health. 2005;10(10):956–60.
contributed to the trial design. The study sponsor is Makerere University, 2. Krishnan A, Srivastava R, Dwivedi P, Ng N, Byass P, Pandav CS. Non-specific sex-
College of Health Sciences. PO Box 7072, Kampala, Uganda. While the differential effect of DTP vaccination may partially explain the excess girl child
University’s scientists (VN, ON, JKT and GN) are deeply involved in the mortality in Ballabgarh, India. Trop Med Int Health. 2013;18(11):1329–37.
development and implementation of the trial, the University’s central 3. Aaby P, Benn CS, Nielsen J, Ravn H. Sex-differential non-specific effects of
authority is not involved in the day-to-day running of the study. BCG and DTP in Cebu, The Philippines. Int J Epidemiol. 2009;38(1):320–3.
author reply 323-324.
4. Shann F. The nonspecific effects of vaccines and the expanded program on
Availability of data and materials immunization. J Infect Dis. 2011;204(2):182–4.
The SPIRIT Checklist (Additional file 1). 5. Shann F. The non-specific effects of vaccines. Arch Dis Child. 2010;95(9):
662–7.
Authors’ contributions 6. Hirve S, Bavdekar A, Juvekar S, Benn CS, Nielsen J, Aaby P. Non-specific and
VN conceptualized and designed the study, drafted the first manuscript, sex-differential effects of vaccinations on child survival in rural western
participated in subsequent manuscript writing and read and approved the India. Vaccine. 2012;30(50):7300–8.
final manuscript. JKT participated in manuscript writing and read and 7. Roth A, Garly ML, Jensen H, Nielsen J, Aaby P. Bacillus Calmette-Guerin
approved the final manuscript. ON participated in manuscript writing and vaccination and infant mortality. Expert Rev Vaccines. 2006;5(2):277–93.
read and approved the final manuscript. TT participated in manuscript 8. Shann F. Commentary: BCG vaccination halves neonatal mortality. Pediatr
writing and read and approved the final manuscript. GN participated in Infect Dis J. 2012;31(3):308–9.
manuscript writing and read and approved the final manuscript. BR 9. Aaby P, Roth A, Ravn H, Napirna BM, Rodrigues A, Lisse IM, Stensballe L,
participated in manuscript writing and read and approved the final Diness BR, Lausch KR, Lund N, et al. Randomized trial of BCG vaccination at
manuscript. MGN participated in manuscript writing and read and approved birth to low-birth-weight children: beneficial nonspecific effects in the
the final manuscript. HS conceptualized and designed the study, provided neonatal period? J Infect Dis. 2011;204(2):245–52.
comments, participated in subsequent protocol writing and read and 10. Biering-Sorensen S, Aaby P, Napirna BM, Roth A, Ravn H, Rodrigues A,
approved the final manuscript. Whittle H, Benn CS. Small randomized trial among low-birth-weight
children receiving bacillus Calmette-Guerin vaccination at first health center
contact. Pediatr Infect Dis J. 2012;31(3):306–8.
Authors’ information 11. Roth AE, Stensballe LG, Garly ML, Aaby P. Beneficial non-targeted effects of
Dr. Victoria Nankabirwa is a senior lecturer in the Department of BCG—ethical implications for the coming introduction of new TB vaccines.
Epidemiology and Biostatistics, School of Public Health, College of Health Tuberculosis (Edinb). 2006;86(6):397–403.
Sciences, Makerere University and a researcher at the University of Bergen’s 12. Stensballe LG, Nante E, Jensen IP, Kofoed PE, Poulsen A, Jensen H, Newport
Center for International Health, Norway. M, Marchant A, Aaby P. Acute lower respiratory tract infections and
respiratory syncytial virus in infants in Guinea-Bissau: a beneficial effect of
Competing interests BCG vaccination for girls community based case-control study. Vaccine.
The authors declare that they have no competing interests. 2005;23(10):1251–7.
13. Jason J, Archibald LK, Nwanyanwu OC, Kazembe PN, Chatt JA, Norton E,
Dobbie H, Jarvis WR. Clinical and immune impact of Mycobacterium bovis
Consent for publication
BCG vaccination scarring. Infect Immun. 2002;70(11):6188–95.
Consent for publication is obtained from all participants as part of the
14. Nankabirwa V, Tumwine JK, Mugaba PM, Tylleskar T, Sommerfelt H, Group
informed consent process.
P-ES. Child survival and BCG vaccination: a community based prospective
cohort study in Uganda. BMC Public Health. 2015;15:175.
Ethics approval and consent to participate 15. Rousseau MC, Parent ME, St-Pierre Y. Potential health effects from non-
Ethics permission to conduct the study has been obtained from the School specific stimulation of the immune function in early age: the example of
of Medicine, Research and Ethics Committee (Makerere University), reference BCG vaccination. Pediatr Allergy Immunol. 2008;19(5):438–48.
number: 2015–114 as well as from the Regional Committees for Medical and 16. Aaby P, Shaheen SO, Heyes CB, Goudiaby A, Hall AJ, Shiell AW, Jensen H,
Health Research Ethics in Norway (REK; reference 2015/1854). The trial has Marchant A. Early BCG vaccination and reduction in atopy in Guinea-Bissau.
also been approved by the National Council of Science and Technology Clin Exp Allergy. 2000;30(5):644–50.
(reference number: HS 1925 and the National Drug Authority (reference 17. Ota MO, van der Sande MA, Walraven GE, Jeffries D, Nyan OA, Marchant A,
number: 472/NDA/DID/02/2016) in Uganda. Written individual informed McAdam KP. Absence of association between delayed type hypersensitivity
consent for study participation, in the local language is obtained from each to tuberculin and atopy in children in The Gambia. Clin Exp Allergy. 2003;
of the participating mothers. 33(6):731–6.
Nankabirwa et al. Trials (2017) 18:152 Page 16 of 17

18. Choi IS, Koh YI. Therapeutic effects of BCG vaccination in adult asthmatic 39. Ota MO, O’Donovan D, Marchant A, Yamuah L, Harding E, Jaffar S, McAdam
patients: a randomized, controlled trial. Ann Allergy Asthma Immunol. 2002; KP, Corrah T, Whittle H. HIV-negative infants born to HIV-1 but not HIV-2-
88(6):584–91. positive mothers fail to develop a Bacillus Calmette-Guerin scar. AIDS. 1999;
19. Elliott AM, Nakiyingi J, Quigley MA, French N, Gilks CF, Whitworth JA. Inverse 13(8):996–8.
association between BCG immunisation and intestinal nematode infestation 40. Msellati P, Dabis F, Lepage P, HItimana DG, Van goethem C, Van De Perre P.
among HIV-1-positive individuals in Uganda. Lancet. 1999;354(9183):1000–1. BCG vaccination and pediatric HIV infection—Rwanda, 1988-1990. MMWR.
20. Barreto ML, Rodrigues LC, Silva RC, Assis AM, Reis MG, Santos CA, Blanton 1991;40(48):833–6.
RE. Lower hookworm incidence, prevalence, and intensity of infection in 41. Garly ML, Martins CL, Bale C, Balde MA, Hedegaard KL, Gustafson P, Lisse IM,
children with a Bacillus Calmette-Guerin vaccination scar. J Infect Dis. 2000; Whittle HC, Aaby P. BCG scar and positive tuberculin reaction associated
182(6):1800–3. with reduced child mortality in West Africa. A non-specific beneficial effect
21. Randall AE, Perez MA, Floyd S, Black GF, Crampin AC, Ngwira B, Pistoni WN, of BCG? Vaccine. 2003;21(21-22):2782–90.
Mulawa D, Sichali L, Mwaungulu L, et al. Patterns of helminth infection and 42. Clerici M, Saresella M, Colombo F, Fossati S, Sala N, Bricalli D, Villa ML,
relationship to BCG vaccination in Karonga District, northern Malawi. Trans R Ferrante P, Dally L, Vigano A. T-lymphocyte maturation abnormalities in
Soc Trop Med Hyg. 2002;96(1):29–33. uninfected newborns and children with vertical exposure to HIV. Blood.
22. World Health Organization. Global tuberculosis report 2013. Geneva: World 2000;96(12):3866–71.
Health Organization; 2013. 43. Nielsen SD, Jeppesen DL, Kolte L, Clark DR, Sorensen TU, Dreves AM, Ersboll
23. SAGE Working Group on Non-specific Effects of Vaccines (March 2013 – AK, Ryder LP, Valerius NH, Nielsen JO. Impaired progenitor cell function in
June 2013). http://www.who.int/immunization/sage/sage_wg_non_specific_ HIV-negative infants of HIV-positive mothers results in decreased thymic
effects_vaccines_march2013/en/. Accessed 15 Dec 2015. output and low CD4 counts. Blood. 2001;98(2):398–404.
24. Fine PE, Williams TN, Aaby P, Kallander K, Moulton LH, Flanagan KL, Smith 44. Embree J, Bwayo J, Nagelkerke N, Njenga S, Nyange P, Ndinya-Achola J,
PG, Benn CS, Working Group on Non-specific Effects of Vaccines. Pamba H, Plummer F. Lymphocyte subsets in human immunodeficiency
Epidemiological studies of the “non-specific effects” of vaccines: I–data virus type 1-infected and uninfected children in Nairobi. Pediatr Infect Dis J.
collection in observational studies. Trop Med Int Health. 2009;14(9):969–76. 2001;20(4):397–403.
25. Farrington CP, Firth MJ, Moulton LH, Ravn H, Andersen PK, Evans S, Working 45. Chougnet C, Kovacs A, Baker R, Mueller BU, Luban NL, Liewehr DJ,
Group on Non-specific Effects of Vaccines. Epidemiological studies of the Steinberg SM, Thomas EK, Shearer GM. Influence of human
non-specific effects of vaccines: II–methodological issues in the design and immunodeficiency virus-infected maternal environment on development of
analysis of cohort studies. Trop Med Int Health. 2009;14(9):977–85. infant interleukin-12 production. J Infect Dis. 2000;181(5):1590–7.
26. Kristensen I, Aaby P, Jensen H. Routine vaccinations and child survival: 46. Kuhn L, Coutsoudis A, Moodley D, Mngqundaniso N, Trabattoni D, Shearer
follow up study in Guinea-Bissau, West Africa. BMJ. 2000;321(7274):1435–8. GM, Clerici M, Coovadia HM. Interferon-gamma and interleukin-10
27. Vaugelade J, Pinchinat S, Guiella G, Elguero E, Simondon F. Non-specific production among HIV-1-infected and uninfected infants of HIV-1-infected
effects of vaccination on child survival: prospective cohort study in Burkina mothers. Pediatr Res. 2001;50(3):412–6.
Faso. BMJ. 2004;329(7478):1309. 47. Levy JA, Hsueh F, Blackbourn DJ, Wara D, Weintrub PS. CD8 cell
28. Kandasamy R, Voysey M, McQuaid F, de Nie K, Ryan R, Orr O, Uhlig U, Sande noncytotoxic antiviral activity in human immunodeficiency virus-infected
C, O’Connor D, Pollard AJ. Non-specific immunological effects of selected and -uninfected children. J Infect Dis. 1998;177(2):470–2.
routine childhood immunisations: systematic review. BMJ. 2016;355:i5225. 48. Kagina BM, Abel B, Bowmaker M, Scriba TJ, Gelderbloem S, Smit E, Erasmus
29. Higgins JP, Soares-Weiser K, Lopez-Lopez JA, Kakourou A, Chaplin K, M, Nene N, Walzl G, Black G, et al. Delaying BCG vaccination from birth to
Christensen H, Martin NK, Sterne JA, Reingold AL. Association of BCG, DTP, 10 weeks of age may result in an enhanced memory CD4 T cell response.
and measles containing vaccines with childhood mortality: systematic Vaccine. 2009;27(40):5488–95.
review. BMJ. 2016;355:i5170. 49. Blakney AK, Tchakoute CT, Hesseling AC, Kidzeru EB, Jones CE, Passmore JA,
30. Wilson CB. Applying contemporary immunology to elucidate heterologous Sodora DL, Gray CM, Jaspan HB. Delayed BCG vaccination results in minimal
effects of infant vaccines and to better inform maternal-infant immunization alterations in T cell immunogenicity of acellular pertussis and tetanus
practices. Front Immunol. 2015;6:64. immunizations in HIV-exposed infants. Vaccine. 2015;33(38):4782–9.
31. Netea MG, Quintin J, van der Meer JW. Trained immunity: a memory for 50. UNAIDS. Global report: UNAIDS report on the global AIDS epidemic 2010. 2010.
innate host defense. Cell Host Microbe. 2011;9(5):355–61. 51. AIDS Epidemic Update, Regional Summary: Sub-Saharan Africa. http://data.
32. Saeed S, Quintin J, Kerstens HH, Rao NA, Aghajanirefah A, Matarese F, unaids.org/pub/Report/2008/JC1526_epibriefs_subsaharanafrica_en.pdf.
Cheng SC, Ratter J, Berentsen K, van der Ent MA, et al. Epigenetic Accessed 20 Jan 2014.
programming of monocyte-to-macrophage differentiation and trained 52. Mofenson LM. Advances in the prevention of vertical transmission of human
innate immunity. Science. 2014;345(6204):1251086. immunodeficiency virus. Semin Pediatr Infect Dis. 2003;14(4):295–308.
33. Kleinnijenhuis J, Quintin J, Preijers F, Joosten LA, Ifrim DC, Saeed S, Jacobs C, 53. Agangi A, Thorne C, Newell ML. Increasing likelihood of further live births in
van Loenhout J, de Jong D, Stunnenberg HG, et al. Bacille Calmette-Guerin HIV-infected women in recent years. BJOG. 2005;112(7):881–8.
induces NOD2-dependent nonspecific protection from reinfection via 54. Filteau S. The HIV-exposed, uninfected African child. Trop Med Int Health.
epigenetic reprogramming of monocytes. Proc Natl Acad Sci U S A. 2012; 2009;14(3):276–87.
109(43):17537–42. 55. Afran L, Garcia Knight M, Nduati E, Urban BC, Heyderman RS, Rowland-
34. Selin LK, Varga SM, Wong IC, Welsh RM. Protective heterologous antiviral Jones SL. HIV-exposed uninfected children: a growing population with a
immunity and enhanced immunopathogenesis mediated by memory T cell vulnerable immune system? Clin Exp Immunol. 2014;176(1):11–22.
populations. J Exp Med. 1998;188(9):1705–15. 56. Slogrove A, Reikie B, Naidoo S, De Beer C, Ho K, Cotton M, Bettinger J,
35. Mazzola TN, da Silva MT, Abramczuk BM, Moreno YM, Lima SC, Zorzeto Speert D, Esser M, Kollmann T. HIV-exposed uninfected infants are at
TQ, Passeto AS, Vilela MM. Impaired Bacillus Calmette-Guerin cellular increased risk for severe infections in the first year of life. J Trop Pediatr.
immune response in HIV-exposed, uninfected infants. AIDS. 2012;58(6):505–8.
2011;25(17):2079–87. 57. Slogrove AL, Cotton MF, Esser MM. Severe infections in HIV-exposed
36. Hesseling AC, Cotton MF, Fordham von Reyn C, Graham SM, Gie RP, Hussey uninfected infants: clinical evidence of immunodeficiency. J Trop Pediatr.
GD. Consensus statement on the revised World Health Organization 2010;56(2):75–81.
recommendations for BCG vaccination in HIV-infected infants. Int J Tuberc 58. Taron-Brocard C, Le Chenadec J, Faye A, Dollfus C, Goetghebuer T, Gajdos V,
Lung Dis. 2008;12(12):1376–9. Labaune JM, Perilhou A, Mandelbrot L, Blanche S, et al. Increased risk of
37. Hesseling AC, Caldwell J, Cotton MF, Eley BS, Jaspan HB, Jennings K, Marais serious bacterial infections due to maternal immunosuppression in HIV-
BJ, Nuttall J, Rabie H, Roux P, et al. BCG vaccination in South African HIV- exposed uninfected infants in a European country. Clin Infect Dis. 2014;
exposed infants—risks and benefits. S Afr Med J. 2009;99(2):88–91. 59(9):1332–45.
38. Miles DJ, Gadama L, Gumbi A, Nyalo F, Makanani B, Heyderman RS. Human 59. Heresi GP, Caceres E, Atkins JT, Reuben J, Doyle M. Pneumocystis carinii
immunodeficiency virus (HIV) infection during pregnancy induces CD4 T- pneumonia in infants who were exposed to human immunodeficiency
cell differentiation and modulates responses to Bacille Calmette-Guerin virus but were not infected: an exception to the AIDS surveillance case
(BCG) vaccine in HIV-uninfected infants. Immunology. 2010;129(3):446–54. definition. Clin Infect Dis. 1997;25(3):739–40.
Nankabirwa et al. Trials (2017) 18:152 Page 17 of 17

60. Adhikari M, Kauchali S, Moodley A. Clinical profile and morbidity pattern of

infants born to HIV infected mothers in Durban South Africa. Indian Pediatr.
2006;43(9):804–8.
61. von Mollendorf C, von Gottberg A, Tempia S, Meiring S, de Gouveia L, Quan
V, Lengana S, Avenant T, du Plessis N, Eley B, et al. Increased risk for and
mortality from invasive pneumococcal disease in HIV-exposed but
uninfected infants aged <1 year in South Africa, 2009-2013. Clin Infect Dis.
2015;60(9):1346–56.
62. Izadnegahdar R, Fox MP, Jeena P, Qazi SA, Thea DM. Revisiting pneumonia
and exposure status in infants born to HIV-infected mothers. Pediatr Infect
Dis J. 2014;33(1):70–2.
63. Manicklal S, van Niekerk AM, Kroon SM, Hutto C, Novak Z, Pati SK, Chowdhury
N, Hsiao NY, Boppana SB. Birth prevalence of congenital cytomegalovirus
among infants of HIV-infected women on prenatal antiretroviral prophylaxis in
South Africa. Clin Infect Dis. 2014;58(10):1467–72.
64. Pifer LL, Hughes WT, Stagno S, Woods D. Pneumocystis carinii infection:
evidence for high prevalence in normal and immunosuppressed children.
Pediatrics. 1978;61(1):35–41.
65. Kuhn L, Kasonde P, Sinkala M, Kankasa C, Semrau K, Scott N, Tsai WY,
Vermund SH, Aldrovandi GM, Thea DM. Does severity of HIV disease in HIV-
infected mothers affect mortality and morbidity among their uninfected
infants? Clin Infect Dis. 2005;41(11):1654–61.
66. Kleinnijenhuis J, Quintin J, Preijers F, Joosten LA, Jacobs C, Xavier RJ, van der
Meer JW, van Crevel R, Netea MG. BCG-induced trained immunity in NK cells:
role for non-specific protection to infection. Clin Immunol. 2014;155(2):213–9.
67. Tylleskar T, Jackson D, Meda N, Engebretsen IM, Chopra M, Diallo AH,
Doherty T, Ekstrom EC, Fadnes LT, Goga A, et al. Exclusive breastfeeding
promotion by peer counsellors in sub-Saharan Africa (PROMISE-EBF): a
cluster-randomised trial. Lancet. 2011;378(9789):420–7.
68. World Health Organization. WHO child growth standards: length/height-for-
age, weight-for-age, weight-for-length, weight-for-height and body mass
index-for-age: methods and development. Geneva: World Health
Organization; 2006. p. 312.
69. World Health Organization. Good clinical laboratory practice (GCLP). Geneva:
World Health Organization; 2009.
70. Cummings P. Methods for estimating adjusted risk ratios. Stata J. 2009;9(2):
175–96.
71. Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence
ratios and differences. Am J Epidemiol. 2005;162(3):199–200.
72. Zou G. A modified Poisson regression approach to prospective studies with
binary data. Am J Epidemiol. 2004;159(7):702–6.
73. The DAMOCLES Study Group. A proposed charter for clinical trial data
monitoring committees: helping them to do their job well. Lancet. 2005;
365(9460):711–22.
74. Grant AM, Altman DG, Babiker AB, Campbell MK, Clemens FJ, Darbyshire JH,
Elbourne DR, McLeer SK, Parmar MK, Pocock SJ, et al. Issues in data
monitoring and interim analysis of trials. Health Technol Assess. 2005;9(7):1–
238. iii-iv.
75. Siegfried N, van der Merwe L, Brocklehurst P, Sint TT. Antiretrovirals for
reducing the risk of mother-to-child transmission of HIV infection. Cochrane
Database Syst Rev. 2011;7:CD003510.
76. Rieder HL. Opportunity for exposure and risk of infection: the fuel for the
tuberculosis pandemic. Infection. 1995;23(1):1–3.
77. Kesho Bora Study G, de Vincenzi I. Triple antiretroviral compared with
zidovudine and single-dose nevirapine prophylaxis during pregnancy and
breastfeeding for prevention of mother-to-child transmission of HIV-1
(Kesho Bora study): a randomised controlled trial. Lancet Infect Dis. 2011;
11(3):171–80. Submit your next manuscript to BioMed Central
78. Bernatowska EA, Wolska-Kusnierz B, Pac M, Kurenko-Deptuch M, Zwolska Z,
Casanova JL, Piatosa B, van Dongen J, Roszkowski K, Mikoluc B, et al. and we will help you at every step:
Disseminated bacillus Calmette-Guerin infection and immunodeficiency.
• We accept pre-submission inquiries
Emerg Infect Dis. 2007;13(5):799–801.
79. Rules and Procedures. https://helseforskning.etikkom.no/ikbViewer/page/ • Our selector tool helps you to find the most relevant journal
reglerogrutiner/loverogregler?p_dim=34770&_ikbLanguageCode=us. • We provide round the clock customer support
Accessed 23 June 2014.
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research

Submit your manuscript at

www.biomedcentral.com/submit