This study was conducted according to the principles expressed in the Declaration of Helsinki. The study was approved by the research committee of the Ministry of Health in Guinea-Bissau and for European Centers by the Danish Central Ethical Committee. All mothers provided written informed consent for the collection of samples of their children.
The two large randomized trials were conducted by the Bandim Health Project (BHP) in Bissau, Guinea-Bissau. The BHP covers a population of around 100,000 in 6 districts in the urban area by a demographic surveillance system. All houses are visited every month to register new pregnancies and births. Once a newborn is identified, the child is followed with a home visit every third month. There is a national hospital (NH) with a maternity ward in the capital and three health centers in the study area, one of them with a maternity ward. Around 60–70% of the mothers from the study area deliver at the NH or at the health center maternity ward. Since 2002 all LBW infants born at the NH have been followed for one year, also infants outside the study area. In trial 1, all LBW infants born at the NH or in the study area were invited to participate in a trial studying the effect of receiving early BCG vaccination on mortality and cytokine levels in LBW infants. In trial 2, NBW infants from the study area were invited to participate in a trial studying the effect of vitamin A supplementation at birth on mortality and cytokine levels. These two EU funded trials were carried out within a collaborative network of European and Guinea-Bissau institutions. The trials were registered at www.clinicaltrials.gov
, registration number, NCT00146302 and NCT00168597.
Enrolment into the trials
Newborns were enrolled throughout 2004 (). Mothers were informed of the purpose of the study. If a mother of a LBW infant accepted to participate, she drew a randomization number indicating whether the children would be BCG vaccinated early or not. If a mother of a NBW infant accepted to participate, she drew a randomization number indicating whether the children would receive vitamin A supplementation (50,000 IU) or placebo. At the time of enrolment, children were examined clinically. The infants were weighed and length, head-circumference and mid-upper-arm-circumference (MUAC) were measured. Newborns that were overtly ill were not enrolled but referred to treatment.
An overview of the two trials, the periods with missing OPV, and the periods with ongoing subgroup studies in Guinea-Bissau 2004.
LBW infants who were randomized to not receiving early BCG vaccination were not included in the present study since we aimed to look at the effect of OPV on the immune response to BCG given at the same time. Hence, the LBW population only includes children who were randomized to BCG at birth. NBW children who had received vitamin A supplementation at birth were not included in the present study as this could have affected their response to OPV and BCG. Hence, the NBW population only includes children who were randomized to placebo at birth.
BCG (0.05 ml; Statens Serum Institut, Copenhagen, Denmark) was provided by the BHP and given intradermally in the upper left deltoid region. OPV was supplied through the national immunization programme and given orally. Three types of OPV were used in Guinea-Bissau, produced by Novartis, Sanofi Pasteur, and GlaxoSmithKline. However, OPV was missing during periods of 2004. As part of our trial routines, we noted at the inclusion form whether the children received OPV or not. Based on the distribution of children who had not received OPV, OPV was missing at one or more inclusion sites 1) from the beginning of February 2004 to the beginning of June 2004, 2) again in the end of June 2004, and 3) from the end of October 2004 to November 2004 ().
Follow-up for in vitro PPD response
We aimed to enroll 200 children from trial 1 and 400 children from trial 2 in a subgroup study of cytokine production. The children were enrolled between April and November 2004 (). We attempted to visit the children at 6 weeks of age. A blood sample was collected by finger prick and analyzed for in vitro
cytokine responses. Whole blood cultures were done as previously described 
. Briefly, heparinized finger prick blood was diluted in RPMI 1640 medium (Invitrogen, Breda, The Netherlands) to a final concentration of 1 in 10 and culture in a total volume of 200 µl in 96-well U plate (Nunc, Roskilde, Denmark) with the presence of purified protein derivative of Mycobacterium tuberculosis
(PPD, 10 µg/ml; Statens Serum Institut, Denmark, Copenhagen). Supernatants were collected on day 3 and stored at −80°C.
Measurement of cytokine levels
IL-5, IL-10, IL-13, IFN-γ and TNF-α from day 3 supernatants were analyzed simultaneously using commercial Luminex cytokine kit and buffer reagent kit (BioSource, Camarillo, CA, USA) and run on a Luminex-100 cytometer (Luminex Corporation, Austin, TX, USA), equipped with StarStation software (Applied Cytometry Systems, Dinnington, UK). The assay was performed with slight modification from the manufacturer's recommendation. Briefly, assays were done in a 96-well U plate at room temperature. Mixes of beads were incubated for 2 hours with a standard, samples, or blank in a final volume of 50 µl for 2 hours under continuous shaking. Beads were washed twice and incubated with cocktail of biotinylated antibodies (25 µl/well) for 1 hour. After removal of excess biotinylated antibodies, Streptavidin-PE was added for 30 minutes. The plate was then washed and analyzed. The lower detection limit of the assays was 3 pg/ml, 5 pg/ml, 10 pg/ml, 5 pg/ml and 10 pg/ml for IL-5, IL-10, IL-13, IFN-γ and TNF-α, respectively. Samples with concentrations below the detection limit were given the value of this threshold.
Follow-up for BCG scar and in vivo PPD response
All LBW children were visited at home and examined for BCG scar and in vivo
PPD response at 2 and 6 months of age. In four of six study area districts NBW children were examined for BCG scar and in vivo
PPD response when they reached 2 and 6 months of age. At the home visits a trained nurse documented health status and vaccination status. She measured the size of the scar following BCG vaccination. Subsequently, 0.1 ml of Tuberculin PPD RT23 (Statens Serum Institut, Copenhagen, Denmark) was injected intradermally on the ventral side of the forearm. Between 48 and 72 hours later, the child was visited again, and the induration was measured by a trained field worker using track-ball technique 
. Size of scar and induration was defined as the average of the height and the width measured to nearest millimeter with a transparent ruler. Children with a measurable scar were categorized as “scar-positive”. Children with a mean diameter of the induration ≥2 mm were categorized as “PPD-positive”. Children who had taken chloroquine within seven days prior to the visit were excluded from the PPD analysis, since chloroquine treatment is known to suppress PPD reaction 
. These children were retained in the analysis of scar reaction since we considered it unlikely that current chloroquine treatment would have affected the scar formation. Children who had received BCG vaccination less than 30 days prior to the visit or with known exposure to tuberculosis in the household within the previous 6 months (information only available for NBW population) were excluded. Children with PPD reactions >10 mm were referred to a medical doctor to be further examined for tuberculosis (TB).
All statistical analysis was performed in Stata 10.0.
In vitro PPD response
Cytokine levels were not normally distributed. The crude cytokine levels were first compared in a univariate analysis using non-parametric Mann-Whitney U
tests, and the results were provided in the figures of the distribution of cytokine levels in the different groups ( and ). Subsequently, the frequency of low and high cytokine producers in BCG and BCG+OPV recipients was compared in Poisson regression model with robust variance estimates 
using the median cytokine level in the total population as cut off for low and high responder. Thus, the relative measure is a prevalence ratio (PR). The study population was divided into 4 groups according to the birth weight and whether OPV was given together with BCG at birth (). As the periods with lack of OPV fell primarily in the dry season (from December to May), all children who had received BCG+OPV had been vaccinated in the rainy season (June to November). The two LBW groups were similar with respect to age and sex. However, MUAC at enrolment was larger in the BCG+OPV group compared with the BCG group, and they had received more OPV and more diphtheria-tetanus-pertussis (DTP) vaccines (recommended at age 6, 10 and 14 weeks) between enrolment and blood sampling (). The two NBW groups were similar with respect to age and MUAC at enrolment (). However, they were significantly different with regard to sex, the NBW BCG+OPV group having fewer boys. The two groups also differed significantly with regard to vaccinations received between enrolment and blood sampling (more OPV and less DTP in the BCG+OPV group). All further analyses were adjusted for age, sex, month of enrolment, MUAC at enrolment, and vaccinations between enrolment and blood sampling.
PPD stimulated IFN-γ (A), IL-5 (B), IL-13 (C), TNF-α (D) and IL-10 (E) production in low birth weight BCG (LBW BCG) and BCG+OPV vaccinated (LBW BCG+OPV) infants.
PPD stimulated IFN-γ (A), IL-5 (B), IL-13 (C), TNF-α (D) and IL-10 (E) production in normal birth weight BCG (NBW BCG) and BCG+OPV vaccinated (NBW BCG+OPV) infants.
Description of the cytokine study population.
BCG scar and in vivo PPD response
PPD response (yes/no) and scar response (yes/no) were analyzed as prevalence data using a Poisson regression model with robust variance estimates providing prevalence ratios (PR) 
. Children with a PPD response above 15 mm were excluded from the analyses as they had potentially been exposed to TB. There were no such children among LBW children. Among NBW children one boy (BCG+OPV) had a PPD response above 15 mm at age 2 months and was excluded from further analyses. At age 6 months, one boy (BCG+OPV) and one girl (BCG), had a PPD response above 15 mm and were excluded from the 6 month-analysis. As in the cytokine analysis, there was a clear overweight of children who had received BCG+OPV in the rainy season, though there was more variation because the follow-up for PPD/scar had been conducted throughout 2004 (). Among LBW infants, the BCG+OPV recipients only differed from the BCG only recipients with regard to season (). Among NBW infants, the two groups also differed with regard to MUAC at enrolment, and number of OPV received between enrolment and the PPD/Scar assessment (). We adjusted the analyses for sex, season (the larger variation in this analysis compared with the cytokine analysis allowed us to control for season rather than month), MUAC, and subsequent vaccinations. As in previous studies of PPD responses, the reading of the PPD reaction differed significantly by the assistant doing the reading and the analyses of PPD responses were therefore furthermore adjusted for assistant. Among PPD responders and/or scar reactors, we analyzed the size of the reaction in linear regression models controlling for the same potential confounders.
Description of the scar and PPD study population.