SummaryBackgroundTrials done in infants with low birthweight in west Africa suggest that BCG vaccination reduces all-cause mortality in the neonatal period, probably because of heterologous protection against non-tuberculous infections. This study investigated whether BCG alters all-cause infectious disease morbidity in healthy infants in a different high-mortality setting, and explored whether the changes are mediated via trained innate immunity.MethodsThis was an investigator-blind, randomised, controlled trial done at one hospital in Entebbe, Uganda. Infants who were born unwell (ie, those who were not well enough to be discharged directly home from the labour ward because they required medical intervention), with major congenital malformations, to mothers with HIV, into families with known or suspected tuberculosis, or for whom cord blood samples could not be taken, were excluded from the study. Any other infant well enough to be discharged directly from the labour ward was eligible for inclusion, with no limitation on gestational age or birthweight. Participants were recruited at birth and randomly assigned (1:1) to receive standard dose BCG 1331 (BCG-Danish) on the day of birth or at age 6 weeks (computer-generated randomisation, block sizes of 24, stratified by sex). Investigators and clinicians were masked to group assignment; parents were not masked. Participants were clinically followed up to age 10 weeks and contributed blood samples to one of three immunological substudies. The primary clinical outcome was physician-diagnosed non-tuberculous infectious disease incidence. Primary immunological outcomes were histone trimethylation at the promoter region of TNF, IL6, and IL1B; ex-vivo production of TNF, IL-6, IL-1β, IL-10, and IFNγ after heterologous stimulation; and transferrin saturation and hepcidin levels. All outcomes were analysed in the modified intention-to-treat population of all randomly assigned participants except those whose for whom consent was withdrawn. This trial is registered with the International Standard Randomised Controlled Trial Number registry (#59683017).FindingsBetween Sept 25, 2014, and July 31, 2015, 560 participants were enrolled and randomly assigned to receive BCG at birth (n=280) or age 6 weeks (n=280). 12 participants assigned to receive BCG at birth and 11 participants assigned to receive BCG at age 6 weeks were withdrawn from the study by their parents shortly after randomisation and were not included in analyses. During the first 6 weeks of life before the infants in the delayed vaccination group received BCG vaccination, physician-diagnosed non-tuberculous infectious disease incidence was lower in infants in the BCG at birth group than in the delayed group (98 presentations in the BCG at birth group vs 129 in the delayed BCG group; hazard ratio [HR] 0·71 [95% CI 0·53–0·95], p=0·023). After BCG in the delayed group (ie, during the age 6–10 weeks follow-up), there was no significant difference in non-tuberculous infectious disease incidence between the groups (88 presentations vs 76 presentations; HR 1·10 [0·87–1·40], p=0·62). BCG at birth inhibited the increase in histone trimethylation at the TNF promoter in peripheral blood mononuclear cells occurring in the first 6 weeks of life. H3K4me3 geometric mean fold-increases were 3·1 times lower at the TNF promoter (p=0·018), 2·5 times lower at the IL6 promoter (p=0·20), and 3·1 times lower at the IL1B promoter (p=0·082) and H3K9me3 geometric mean fold-increases were 8·9 times lower at the TNF promoter (p=0·0046), 1·2 times lower at the IL6 promoter (p=0·75), and 4·6 times lower at the IL1B promoter (p=0·068), in BCG-vaccinated (BCG at birth group) versus BCG-naive (delayed BCG group) infants. No clear effect of BCG on ex-vivo production of TNF, IL-6, IL-1β, IL-10, and IFNγ after heterologous stimulation, or transferrin saturation and hepcidin concentration, was detected (geometric mean ratios between 0·68 and 1·68; p≥0·038 for all comparisons).InterpretationBCG vaccination protects against non-tuberculous infectious disease during the neonatal period, in addition to having tuberculosis-specific effects. Prioritisation of BCG on the first day of life in high-mortality settings might have significant public-health benefits through reductions in all-cause infectious morbidity and mortality.FundingWellcome Trust.TranslationsFor the Luganda and Swahili translations of the abstract see Supplementary Materials section.