BackgroundNecrotising enterocolitis (NEC) and late-onset sepsis remain important causes of death and morbidity in preterm babies. Probiotic administration might strengthen intestinal barrier function and provide protection; this is supported by published meta-analyses, but there is a lack of large well-designed trials.ObjectiveTo test the use of the probioticBifidobacterium brevestrain BBG-001 to prevent NEC, late-onset sepsis and death in preterm babies while monitoring probiotic colonisation of participants.DesignDouble-blind, randomised, placebo-controlled trial.SettingRecruitment was carried out in 24 hospitals, and the randomisation programme used a minimisation algorithm. Parents, clinicians and outcome assessors were blinded to the allocation.ParticipantsBabies born between 23 and 30 weeks’ gestation and randomised within 48 hours of birth. Exclusions included life-threatening or any gastrointestinal malformation detected within 48 hours of birth and no realistic chance of survival.InterventionsActive intervention: 1 ml ofB. breveBBG-001 in one-eighth-strength infant formula Neocate®(Nutricia Ltd, Trowbridge, UK), (6.7 × 107to 6.7 × 109colony-forming units) per dose administered enterally. Placebo: 1 ml of one-eighth-strength infant formula Neocate. Started as soon as practicable and continued daily until 36 weeks’ postmenstrual age.Main outcome measuresPrimary outcomes were an episode of bloodstream infection, with any organism other than a skin commensal, in any baby between 72 hours and 46 weeks’ postmenstrual age; an episode of NEC Bell stage ≥ 2 in any baby; and death before discharge from hospital. Secondary outcomes included stool colonisation withB. breve.ResultsIn total, 654 babies were allocated to receive probiotic and 661 to receive placebo over 37 months from July 2010. Five babies were withdrawn; 650 babies from the probiotic group and 660 from the placebo group were included in the primary analysis. Baseline characteristics were well balanced. There was no evidence of benefit for the primary outcomes {sepsis: 11.2% vs. 11.7% [adjusted relative risk (RR) 0.97, 95% confidence interval (CI) 0.73 to 1.29]; NEC Bell stage ≥ 2: 9.4% vs. 10.0% [adjusted RR 0.93, 95% CI 0.68 to 1.27]; and death: 8.3% vs. 8.5% [adjusted RR 0.93, 95% CI 0.67 to 1.30]}.B. brevecolonisation status was available for 1186 (94%) survivors at 2 weeks’ postnatal age, of whom 724 (61%) were positive: 85% of the probiotic group and 37% of the placebo group. There were no differences for subgroup analyses by minimisation criteria and by stool colonisation withB. breveat 2 weeks. No harms associated with the interventions were reported.LimitationsCross-colonisation of the placebo arm could have reduced statistical power and confounded results; analyses suggest that this did not happen.ConclusionsThis is the largest trial to date of a probiotic intervention. It shows no evidence of benefit and does not support routine use of probiotics for preterm infants.Future work recommendationsThe increasing understanding of the pathogenesis of NEC and sepsis will inform the choice of probiotics for testing and better define the target population. Future Phase III trials should incorporate monitoring of the quality and viability of the intervention and colonisation rates of participants; cluster design should be considered.Trial registrationCurrent Controlled Trials ISRCTN05511098 and EudraCT 2006-003445-17.FundingThis project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full inHealth Technology Assessment; Vol. 20, No. 66. See the NIHR Journals Library website for further project information.