Twenty RCT studies were included (3,816 participants). The Jadad score for individual trials ranged from 3 to 5.
Risk of necrotizing enterocolitis was statistically significantly lower for probiotic groups versus placebo groups (RR 0.33, 95% CI 0.24 to 0.46; 20 trials, Ι²=0%). The risk remained significantly lower in subgroup analyses by probiotic agent: bifidobacteria (RR 0.30, 95% CI 0.16 to 0.58; eight trials; Ι²=0%), lactobacillus (RR 0.37, 95% CI 0.19 to 0.73; four trials; Ι²=0%), and both bifidobacteria and lactobacillus (RR 0.33, 95% CI 0.19 to 0.58; six trials; Ι²=0%).
Risk of mortality was statistically significantly lower for probiotic groups versus placebo groups (RR 0.56, 95% CI 0.43 to 0.73; 13 trials; Ι²=0%). The risk remained significantly lower in subgroup analyses for lactobacillus (RR 0.61, 95% CI 0.38 to 0.97; four trials; Ι²=0%), and both bifidobacteria and lactobacillus (RR 0.47, 95 CI 0.26 to 0.87; five trials; Ι²=49%). The risk of mortality remained lower for bifidobacteria versus placebo, but was not statistically significant (RR 0.74, 95% CI 0.18 to 2.97; three trials; Ι²=0%).
No statistically significant differences for risk of culture-positive sepsis were found between probiotic groups and placebo groups (RR 0.90, 95% CI 0.71 to 1.15; 14 trials; Ι²=56%). Differences in risk for sepsis remained non-significant in the three subgroup analyses (reported in paper).
Meta-regression analysis suggested that major sources of heterogeneity between studies might be type of probiotic agent and eligibility criteria for inclusion of pre-term very low-birth-weight infants. No evidence was found for publication bias (reported in paper).