Absence of Evidence of Beneficial Effects of Human Milk-Based Fortifier: A Bayesian Model-Averaged Meta-Analysis.

Aim: Human milk-based fortifiers (HMBFs) have been adopted in neonatal care despite limited efficacy data. Our objective was to conduct a Bayesian re-analysis of the current evidence on the protective effect of HMBF against necrotising enterocolitis (NEC, stage II-III).

Methods: We searched for randomised controlled trials (RCTs) comparing HMBF versus bovine milk-based fortifier (BMBF) in infants with gestational age < 32 weeks or birthweight < 1500 g. Bayesian model-averaged (BMA) meta-analysis was used to calculate Bayes factors (BFs). The BF₁₀ is the ratio of the probability of the data under the alternative hypothesis (H₁, presence of effect) over the probability of the data under the null hypothesis (H₀, absence of effect).

Results: We included 3 RCTs (405 infants). The BMA analysis did not show conclusive evidence in favour of H₁ (BF₁₀ > 3) or in favour of H₀ (BF₁₀ < 1/3) for NEC (BF₁₀ = 0.86), mortality (BF₁₀ = 1.07), late onset sepsis (BF₁₀ = 0.69), bronchopulmonary dysplasia (BPD, BF₁₀ = 0.92), or severe retinopathy of prematurity (ROP, BF₁₀ = 0.75).

Conclusions: The extant corpus of RCTs does not provide sufficient evidence to either confirm or exclude a potential effect of HMBF compared with BMBF on the risk of NEC, mortality, sepsis, BPD, or severe ROP.