Do Infant Probiotics Prevent Eczema? The Evidence Gap

The promising eczema numbers shrink when study quality improves

Infant probiotics have an appealing pitch: alter the early gut microbiome, teach the immune system some manners, and reduce eczema before it starts. Several meta-analyses report lower eczema risk. Yet the best recent review finds that the apparent benefit disappears when it restricts the analysis to the few trials at low risk of bias.

That does not prove probiotics cannot prevent eczema. It does mean the evidence does not support treating every probiotic product, every infant, and every feeding pattern as equivalent. They plainly are not.

A 2025 Cochrane review included 24 studies and 7,077 mother-infant pairs. It rated the evidence for preventing eczema as low certainty, and the evidence for food allergy and synbiotic-related eczema outcomes as very low certainty. In a prespecified sensitivity analysis limited to four low-risk-of-bias studies involving 892 participants, the eczema estimate was a relative risk of 0.86 (95% CI 0.69–1.07). That confidence interval includes no effect, so the result was not statistically significant.

The Cochrane authors found methodological concerns in 20 of the 24 included studies, including unclear randomization, missing outcome data, and missing prespecified outcome intentions. A large collection of trials does not automatically become reliable by being large; it can also become a large collection of the same avoidable problems.

Why earlier meta-analyses look more optimistic

Earlier and narrower syntheses report more favorable estimates. A 2021 meta-analysis by Sun et al. combined nine randomized trials with 2,093 children younger than 3 years and reported a pooled relative risk of 0.60 for eczema with mixed Lactobacillus and Bifidobacterium regimens. In subgroup comparisons across trials, protocols that began during pregnancy had an estimated relative risk of 0.59, versus 0.63 for postnatal-only protocols. Those two estimates came from indirect comparisons, not a randomized head-to-head test of timing.

A 2025 review of infant-type Bifidobacteria reported a relative risk of 0.78 for eczema. It focused on B. infantis, B. bifidum, B. breve, and B. longum subsp. longum, strains selected partly for their ability to use human milk oligosaccharides, or HMOs.

These results should not be averaged into a comforting middle number. The reviews ask related but different questions, include different interventions, and draw on studies with uneven quality. The Cochrane sensitivity analysis is especially informative because it tests whether the overall result survives after excluding trials with the clearest methodological concerns. It did not.

Strain choice also matters. In the Sun analysis, the combination of LGG and B. longum did not reduce eczema incidence (p=0.18), while other Lactobacillus-Bifidobacterium combinations did. "Probiotic" is a category on a store shelf, not a single treatment. An eczema-prevention claim that does not name the strains tells parents very little.

Formula-fed infants are a separate evidence problem

Formula-fed infants often get swept into the same conversation, although their microbial environment differs from that of breastfed infants. A 2024 EDEN mother-child cohort study followed about 1,243 formula-fed infants to age 8 and examined formula type at 4 months in relation to allergy multimorbidity clusters.

Probiotic-enriched formula was associated with lower odds of membership in the "allergies without asthma" cluster, with an odds ratio of about 0.63. But this was an observational association, not evidence that probiotic formula caused lower risk. The same analysis found that partially hydrolyzed formula was associated with markedly higher odds of that cluster, about 2.73. That pattern fits confounding by indication: clinicians may steer infants with a family history of allergy or early symptoms toward partially hydrolyzed formula, while parents who choose probiotic formula may differ in other health-related ways.

The study therefore cannot establish an independent protective effect from probiotic formula. It also cannot prove the formula caused harm. Formula choice was not randomized, and the opposing associations warn against reading either estimate as a treatment effect.

The HMO hypothesis makes biological sense—but remains a hypothesis

One plausible explanation for differing results by feeding mode involves HMOs, complex carbohydrates in human milk. Some infant-type Bifidobacteria are adapted to use HMOs as an energy source. In a study of breastfed infants, early B. infantis supplementation produced persistent colonization through 1 year of age. That finding supports colonization durability in breastfed infants; it does not show that the same result occurs in formula-fed infants.

The missing direct test is obvious: researchers have not compared colonization over time in breastfed and formula-fed infants given identical doses of the same strain. Without that study, the claim that conventional formula provides too little substrate for durable colonization remains biologically coherent but unproven.

There is relevant proof of concept. A 2025 randomized clinical trial found that adding 2'-fucosyllactose, an abundant HMO, to formula enriched Bifidobacterium populations in formula-fed infants. That is a microbiome outcome, not an eczema-prevention result. Likewise, a randomized trial in 540 European infants found that synbiotic formula containing Limosilactobacillus fermentum CECT5716 and galacto-oligosaccharides changed gut microbiota profiles compared with control formula at 4 months. The synbiotic formula trial supports the idea that formula can shift microbial ecology, but it does not establish that those shifts prevent eczema.

A small 2025 pilot study of 23 infants also found distinct microbiota patterns between exclusively breastfed and formula-fed groups over six months. Its size prevents firm clinical conclusions, but it reinforces the basic point: feeding mode is not background scenery in a probiotic trial.

What parents can reasonably conclude

For low-risk infants in the general population, routine probiotic supplementation for allergy prevention is not supported by current evidence. The Cochrane review found low-certainty evidence for eczema and no statistically significant benefit for asthma, allergic rhinitis, or food allergy in the reported analyses.

For infants with a strong family history of eczema or allergy, there is a narrower, uncertain signal worth discussing with a pediatrician—particularly for named mixed-strain or infant-type Bifidobacterium interventions and protocols that begin during pregnancy. But the low-bias Cochrane analysis did not confirm eczema prevention, so this is not a settled preventive treatment.

For formula-fed high-risk infants, a synbiotic approach that pairs an HMO-type prebiotic such as 2'-FL with a specified strain is more biologically plausible than probiotic-only formula. Plausibility is not clinical proof. No low-bias randomized trial has shown that this strategy prevents eczema, and no direct study has shown that it recreates the durable B. infantis colonization observed in breastfed infants.

The practical standard is simple: ask for the strain, the population studied, the comparator, and the actual outcome. If a product offers only the word "probiotic," it has skipped the details that determine whether the claim means anything.

Bottom line: Infant probiotics may reduce eczema risk in specific settings, but current low-certainty evidence does not justify routine use. The biggest unanswered question is whether a properly tested synbiotic formula can deliver a real eczema benefit for formula-fed infants.