Probiotics moved older adults' cognitive scores. Half of trials never reported harms

"Gut-brain" supplements are sold on a general promise: feed the microbiome, sharpen the mind. Two papers published this month test the two halves of that proposition separately — does it work, and is it safe — and neither answer is as broad as the marketing [s1][s2].

The efficacy question

The first is a meta-analysis of randomised controlled trials examining probiotics and cognitive function across the human lifespan, drawing on four databases searched from inception to December 2023 and pooling 21 RCTs [s1].

Its organising finding is that the effect depends on age. In older age groups, probiotics were associated with significant improvements in overall cognitive performance (SMD 0.40, 95% CI 0.13 to 0.67, P = 0.004), processing speed (SMD 0.37, 95% CI 0.11 to 0.63, P = 0.006), memory (SMD 0.51, 95% CI 0.25 to 0.78, P = 0.0002) and spatial ability (SMD 0.35, 95% CI 0.09 to 0.62, P = 0.008) [s1].

Outside that group the picture is narrower. In infants and children, probiotics improved executive function (SMD 0.30, 95% CI 0.04 to 0.56, P = 0.03) [s1]. In young adults, the effect appeared mainly in verbal ability (SMD 0.43, 95% CI 0.15 to 0.71, p = 0.003) [s1].

The authors state that the best results were observed after 12 weeks of daily intake of approximately 2 × 10¹⁰ colony-forming units [s1].

Two cautions belong with those numbers. Standardised mean differences in the 0.3 to 0.5 range are small-to-moderate effects on test scores; they are not evidence of prevented dementia, and the meta-analysis does not report any clinical endpoint. And 21 trials spread across infants, children, young adults and older adults leaves few studies per age band, which is a thin base for age-specific conclusions.

The safety question

The second paper takes the strain that is arguably the most widely consumed of all and asks a question the supplement literature rarely asks directly: what adverse events have actually been recorded [s2].

The target was Bifidobacterium animalis subsp. lactis (BB-12). The authors note that probiotics are widely used in everyday foods and supplements but that very few studies have thoroughly evaluated safety [s2]. Two independent reviewers screened for double-blinded, randomised or randomly allocated, placebo-controlled trials of BB-12 published in English. They identified 181 studies, of which 35 qualified for inclusion [s2]. The authors then contacted the authors of all 35 to assemble an individual patient data meta-analysis [s2].

In the conventional meta-analysis, adverse event rates were 15.2% in the probiotic groups and 14% in the placebo groups — no statistical difference (odds ratio 1.10, 95% CI 0.88 to 1.36, p = .40) [s2]. The individual patient data analysis found statistically similar unadjusted rates (27% versus 25%) and adjusted rates (43% versus 41%), with a relative risk of 1.05 (95% CI 0.94 to 1.15) [s2].

Both approaches point the same way, and the authors conclude the data support the safety of wide usage of BB-12 [s2].

The finding underneath the finding

The paper's most transferable result is not about BB-12 at all. Half of the randomised controlled trials analysed did not report adverse event data, which the authors identify as a major limitation of the published literature [s2].

That is the number to carry away. A reassuring safety meta-analysis assembled from trials where half report no harm data is describing what was recorded, not what occurred. The finding is best read as an absence of a detected signal in the trials that bothered to look — which is meaningful, and is not the same as an established safety profile.

It also reframes the efficacy paper. If half of probiotic RCTs omit adverse event reporting, the cognitive trials pooled in the first study are drawn from the same reporting culture [s1][s2].

Limits

The safety review covers one strain. Probiotic products routinely contain multiple strains at doses and combinations that have never been trialled together, and safety findings for BB-12 do not transfer to them. The cognition meta-analysis covers a mix of strains, doses and durations, which is part of why heterogeneity in this literature is high and why the 12-week, 2 × 10¹⁰ CFU pattern the authors describe is a summary observation rather than a tested protocol [s1].

Neither paper addresses immunocompromised or critically ill populations, where the risk calculus for live-organism supplements differs and where case reports of invasive infection exist in the wider literature.

This article does not recommend for or against taking any supplement.

What to watch

Whether adverse event reporting becomes a condition of publication for probiotic trials. The efficacy literature is growing faster than the safety literature, and the 50% non-reporting figure [s2] is the clearest measurement anyone has published of that gap.

Sources

• Effects of probiotics on cognitive function across the human lifespan: a meta-analysis — European Journal of Clinical Nutrition, 2025-09-23
• Safety of One of the Most Commonly Used Probiotic Strains: Systematic Review and Meta-analysis of Reported Adverse Events — Probiotics and Antimicrobial Proteins, 2025-09-02