Probiotics

What probiotics are

Probiotics are live microorganisms that are supposed to have a health benefit in sufficient amounts. In commercial products these are almost always lactobacilli and bifidobacteria, plus the yeast Saccharomyces boulardii. They are measured in colony-forming units, CFU for short.

The decisive point is the third level of naming: genus, species and strain. Lactobacillus rhamnosus GG is something different from any Lactobacillus rhamnosus. The studies are run with strains, advertising talks about genera. Most of the confusion arises in this gap.

How they are supposed to work

Three mechanisms are discussed: competition against pathogens, strengthening of the gut barrier, signals to the mucosal immune system. All three are well established in cell culture and animal models.

In humans, the picture becomes more nuanced. When a research group sampled the intestinal mucosa endoscopically instead of only the stool, colonization by an eleven-strain preparation turned out completely different depending on the person, the gut segment and the strain — and could not be read from the stool. This removes the idea that a capsule permanently remodels the gut flora: a review of 45 studies in healthy people found only a temporary enrichment of the bacteria taken. Probiotics pass through. What they do, they do in passing.

Where the evidence is densest

Antibiotic-associated diarrhea is the field with the best data. A Cochrane review of 33 studies with 6,352 children found 8 versus 19 percent, number needed to treat 9. For Clostridioides difficile diarrhea, a second review of 31 studies with 8,672 participants arrives at 1.5 versus 4.0 percent — carried, however, by studies with a baseline risk above 5 percent, where it was 3.1 versus 11.6 percent; below that, no difference was seen. This is a statement about hospital patients on broad-spectrum antibiotics, not about a week of amoxicillin at home.

In irritable bowel syndrome, a closer look pays off: a systematic review evaluated 32 papers on 10 single strains separately. Core symptoms were improved by Bifidobacterium longum 35624, Lactobacillus rhamnosus GG, Lactiplantibacillus plantarum 299v, Saccharomyces cerevisiae CNCM I-3856 and Bacillus coagulans Unique IS2. Escherichia coli Nissle 1917, Lactobacillus gasseri BNR17 and Lactobacillus casei Shirota showed no effect. The same sharp distinction in constipation: Bifidobacterium lactis increased stool frequency by 1.5 bowel movements per week, L. casei Shirota did not.

The uncomfortable part

For acute gastrointestinal infection in young children, there are two large studies. Lactobacillus rhamnosus GG in 971 children: 11.8 versus 12.6 percent with a moderate to severe course; a two-strain combination in 886 children: 26.1 versus 24.7 percent. The same strain that helps with antibiotic-associated diarrhea does nothing here — that is the core of indication specificity.

Most food for thought comes from an invasive study on recovery after antibiotics. The probiotic colonized the mucosa particularly well — and in doing so delayed the return of the native flora, which remained permanently incomplete. With waiting, it went faster. The recommendation to take a capsule after every course of antibiotics faces a serious contrary finding here.

What the studies show

Cochrane 2019: antibiotic-associated diarrhea in children

Meta-analysis of 33 randomized studies with 6,352 children on antibiotics, follow-up 5 days to 12 weeks. Primary endpoint diarrhea frequency: 8 versus 19 percent, relative risk 0.45, number needed to treat 9, moderate certainty. In 19 studies, 1 to 46 percent were missing at follow-up; if all dropouts are assumed to have had the worst outcome, the result still holds (relative risk 0.61). Of all subgroup analyses, only the one by dose held up. 20 of the 33 studies had a high risk of bias.

Cochrane 2017: Clostridioides difficile diarrhea

Review of 39 studies with 9,955 participants, 31 studies with 8,672 participants evaluable: 1.5 versus 4.0 percent, relative risk 0.40, number needed to treat 42. The effect was independent of age, setting, type and dose — with one exception found post hoc: at a baseline risk above 5 percent, the figures were 3.1 versus 11.6 percent, below that no difference. Detection in the stool did not change: what is prevented is the disease, not the colonization.

PLACIDE 2013: the largest study was negative

Randomized, double-blind, placebo-controlled, in inpatients aged 65 and over on antibiotics. Of 17,420 screened, 1,493 and 1,488 were randomized, 1,470 and 1,471 evaluated; a multi-strain preparation with 6 times 10 to the 10th organisms daily was given for 21 days. Diarrhea within 8 weeks: 10.8 versus 10.4 percent; C. difficile diarrhea within 12 weeks: 0.8 versus 1.2 percent. Both primary endpoints missed.

Cell 2018: recovery after antibiotics

Study with endoscopic sampling from the intestinal mucosa instead of mere stool analysis. After a course of antibiotics, three approaches were compared: waiting, a multi-strain probiotic and the retransplantation of the person's own, previously frozen stool. Under the probiotic, the native flora returned with a delay and remained permanently incomplete; the transcriptome of the gut wall also recovered more slowly than with waiting. The stool retransplantation restored the baseline state within days.

Where the data stop

Skin is the clearest case. A Cochrane review of 39 studies with 2,599 participants found, for eczema symptoms rated by patients and parents, evaluated in 13 studies with 754 participants, a difference of 0.44 points on a scale from 0 to 20, with a confidence interval that includes zero; quality of life remained unchanged. The authors consider further studies with similar strains futile.

On mood, the analyses contradict each other: one meta-analysis of 13 studies with 710 patients with diagnosed depression found a standardized mean difference of 0.38 in favor of probiotics, another of 13 studies with 893 participants found no evidence of a consistent effect. An open field, not a proven benefit.

In irritable bowel syndrome, too, the figures are less stable than the recommendation suggests: a meta-analysis on multi-strain preparations found the symptom scale lowered by 43.66 points, with a heterogeneity of 99 percent and no difference in abdominal pain and disease-specific quality of life.

Status, approval and legal

Preparations with microorganisms are legally foods, not medicines, and do not require approval; there are no statutory maximum amounts. New strains, by contrast, require novel food authorization: Clostridium butyricum since 2014 with at most 1.35 times 10 to the 8th CFU daily, pasteurized, that is, killed Akkermansia muciniphila since 2022. For advertising, the situation is tight: in the EU no health claim for probiotics is authorized, and all applications to EFSA received negative opinions. The term “probiotic” itself counts as such a claim; the German Federal Court of Justice ruled this for baby food on February 26, 2014, and since then German packages say “with special bacterial cultures”. Only a related claim is authorized: live cultures in yogurt improve lactose digestion. Therapeutically, the substance is anchored through the German S3 guideline on irritable bowel syndrome of June 2021; in 2020 the American Gastroenterological Association (AGA) conditionally recommends only the prevention of necrotizing enterocolitis in premature infants and of C. difficile infection during antibiotic treatment, as well as a multi-strain preparation in pouchitis.

Safety

In people who are not immunocompromised, probiotics are well tolerated in controlled studies. In the Cochrane review on C. difficile diarrhea, adverse events across 32 studies with 8,305 participants were even less frequent than under control; the most common were abdominal cramps, nausea, soft stools and bloating. The rare but real risk is infection by the administered microbe itself. A systematic review collected 93 patients with probiotic-associated infections between 1976 and 2018: the most common complication was fungemia with 35 cases, mainly involving Saccharomyces with 47 and Lactobacillus with 26, overall mortality 19.6 percent. Those affected were predominantly seriously ill hospital patients at the extremes of age: 33 cases over 60 years, 25 under one year. In 2023 the FDA issued a warning after a premature infant weighing under 1,000 grams died of a bloodstream infection whose pathogen genetically matched the Bifidobacterium longum from the product. Live microbes are therefore not an option for immunosuppressed people after organ or bone marrow transplantation, for the seriously ill and in acute gastroenteritis; for pregnant women, breastfeeding women and children, the decision belongs in the hands of a physician.

Sources

• Guo et al., Cochrane Database of Systematic Reviews 2019
• Goldenberg et al., Cochrane Database of Systematic Reviews 2017
• Allen et al., The Lancet 2013 (PLACIDE)
• Suez et al., Cell 2018
• Zmora et al., Cell 2018
• Maslennikov et al., Journal of Clinical Medicine 2026
• Zhao et al., Cochrane Database of Systematic Reviews 2022
• Makrgeorgou et al., Cochrane Database of Systematic Reviews 2018
• Costa et al., BMC Complementary and Alternative Medicine 2018
• Verbraucherzentrale (German consumer advice center), food supplements with microorganisms