Probiotic capsules: which ones have trial evidence, what they can do, and when to skip them

Two large pieces of evidence, both about probiotics and antibiotics. In a Cochrane pool of 33 trials in children, 8% of those given a probiotic got antibiotic diarrhoea, against 19% of those who weren't. In the biggest single trial in older hospital patients, 2,941 people aged 65 and over, it was 10.8% on the capsule and 10.4% on placebo.

Neither result is wrong. They tested different people and different products, and that's the whole story of probiotics. "Probiotic" isn't one thing, the way vitamin C is one thing. Lacticaseibacillus rhamnosus GG, the yeast Saccharomyces boulardii and an eight-strain mix for pouchitis are different treatments that happen to share a shelf. So the useful question is which strain, for which problem, at what dose.

What the guideline actually recommends

The American Gastroenterological Association went through the trials condition by condition in 2020. Its list is much shorter than the supplement aisle, and even the yes answers are conditional ones.

A healthy adult wanting a "gut reset" doesn't appear on that list at all.

Antibiotics: the best everyday case, in real numbers

This is where most people meet the question. NNT below means number needed to treat: how many people take the capsule for one of them to avoid the problem.

Three things stand out. First, dose seemed to matter in children: the benefit was clear at high doses and fell short of statistical significance at lower ones. The Cochrane authors singled out L. rhamnosus or S. boulardii at 5 to 40 billion CFU a day as looking most suitable, while saying this still needs a large confirming trial.

Second, the older pooled figures may flatter the capsule. Hempel's reviewers said strains were "poorly documented" across the trials, which makes it hard to know what the pooled number refers to. And the big PLACIDE trial, in the older patients you'd most want to protect, found nothing, though it tested one specific mix and can't speak for LGG or S. boulardii.

Third, the C. difficile numbers have shrunk as more trials came in. The 2017 Cochrane review gave an NNT of 42. The 2025 update, with 47 studies, gives 65, rates the certainty as low, and notes that 28 studies had authors linked to or funded by probiotic companies. In the 2017 review the benefit sat in high-risk settings: where the control group's risk was above 5% it was 3.1% vs 11.6% (NNT 12), and below that there was no clear difference. That was a post hoc analysis, so treat it as a pointer. It does match common sense, though. If you're a healthy 30-year-old on five days of amoxicillin, you're probably not the person these trials helped most.

One caution about recovery

A small study from Israel (Suez and colleagues, Cell 2018) took gut samples after antibiotics and found that people on a multi-strain probiotic took markedly longer to get their own gut microbes back than people left to recover by themselves. It's one product and few people, and nobody has shown it changes health. Still, it's a good reason to treat a probiotic as a short course with an end date, not something to carry on with "to rebuild" afterwards.

IBS: three guidelines, three different answers

A 2018 review pulled together 53 trials and 5,545 patients with IBS. Some strains and combinations looked helpful for overall symptoms and pain, but the authors couldn't draw firm conclusions about which. The guidelines read those uncertain data in three ways:

If you do want to try one, the British version is the fair way to go about it. Pick one product. Choose the one symptom you care about, such as bloating days per week, and count it for two weeks first. Keep your diet steady. Reassess at 12 weeks and stop if the count hasn't moved. Rotating bottles every month tells you nothing. For many people the better-tested next step is food: our IBS and low-FODMAP guide covers those trials.

Children with a stomach bug: two trials, no benefit

Parents often reach for probiotics when a child has vomiting and diarrhoea. Two large trials in the New England Journal of Medicine in 2018 tested this properly, and neither found a benefit. With LGG in 943 American children, moderate-to-severe illness followed in 11.8% vs 12.6% on placebo, and diarrhoea lasted 49.7 vs 50.9 hours. With a two-strain product in Canadian children it was 26.1% vs 24.7%. Fluids are still the treatment.

Reading the label: CFU is a count, not a quality score

CFU means colony-forming units, the microbes able to grow in a lab test. It only means something next to the strain name and the date it's promised for. Labels aren't always reliable either: when researchers tested 16 commercial bifidobacteria products, only 1 matched its label claims in every sample, with differences between lots and even between pills (Lewis and colleagues, Pediatric Research 2016). That's one small study of one group of bacteria, but it's a reason to favour products with independent testing.

A five-line check before you buy:
1. Full strain name. L. rhamnosus GG (ATCC 53103) or S. boulardii CNCM I-745, not just "Lactobacillus blend". The letters and numbers after the species are the part the trials tested.
2. Dose at the end of shelf life. "5 billion through expiry" is a stronger promise than "20 billion at manufacture".
3. A trial for your problem. Evidence for antibiotic diarrhoea doesn't transfer to IBS or immunity.
4. Storage. Check whether it needs the fridge, and think about a hot delivery van if it does.
5. Other ingredients. Added inulin or other prebiotics can make gas worse, especially with IBS.

On timing with antibiotics, pharmacists commonly suggest taking a bacterial probiotic a couple of hours apart from the dose. That's sensible practice more than a trial finding. S. boulardii is a yeast, so antibacterial antibiotics don't target it.

What it costs, and what a prevented case costs

Listed prices we found on 2 November 2026, for products that name a strain tested in the trials above. We have no link to these brands and these aren't recommendations; they're just what the sums look like. The 14-day column assumes a one-week antibiotic course plus a week after.

Now multiply by the NNT, which is pooled across products, so read this as an order of magnitude and not a promise for any one brand:
Antibiotic diarrhoea, NNT 13: 13 × $11.20 = about $146 per case avoided with the LGG product, or 13 × $19.04 = about $248 with the yeast.
C. difficile, average risk, NNT 65: 65 × $19.04 = about $1,238 per case avoided.
C. difficile, high-risk settings, NNT 12: 12 × $19.04 = about $228.

That last gap is the argument in one line. Someone who's had C. difficile before, or an older person on a long course of broad-spectrum antibiotics, has a real conversation to have with their doctor. For a short course in a healthy adult, a mild bout of loose stools is rarely worth $150 to avoid, though if it's the third course this year you may feel differently.

When to skip it, or ask first

For most healthy people the trials show few side effects beyond gas and bloating, and the 2025 Cochrane review concluded short-term use is likely fine for people on antibiotics who aren't immunocompromised. The risk sits with a specific group:
Premature babies. Don't give these outside the neonatal team's plan. In October 2023 the FDA warned hospitals after a preterm infant died of sepsis from bacteria genetically matching the probiotic given, and said it had more than two dozen injury reports since 2018. The American Academy of Pediatrics doesn't support routine use, especially under 1,000 g.
Intensive care or a central line. At one Boston children's hospital, 1.1% of ICU patients given a probiotic developed bloodstream infection, and genome sequencing traced the bacteria to the capsules (Yelin and colleagues, Nature Medicine 2019). Five of the six had central lines.
Serious immune suppression, including chemotherapy, transplant drugs or high-dose steroids. Ask your specialist first.
Active C. difficile. A probiotic isn't the treatment. Severe diarrhoea on or after antibiotics needs a doctor, not a capsule.

See a doctor as well for fever, blood in the stool, dehydration or severe pain. And a capsule doesn't replace fibre-rich food, which feeds the microbes you already have. Our fermented foods article looks at what yogurt, kefir and kimchi did in a Stanford trial, and at why a jar of them isn't the same as a capsule from a trial.

Before buying one, try finishing this sentence: this strain, at this dose, for this problem, until this date. If you can't, the bottle can stay on the shelf.

This article is general information, not medical advice. Check with your clinician or pharmacist before giving a probiotic to a child, or using one in pregnancy, serious illness or alongside immune-suppressing medicines.

Sources

Su G.L. et al., AGA Clinical Practice Guidelines on the Role of Probiotics in the Management of Gastrointestinal Disorders, Gastroenterology 159:697-705 (2020). Guo Q. et al., Probiotics for the prevention of pediatric antibiotic-associated diarrhea, Cochrane Database Syst Rev (2019). Hempel S. et al., Probiotics for the prevention and treatment of antibiotic-associated diarrhea, JAMA 307:1959-69 (2012). Videlock E.J., Cremonini F., Meta-analysis: probiotics in antibiotic-associated diarrhoea, Aliment Pharmacol Ther 35:1355-69 (2012). Allen S.J. et al., Lactobacilli and bifidobacteria in the prevention of antibiotic-associated diarrhoea and C. difficile diarrhoea in older inpatients (PLACIDE), Lancet 382:1249-57 (2013).

Goldenberg J.Z. et al., Probiotics for the prevention of Clostridium difficile-associated diarrhea, Cochrane Database Syst Rev (2017), and Esmaeilinezhad Z. et al., 2025 update. Suez J. et al., Post-antibiotic gut mucosal microbiome reconstitution is impaired by probiotics, Cell 174:1406-23 (2018). Ford A.C. et al., Systematic review with meta-analysis: prebiotics, probiotics, synbiotics and antibiotics in IBS, Aliment Pharmacol Ther 48:1044-60 (2018). Lacy B.E. et al., ACG Clinical Guideline: Management of Irritable Bowel Syndrome, Am J Gastroenterol 116:17-44 (2021). Vasant D.H. et al., British Society of Gastroenterology guidelines on the management of IBS, Gut 70:1214-40 (2021).

Schnadower D. et al., Lactobacillus rhamnosus GG versus placebo for acute gastroenteritis in children, NEJM 379:2002-14 (2018). Freedman S.B. et al., Multicenter trial of a combination probiotic for children with gastroenteritis, NEJM 379:2015-26 (2018). Lewis Z.T. et al., Validating bifidobacterial species and subspecies identity in commercial probiotic products, Pediatr Res 79:445-52 (2016). Yelin I. et al., Genomic and epidemiological evidence of bacterial transmission from probiotic capsule to blood in ICU patients, Nat Med 25:1728-32 (2019). FDA warning on probiotics for preterm infants, as reported by HealthDay, October 2023. Prices: Pasteur Pharmacy, Target, Tata 1mg.