Probiotics for gut health are usually sold by genus name, but a systematic review published in the Journal of Clinical Medicine in February 2026 shows that the genus tells you almost nothing.
Of 67 randomised placebo-controlled trials that named a specific strain, only a handful of strains held up when their results were pooled.[1]
Why the review counted strains instead of genera
The team at Sechenov University searched PubMed and Scopus and assessed 2,643 records, later adding a Cochrane Central search that found nothing new. They registered the protocol in PROSPERO before starting.[1]
Sixty-seven randomised placebo-controlled trials tested a single named strain in irritable bowel syndrome. Only ten of those strains had been studied at least twice, which is the minimum for pooling, leaving 32 studies in the meta-analyses.[1]
Which strains improved symptoms
Bifidobacterium longum 35624, previously classified as Bifidobacterium infantis, reduced abdominal pain, straining during defecation and overall symptom severity. It did not significantly change bloating, urgency, the feeling of incomplete evacuation or passage of gas.[1]
Bacillus coagulans Unique IS2 had the broadest effect in the pooled data, improving pain, bloating, urgency, incomplete evacuation, straining, flatulence and overall symptoms, and normalising stool consistency more often than placebo.[1]
Lactiplantibacillus plantarum 299v reduced abdominal pain and the sense of incomplete evacuation, and more patients on it reported meaningful improvement in pain and flatulence. It did not resolve constipation or reduce bloating severity.[1]
The dose and duration findings are the most useful part
Lactobacillus rhamnosus GG reduced pain overall, but the subgroup analyses are where it gets interesting. The benefit appeared at a daily dose of 6 billion cells and disappeared at 20 billion and 40 billion.[1]
Duration behaved the same way. Benefit was seen only in trials that ran eight weeks, not in those running four or six weeks. More is not better and faster is not available, which runs against how these products are usually marketed.[1]
Saccharomyces cerevisiae CNCM I-3856 reduced abdominal pain overall, but when the data were split by bowel pattern the effect held only in the constipation-predominant group, not in diarrhoea-predominant or mixed patients. It also improved stool consistency in that same group.[1]
Which strains did not work
Three strains failed to beat placebo in the pooled analyses: Escherichia coli Nissle 1917, Lactobacillus gasseri BNR17 and Lactobacillus casei Shirota. Results for Saccharomyces boulardii CNCM I-745 conflicted depending on how outcomes were reported.[1]
Negative results here mean the pooled evidence did not show a benefit, not that the strain is harmful. Adverse event rates did not differ from placebo for any strain in the review.[1]
Diet is still doing more of the work
A 2026 review in Frontiers in Cellular and Infection Microbiology reached the same conclusion about strain specificity and added a comparison.
Structured dietary change, particularly low-FODMAP approaches, produced consistent improvements in abdominal pain, bloating and overall symptom severity across network meta-analyses.[2]
Fibre intake shapes the same system from the other direction. When fibre runs short, gut bacteria have been observed turning to the mucus layer instead, which we covered in an earlier article on fibre and the gut lining.
Probiotics do not settle in the gut
The common description has swallowed bacteria taking up residence, but most strains do not. Within days to weeks of stopping, they disappear from stool samples and the original bacterial community returns.
The effect is therefore explained by what the organisms produce while passing through and the signals they give to gut lining cells: short-chain fatty acid production, competition with pathogenic bacteria, and modulation of mucosal immune signalling.
That structure matches the trial results. Symptoms often creep back after the product is stopped, and for one strain the benefit appeared only in trials that ran a full eight weeks.[1]
Why two strains of one species behave differently
Strains of the same species differ in their gene content. One strain may carry genes for surviving bile acid while another does not, and the surface proteins that let a strain stick to the gut lining vary as well.
Because the proportion that survives stomach acid and bile to reach the colon alive already differs, evidence from one strain cannot be borrowed for another strain of the same species. That is precisely why the 2026 review split its data by strain.[1]
What to check on the label
Look first for the strain code after the species name. Numbers such as 35624, 299v or Unique IS2 are what can be matched to trial data. A box listing only a genus cannot be checked against anything.
Second is the dose. The finding that Lactobacillus rhamnosus GG worked at 6 billion cells a day and not at 20 or 40 billion runs against the assumption that a higher count on the box is better.[1]
Third is duration. Trials that showed benefit generally ran four to eight weeks. Judging a product after a few days does not match how the evidence was generated.
Fermented foods are not the same thing
Kimchi, yoghurt and fermented soybean pastes do contain live organisms. Which strains and how many varies with the product and the fermentation conditions, and it differs between production batches of the same brand.
So fermented food cannot be treated as equivalent to a trial strain. The 2026 review covered single-strain products with a declared strain and dose, and fermented foods were not part of it.[1]
That does not make fermented food pointless. It means it cannot be used as evidence for a specific symptom, and belongs in the diet-as-a-whole column instead.
Heating also matters. Live organisms do not survive cooking, so a simmered fermented dish delivers other fermentation products rather than probiotics.
Side effects and who should be careful
Adverse event rates did not differ from placebo for any strain in the review. Some participants reported more gas or fullness in the first days, and this generally settled.[1]
That safety record comes mostly from healthy adults and people with irritable bowel syndrome. Bacteraemia has been reported rarely in patients on immunosuppressants, during cancer treatment, or with a central venous catheter in place.
Anyone with serious underlying illness should raise it with their clinician before starting. Being sold as a supplement does not mean the product is appropriate in every situation.
What is still unknown
Splitting the data by strain meant discarding every strain without repeat trials. Only 32 of 67 studies survived, which also says that most strains on the market have been tested exactly once.[1]
Multi-strain products were outside the scope. Whether a combination adds to or cancels the effect of its components has to be tested as that combination, and such data remain thin.
There is also no way to predict individual response. People taking the same strain at the same dose respond differently, and whether their existing bacterial community or diet explains it has not been established.[2]
What irritable bowel syndrome is
Irritable bowel syndrome means abdominal pain and altered bowel habit lasting three months or more without a clear abnormality on testing. Diagnosis uses the Rome symptom criteria, which the included trials also applied.[1]
It is divided by bowel pattern into diarrhoea-predominant, constipation-predominant and mixed. That split matters because the same strain gave different results by subtype, as with Saccharomyces cerevisiae CNCM I-3856 working only in the constipation group.[1]
Knowing your own subtype therefore changes which strain the evidence supports. Choosing a strain validated in constipation while having the diarrhoea pattern is a mismatch.
Weight loss, blood in the stool or pain that wakes you at night are not features of this diagnosis. Those signs call for looking at other causes first.
The placebo response here is large
In irritable bowel trials, 30 to 40 percent of placebo recipients report improvement. Symptoms fluctuate on their own, and joining a trial tends to change diet and daily habits at the same time.
So feeling better after starting a product is not by itself evidence that the strain did it. That is why placebo control is required, and why this review included only placebo-controlled trials.[1]
For the same reason a single small trial is weak evidence. Restricting the analysis to strains studied at least twice was a step aimed at that problem.[1]
Storage and timing are part of the dose
Because the product is live organisms, storage conditions affect the count directly. Leaving a refrigerated product at room temperature means taking fewer organisms than the label states.
Some products are shelf stable. Spore-forming organisms such as Bacillus coagulans tolerate heat and stomach acid better, so their storage and transit conditions are less demanding.[1]
Timing advice differs between products. There is evidence for taking it before food and evidence for after, and trials often did not standardise it, so this remains unsettled.
What matters more is keeping the same product and the same routine for a set period. Switching products and timing frequently makes it impossible to tell what worked.
| Strain | What the 2026 pooled analysis showed |
|---|---|
| B. longum 35624 | Reduced abdominal pain, straining and overall symptoms. No effect on bloating, urgency or gas |
| B. coagulans Unique IS2 | Broadest benefit: pain, bloating, urgency, straining, flatulence, stool consistency |
| L. plantarum 299v | Reduced pain and incomplete evacuation. No effect on constipation or bloating severity |
| L. rhamnosus GG | Reduced pain only at 6 billion cells per day and only after 8 weeks |
| S. cerevisiae CNCM I-3856 | Reduced pain in constipation-predominant patients only; improved stool consistency |
| E. coli Nissle 1917, L. gasseri BNR17, L. casei Shirota | No significant benefit over placebo in the pooled data |
Limitations
Heterogeneity was high for several strains, reaching 94.7% for one pain outcome, meaning the underlying trials disagreed. Many included trials were small, and several reported results in formats that could not be pooled at all.
The review covers irritable bowel syndrome specifically. It does not tell you what any of these strains do for general digestion, immunity or the other claims printed on supplement boxes, because those trials were not part of the question.
References
- Ivashkin V, et al. Strain-Specific Systematic Review with Meta-Analysis of Probiotics Efficacy in the Treatment of Irritable Bowel Syndrome. Journal of Clinical Medicine, 2026 Feb 2;15(3):1152. PROSPERO CRD420251047092.
- Probiotics in irritable bowel syndrome: strain-specific effects, diet, and biomarker timing. Frontiers in Cellular and Infection Microbiology, 2026.
