Wuxi Baifude Biotechnology Co., Ltd.
← Technical Notes

Technical Note

A clean conclusion, tested against its data: 299v in IBS as a worked example

Supplier dossiers and sell sheets regularly carry a line like "supported by a meta-analysis of six randomised controlled trials, 527 participants." A meta-analysis that came out this week in Probiotics and Antimicrobial Proteins (14 November 2026) is a clean example: it concludes that Lactiplantibacillus plantarum 299v relieves abdominal pain and bloating in irritable bowel syndrome (IBS). We traced the six trials it pooled. Three were positive and three were null, and the pooled result rests largely on two trials whose placebo arms responded unusually little. This note does not assess the strain. It walks through a check any buyer or formulator can run in under twenty minutes before a meta-analysis goes into a dossier.

Published 19 November 2026 · 8 min read

What the meta-analysis reported

Design

Searches of PubMed/MEDLINE, Embase, Scopus and CENTRAL for randomised placebo-controlled trials of L. plantarum 299v in adults with IBS, pooled with a random-effects model. Protocol registered in PROSPERO. Six trials, 527 participants.

Result

  • Abdominal pain: pain score SMD -2.16 (95% CI -3.59 to -0.72); pain severity SMD -0.39; daily pain frequency SMD -0.66.
  • Responder rates: pain improvement RR 4.83, bloating/flatulence RR 1.96, global symptoms RR 4.69.
  • Constipation: no significant effect.

The full text is paywalled; the figures above are from the abstract.

The six trials, one by one

The abstract does not list the included trials. The reference list on Europe PMC contains six randomised placebo-controlled trials of 299v in IBS. Five report their size, totalling 446; the sixth (South Africa) does not state it in the abstract, and the difference from the reported 527 puts it at about 81.

  • Sweden, 2000, n = 60, 4 weeks: positive. Faster reduction in flatulence; delivered in a rose-hip drink.
  • Poland, 2001, n = 40, 4 weeks: positive, very large. Abdominal pain resolved in all 20 treated patients versus 11 on placebo; global improvement 95% versus 15%.
  • United Kingdom, 2002, n = 12, crossover: null. No difference in symptoms or colonic fermentation.
  • India, 2012, n = 214, 4 weeks: positive, very large. 78.1% rated the effect excellent or good, against 8.1% on placebo.
  • South Africa, 2014, about 81, 8 weeks: null. No difference in pain relief or quality of life; the authors note both arms improved markedly, "indicating a large placebo effect."
  • Pakistan, 2022, n = 120 randomised: null. No difference in pain, bloating or rectal emptying.

Three and three. The pooled result is positive not because most trials were, but because the positive ones were large.

What pulled the pooled estimate up

Placebo arms that barely responded. IBS is a high-placebo condition; placebo improvement rates in the region of a third are unremarkable, and the South African trial saw exactly that. Against that background, 8.1% (India) and 15% (Poland) are low. A suppressed placebo arm widens the between-arm gap and inflates responder RRs, and the RRs of 4.7-4.8 in the abstract lean heavily on these two trials.

An implausible effect size. By convention an SMD of 0.2 is small, 0.5 moderate and 0.8 large; probiotic effects in IBS typically sit in the small-to-moderate band. An SMD of -2.16 with a confidence interval that wide signals strong disagreement between trials. A rough calculation from the means and SDs in the Indian trial's abstract gives roughly -0.4 for pain severity and -0.8 for frequency, neither near -2. Which trial drives the -2.16 cannot be seen from the abstract. It may be one extreme trial, or a data-extraction issue (standard errors entered as standard deviations is a recurring one); the full text would settle it.

Decline over time. The three positive trials date from 2000-2012. The two later trials, from different countries and research groups, found no difference. Large early effects that later trials fail to reproduce are a well-known warning pattern.

None of this shows that any individual trial was wrong; abstracts are not enough to say why the results diverge. What it does show is that "six RCTs support it" is tidier than the data underneath.

What it does and does not establish

Supportable: 299v is among the more studied organisms in IBS, with six randomised trials, and the pooled direction favours pain and bloating, with no effect on constipation.

Needs a discount: individual trials are inconsistent, half null; the pain-score SMD of -2.16 should not be quoted; later independent trials did not replicate.

Not supportable:

  • "299v does not work." Inconsistent is not ineffective.
  • Carrying the finding to another endpoint. For iron absorption, 299v is the only organism with poolable human data, from a different set of trials and a different mechanism; see Probiotics alongside iron. Weak evidence on one endpoint says nothing about another, in either direction.
  • "Strains within a species differ widely." The same strain gave opposite results in different populations. Population, trial conduct and placebo response are at least as plausible explanations as strain biology.

A five-step check before citing a meta-analysis

  • Read the abstract, not the summary of it. Confirm size, duration, primary endpoint and result.
  • List the included trials and count positives and nulls. Europe PMC usually exposes the reference list; open each primary trial's abstract.
  • Look at the placebo arm. Unusually low placebo response, or a near-100% response on treatment, warrants a question.
  • Sanity-check the effect size. An SMD above 1 with a wide interval calls for checking the data before trusting the conclusion.
  • Check the timeline and independent replication, and note the funding. Sponsor funding does not disqualify a trial; it raises the value of independent confirmation.

It is the same reading we applied in probiotics and mood and sleep in healthy adults and the broiler synbiotic meta-analysis.

A note from an ingredient manufacturer

Baifude manufactures lactic acid bacteria, bifidobacteria, Bacillus, yeast and blended functional powders, including Lactiplantibacillus plantarum. This note concerns a commercial strain of a species we also supply, and the interest should be stated.

It is not an assessment of 299v and is not directed at any company. We chose this paper because it appeared this week and pools few enough trials to trace completely. We do not use the inconsistency shown here against any competing product; we reported 299v's iron-absorption evidence as published and continue to.

Nor do we suggest that another supplier's inconsistent evidence makes our own strains better. We apply the same check to our own data, and we do not rely on any study cited here to support claims for our own powders.

What this means for brands and formulators

1. Ask for the trial list, not just the pooled line

When a supplier cites a meta-analysis, request the included trials and the result of each. A pooled sentence is not a substantiation file.

2. Weight placebo response on subjective endpoints

Pain, bloating, sleep and mood are self-reported and placebo-sensitive. A low placebo response is a common reason a pooled effect looks larger than it is.

3. Check outsized effect sizes before using them

Do not carry an SMD above 1 with a wide interval into marketing copy until you know which trial produced it and whether it was extracted correctly.

4. Treat early-positive, later-null as inconsistent

Citing only the earliest and largest results is the easiest way to overstate a position. Replicated findings are what a positioning can safely rest on.

5. Substantiate endpoint by endpoint

The same strain can have solid evidence on one endpoint and inconsistent evidence on another. Claims should be checked per endpoint, not bundled per strain.

6. Power your own trials for a realistic placebo response

A small trial with an unusually quiet placebo arm can produce a positive result that later trials will not reproduce.

If you are assessing this direction

We can help at the documentation level: taking apart a meta-analysis a claim relies on, listing its trials and their results, and checking whether the organism, dose and population match yours. Strain identification and formulation feasibility can be discussed alongside. We do not use any study to support the efficacy of any product, ours or anyone else's.

Contact us

References

  1. Brindicci VF, Dargenio VN, La Grasta G, et al. The Effects of Lactiplantibacillus plantarum 299v on Irritable Bowel Syndrome: A Systematic Review with Meta-analysis. Probiotics and Antimicrobial Proteins, 14 November 2026. doi:10.1007/s12602-026-11296-0. PubMed 43033380
  2. Nobaek S, Johansson ML, Molin G, et al. Alteration of intestinal microflora is associated with reduction in abdominal bloating and pain in patients with irritable bowel syndrome. Am J Gastroenterol, 2000. PubMed 10811333
  3. Niedzielin K, Kordecki H, Birkenfeld B. A controlled, double-blind, randomized study on the efficacy of Lactobacillus plantarum 299V in patients with irritable bowel syndrome. Eur J Gastroenterol Hepatol, 2001. PubMed 11711768
  4. Sen S, Mullan MM, Parker TJ, et al. Effect of Lactobacillus plantarum 299v on colonic fermentation and symptoms of irritable bowel syndrome. Dig Dis Sci, 2002. PubMed 12452404
  5. Ducrotté P, Sawant P, Jayanthi V. Clinical trial: Lactobacillus plantarum 299v (DSM 9843) improves symptoms of irritable bowel syndrome. World J Gastroenterol, 2012. PubMed 22912552
  6. Stevenson C, Blaauw R, Fredericks E, et al. Randomized clinical trial: effect of Lactobacillus plantarum 299 v on symptoms of irritable bowel syndrome. Nutrition, 2014. PubMed 25194614
  7. Moeen-Ul-Haq, Babar AN, Hassan MK, et al. Role of Lactobacillus plantarum 299v versus Placebo in symptomatic improvement of irritable bowel syndrome patients. J Pak Med Assoc, 2022. PubMed 35320214

This note summarises and interprets published research in order to show how far a meta-analytic conclusion can be cited. The full text of reference 1 is paywalled; we checked its abstract, its reference list as indexed on Europe PMC, and the abstracts of references 2-7. Which trials reference 1 actually pooled, and how data were extracted, are governed by the full paper. Effect-size estimates from abstract figures are approximate and illustrate magnitude only. This note is not a claim about the efficacy or safety of any product, is not directed at any company or brand, and is not medical, nutritional or regulatory advice. Claim substantiation requirements differ between jurisdictions; the rules of the market concerned govern in every case. Findings apply only to the populations, organisms, doses and durations described in the original sources.