Probiotics: Why the Strain on the Label Matters More Than the CFU Count

"Probiotics" is not one thing — it is a regulatory catch-all for live microorganisms from dozens of genera, hundreds of species, and thousands of proprietary strains, most of which have never been tested in a single human trial. The strongest independent human evidence supports specific, named strains for specific problems: Saccharomyces boulardii CNCM I-745 and Lactobacillus rhamnosus GG for antibiotic-associated diarrhea, multi-strain mixtures for preventing C. difficile infection during antibiotic treatment, and select strains for shortening acute infectious diarrhea in children. Claims about general "gut health," immune support, weight loss, and mood are weak, mixed, or actively rejected by regulators — the European Food Safety Authority has turned down essentially every general probiotic health claim it has reviewed. Independent testing regularly finds mislabeled CFU counts, and rare but documented bloodstream infections occur in immunocompromised people, ICU patients, and those with central lines. The practical rule from the field's own scientific consensus body is blunt: match the strain, the dose, and the indication — do not buy "a probiotic" and expect it to do anything in particular (World Gastroenterology Organisation Global Guideline on Probiotics and Prebiotics, 2023).

Table of contents

• Evidence summary
• What probiotics are
• All forms and types of probiotics
• How probiotics work
• What works and what does not
• Benefits with evidence grades
• Risks and all side effects
• All interactions
• Who should avoid probiotics
• Dosage and how to take probiotics
• Infographics
• Related research
• Frequently asked questions
• Sources and funding notes

Evidence summary

What probiotics are

Probiotics are defined by the International Scientific Association for Probiotics and Prebiotics (ISAPP) consensus panel as "live microorganisms that, when administered in adequate amounts, confer a health benefit on the host" (Hill et al., ISAPP consensus statement). That definition contains three separate requirements that most commercial products never demonstrate: the organism has to be alive at the point of consumption, it has to be present in an amount shown to work, and it has to actually confer a specific, testable benefit — not a vague, unverifiable one.

This is why "probiotics" as a single category is close to meaningless from an evidence standpoint. The category spans bacteria from genera as different as Lactobacillus (recently reclassified into more than 25 genera, including Lacticaseibacillus and Limosilactobacillus), Bifidobacterium, Streptococcus, Bacillus, and Enterococcus, plus at least one yeast, Saccharomyces boulardii (a subtype of Saccharomyces cerevisiae). Effects documented for one strain of one species cannot be assumed to transfer to a different strain of the same species, let alone a different species or genus entirely. The World Gastroenterology Organisation's 2023 global guideline is explicit about this: recommendations are made "at the strain level" because effects are strain-specific, not species-specific or genus-specific (WGO Global Guideline, 2023).

Strain naming conventions matter enormously and are routinely stripped off product labels. A properly identified strain looks like Lactobacillus rhamnosus GG (also written as ATCC 53103) or Saccharomyces boulardii CNCM I-745 — genus, species, and a unique strain designator (a code, a name, or a deposit number in a culture collection). A label that says only "Lactobacillus acidophilus" or, worse, "probiotic blend, 50 billion CFU," gives a consumer no way to check whether that specific organism was ever tested in a human trial for the condition they are trying to treat.

Probiotics should also be distinguished from three related but different terms that appear on the same shelves: prebiotics (non-digestible fibers that feed beneficial bacteria already in the gut, such as inulin and fructooligosaccharides), synbiotics (a combination of a probiotic and a prebiotic designed to work together), and postbiotics (inactivated microorganisms or their metabolic byproducts, which do not need to be alive to work). ISAPP has issued a separate consensus definition for each of these categories, precisely because marketing routinely blurs the lines between them (ISAPP consensus definitions roundup).

All forms and grades

The table below groups the genera and species with the most independent human-trial evidence, organized by the type of clinical claim they are most associated with. This is not an exhaustive list of every probiotic species sold commercially — it is a map of where independent human evidence actually concentrates.

Beyond strain identity, format affects viability and, in turn, real-world dose. Capsules and sachets vary in whether they require refrigeration; some modern formulations use spore-forming Bacillus species specifically because spores survive room-temperature storage and stomach acid better than vegetative (non-spore) bacterial cells, though human outcome data for Bacillus-based products remains thinner than for the Lactobacillus/Bifidobacterium/S. boulardii strains above. Fermented foods (yogurt, kefir) contain live cultures but rarely at the specific strain identity or CFU dose used in the clinical trials that generated the evidence in this article — a cup of yogurt is not interchangeable with a capsule of L. rhamnosus GG at 10 billion CFU.

How probiotics work

Probiotic mechanisms are studied mostly through in-vitro and animal work, but the clinically relevant question is simpler: does a given strain, at a given dose, survive transit and produce a measurable clinical effect in a human trial? Proposed mechanisms — competitive exclusion of pathogens, short-chain fatty acid production, modulation of gut barrier integrity, immune signaling through pattern-recognition receptors, and bile salt hydrolysis — are generally derived from in-vitro or animal models and should not be treated as proof of clinical benefit in humans. Per this article's evidence standard, mechanistic claims resting only on animal or non-human-relevant in-vitro data are noted as such and are not used to support any benefit or safety claim; see the excluded animal-study table in the Sources section.

Two practical, human-trial-relevant factors determine whether a probiotic can plausibly work at all:

• Survival through gastric acid and bile. Many organisms die in transit; formats matter (enteric coatings, spore-formers, or yeast such as S. boulardii, which tolerates gastric acid and antibiotics that kill bacterial probiotics, are more resilient by nature).
• Dose delivered versus dose labeled. A strain proven effective at 10 billion CFU/day in a trial provides no assurance of benefit at a different dose, and independent testing shows real-world products frequently do not deliver the labeled count by the time they reach a consumer (see the Product quality section below).

What works and what does not

Antibiotic-associated diarrhea (AAD) — Strong evidence, strain-specific

This is the single best-supported use of probiotics in independent human trials. A 2017 systematic review and meta-analysis of outpatient randomized controlled trials found that specific probiotic strains, taken alongside antibiotics, reduce the risk of developing diarrhea as a side effect (Blaabjerg, Artzi & Aabenhus, Antibiotics, 2017). The Cochrane-indexed pediatric review (search updated through May 2018) pooled 33 trials covering 6,352 children aged 3 days to 17 years and found probiotics — Lactobacillus spp., Bifidobacterium spp., Streptococcus spp., or Saccharomyces boulardii, alone or in combination — reduced AAD risk compared with placebo or no treatment (Cochrane CD004827).

The best-characterized strains for this indication are Saccharomyces boulardii CNCM I-745 and Lactobacillus rhamnosus GG. A 2024 clinical review by academic authors summarizing Szajewska and Kolodziej's meta-analyses reports a relative risk of 0.47 (95% CI 0.38–0.57) for AAD with S. boulardii and 0.48 (95% CI 0.26–0.89) with L. rhamnosus GG, both versus placebo (Advances in Therapy, 2024). Typical studied doses are 250 mg once daily in children and 500 mg once daily (or 250 mg twice daily) for S. boulardii, and at least 10⁹ CFU/day for L. rhamnosus GG per European Society for Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) guidance. Timing matters: benefit is clearest when the probiotic is started within 48 hours of the first antibiotic dose, not added after diarrhea has already begun.

Clostridioides difficile-associated diarrhea (CDAD) prevention — Strong evidence in appropriate patients

The 2017 Cochrane review by Goldenberg and colleagues pooled 31 randomized trials covering 8,672 patients receiving antibiotics and found moderate-certainty evidence that probiotics reduce the risk of developing CDAD (Goldenberg et al., Cochrane Database Syst Rev, 2017). The review also found probiotics were probably not harmful when taken alongside antibiotics, with fewer mild-to-moderate adverse effects in the probiotic arm than placebo — but the review explicitly flags that people with weakened immune systems were underrepresented and that harm cannot be ruled out in that population (Cochrane evidence summary, CD006095). This is one of the few probiotic indications where a WGO/ISAPP-consistent, strain-agnostic pooled result exists at meaningful scale — most trials in the pool used Lactobacillus, Saccharomyces, or multi-strain combinations rather than a single named strain, so the "which exact strain" question is less settled here than for AAD generally.

Acute infectious diarrhea in children — Strong evidence for specific strains, weak for prevention

The WGO's 2023 global guideline states that certain probiotic strains shorten the duration of acute diarrheal illness in children by roughly one day, based on multiple meta-analyses with consistent results (WGO Global Guideline, 2023). This is a treatment effect on an already-established episode, not a prevention effect. The guideline separately notes that a Cochrane meta-analysis on prevention of diarrhea (rather than treatment of an active episode) concluded probiotics "probably make little or no difference" once diarrhea has lasted 48 hours or longer, and that early administration is likely necessary for any benefit. The NIH Office of Dietary Supplements echoes this distinction: "some probiotic strains might help reduce duration of acute diarrhea by about 1 day" (NIH ODS Probiotics Fact Sheet, 2025).

Necrotizing enterocolitis (NEC) prevention in preterm infants — Strong but strain-restricted, and safety is contested

This indication carries the highest stakes in the entire probiotics literature: a severe, sometimes fatal bowel condition in the most vulnerable patients, weighed against real infection risk from live-organism supplementation in immature immune systems.

The Cochrane systematic review on this topic found that probiotic supplementation of milk feeds reduces NEC and all-cause mortality in very preterm and very-low-birth-weight infants, but is explicit that "not all preparations" work and that effect is concentrated in specific multi-strain products (Cochrane CD005496). WGO cites a number needed to treat (NNT) of approximately 20 to prevent one death from any cause. A 2020 ESPGHAN position paper conducted its own network meta-analysis and found efficacy in reducing mortality and morbidity for only a minority of the strains and combinations studied — not probiotics as a category. It issued a conditional, low-certainty-evidence recommendation for either Lactobacillus rhamnosus GG ATCC 53103 alone, or the specific three-strain combination of Bifidobacterium infantis Bb-02, Bifidobacterium lactis Bb-12, and Streptococcus thermophilus TH-4 (ESPGHAN Position Paper). Strains tested with sufficient sample size (≥247 infants per group in randomized trials) included Bifidobacterium breve BBG-001 (YIT4010), Limosilactobacillus reuteri DSM 17938, L. rhamnosus GG ATCC 53103, and S. boulardii CNCM I-745.

The American Academy of Pediatrics has historically taken a more cautious position than European groups. As of the 2020 ESPGHAN position paper's review of the literature, "the AAP stated that current evidence does not support routine administration of probiotics to preterm infants, particularly those with a birth weight of <1000 g," while noting this did not preclude considering probiotics case-by-case. A newer AAP clinical report, "Safety and Efficacy of Probiotics for Preterm Infants," was published in Pediatrics in early 2025, reflecting the continued evolution of the evidence and the ongoing tension between benefit signal and infection risk (AAP Clinical Report, Pediatrics, 2025).

The FDA took direct regulatory action on this exact question in September 2023, warning healthcare providers that probiotic products marketed to prevent or treat NEC in hospitalized preterm infants can be dangerous, and issuing warning letters to two companies for illegally marketing products for this purpose. The agency stated it was aware of at least one infant death in 2023 and more than two dozen other reported adverse events in the United States since 2018 tied to probiotic products used in preterm infants (FDA Press Announcement, September 2023). A 2025 UK neonatal network guideline, by contrast, continues to recommend routine probiotic use in infants born under 32 weeks or under 1.5 kg, citing an estimated invasive-infection rate of less than 1 per 500 treated infants from UK NICU experience and noting the most recent Cochrane review found no probiotic-associated invasive infections among more than 11,000 infants reviewed, while acknowledging a small number of case reports exist in the wider literature (East of England Neonatal Network Probiotic Guideline, 2025).

Irritable bowel syndrome (IBS) — Moderate, strain-specific, and guideline-contested

IBS is the clearest example of guidelines disagreeing because they weigh the same low-quality, heterogeneous evidence differently. The American College of Gastroenterology's 2021 clinical guideline states: "We suggest against probiotics for the treatment of global IBS symptoms" — a conditional recommendation based on what it grades as very low quality evidence, explicitly citing the huge variability between probiotic products as a barrier to specific recommendations (ACG Clinical Guideline, 2021). NCCIH summarizes the same position: "there's not enough strong evidence on the effectiveness of probiotics for IBS to recommend their use" (NCCIH, IBS: What You Need To Know).

The British Society of Gastroenterology's 2021 guideline reaches a different conclusion using the same evidence base, issuing a weak recommendation (very low quality evidence) that probiotics "can be an effective treatment for general symptoms and abdominal pain in IBS," while stating it is "impossible to recommend a specific species or strain," and advising a 12-week trial with discontinuation if there is no improvement. A 2021 review comparing 17 professional-society guidelines and position statements on probiotics for IBS found 13 recommended some form of probiotic use, 4 did not, and quality of evidence ranged from very low to moderate across all of them (Przegląd Gastroenterologiczny review, 2021).

Where specific strains have shown IBS-relevant results in trials, they are named-strain, not category, effects. The WGO guideline specifically cites Bifidobacterium longum subsp. longum 35624 (the strain behind the branded product Align/Bifantis) at a studied dose of 100 million CFU/day as effective for alleviating IBS symptoms. It also cites Lactiplantibacillus plantarum 299v (DSM 9843) for improving IBS symptoms and Bifidobacterium bifidum MIMBb75 (including a heat-inactivated version, SYN-HI-001) for significantly alleviating IBS and improving quality of life. None of these single-strain findings should be read as validating "probiotics" broadly for IBS — they are isolated, strain-specific results embedded in an overall guideline landscape that leans skeptical (ACG) to weakly permissive (BSG).

Helicobacter pylori eradication adjunct — Moderate evidence, adjunct only

S. boulardii CNCM I-745 has the most consistent adjunct evidence: a randomized, open-label trial found it improved outcomes when added to sequential antibiotic therapy for H. pylori (RCT, 2019), and real-world data from the European Registry on H. pylori Management (Hp-EuReg) support its use as adjunct therapy (Hp-EuReg analysis). Evidence for L. rhamnosus GG as an adjunct is more mixed across trials. The WGO guideline is explicit on the limits: "certain probiotics reduce gastrointestinal side effects caused by H. pylori eradication therapies," improving tolerability and potentially completion rates — but there is "no evidence for probiotic alone without antibiotics" actually eradicating the infection. Probiotics in this context are a tolerability adjunct to standard triple or sequential antibiotic therapy, never a substitute for it.

Traveler's diarrhea — Mixed, modest effect at best

An older meta-analysis of 12 randomized trials using various probiotics — including S. boulardii, Lactobacillus species, and mixed-strain products — found a modest reduction in traveler's diarrhea risk (relative risk 0.85, 95% CI 0.79–0.91), summarized in a broader World Journal of Gastroenterology review (World Journal of Gastroenterology, 2010). More recent systematic reviews and meta-analyses have found effects to be inconsistent across trials, destination, and strain, and no single strain has emerged as a clearly superior, reproducibly effective choice for this indication (Systematic review and meta-analysis, Travel Medicine and Infectious Disease; Adaptive meta-analysis). This is a "might help a little, might not, evidence is thin and heterogeneous" category — not a strong recommendation.

Vaginal health — Moderate, but mostly for specialized live biotherapeutic products, not oral capsules

Lactobacillus crispatus-based live biotherapeutic products, delivered as vaginal suppositories rather than oral capsules, have gone through phase 1 through phase 2/3 randomized, placebo-controlled trials for recurrent bacterial vaginosis and urinary tract infection prevention, with positive signal in trials specifically designed around this strain and this delivery route (Phase 2 RCT; Multi-strain vaginal suppository RCT). This is an important distinction from most other probiotic categories in this article: it is a purpose-built, vaginally-administered live biotherapeutic product being tested through a drug-like development pathway, not an oral, over-the-counter dietary supplement. Oral probiotic capsules marketed generically for "vaginal health" or "feminine balance" do not have the same trial support as these specific vaginal formulations, and should not be assumed equivalent.

General "gut health" — Weak, and directly contradicted by regulators

This is the single most commercially exploited and least evidence-supported claim in the entire probiotics market. There is no generic human-trial evidence that taking "a probiotic" improves an undefined concept of "gut health" absent a specific, diagnosed condition (like those addressed above). The European Food Safety Authority's health claims review, covering the period from roughly 2009 to 2012, rejected virtually every general digestive-health and immune-support claim submitted for probiotic products, including from major manufacturers — see the Regulatory reality section below for the full story. When a specific strain has a specific proven use (like AAD prevention or IBS symptom relief in trials), that is evidence for that specific claim — it is not evidence that probiotics generally "support gut health."

Immune support — Weak, mostly small and mixed trials

A systematic review specifically assessing circulating immune and inflammatory markers in healthy adults found probiotic supplementation had "a limited effect" across the randomized controlled trials it reviewed (Systematic review, 2019). A separate systematic review and meta-analysis of salivary cytokines and immunoglobulins across clinical trials found effects that were inconsistent and highly dependent on the specific outcome measured and strain used (Scientific Reports, 2020). Broader reviews of probiotic and synbiotic effects on immune cell counts likewise found mixed, low-confidence results across trials (Nutrition and Cancer, 2019). This is a category where mechanistic plausibility (probiotics interact with gut-associated immune tissue) has run far ahead of consistent clinical proof of a meaningful immune benefit in generally healthy people.

Weight loss and metabolic markers — Very weak human evidence

Multiple systematic reviews and meta-analyses of randomized trials on probiotics/synbiotics for weight loss and metabolic syndrome markers report small, inconsistent effect sizes across a heterogeneous set of strains, doses, and populations (Systematic review and meta-analysis; Systematic review, metabolic syndrome; Systematic review, weight loss). The NIH Office of Dietary Supplements sums up the state of play plainly: "some studies have shown that probiotics might slightly reduce body weight or body fat. Other studies have shown that probiotics have no effect or might even increase body weight. More research is needed" (NIH ODS Probiotics Fact Sheet). There is no independent human-trial basis for marketing any probiotic as a weight-loss product.

Mood and "psychobiotics" — Preliminary, small trials, flag as emerging

"Psychobiotics" — probiotics studied for mental health outcomes via the gut-brain axis — is an active but immature research area. A 2025 review synthesized 19 randomized controlled trials on precision psychobiotics as adjunct or stand-alone therapy in adult depression, describing the space as promising but still early (2025 RCT synthesis). Other reviews of psychobiotic effectiveness for psychiatric and neurological symptoms and for depression treatment specifically report heterogeneous strains, small sample sizes, and short durations across most included trials (Effectiveness review; Depression and comorbidities review). This article flags mood/psychobiotic claims as preliminary: encouraging enough to justify further trials, not strong enough to support a clinical recommendation to treat depression or anxiety with any specific probiotic strain today.

Skin (acne, eczema) — Mixed, inconsistent results

A meta-analysis of randomized controlled trials on probiotics for acne vulgaris found mixed results across a small and heterogeneous trial base (Meta-analysis, acne). An earlier systematic review of probiotics for eczema treatment likewise found inconsistent evidence of benefit (Systematic review, eczema treatment). Separately, the NIH Office of Dietary Supplements notes that when pregnant women take probiotics and their infants are also given probiotics after birth, some studies show a lower risk of developing atopic dermatitis with less severe symptoms in the infant — a prevention signal in a specific perinatal context, not a treatment claim for existing acne or eczema in the general population (NIH ODS Probiotics Fact Sheet).

Long COVID / post-viral recovery — Early trials only

Trials examining probiotics for long COVID and post-viral symptom recovery remain early-stage, with small sample sizes and short follow-up (Post-COVID probiotic/postbiotic review; Post-exposure prophylaxis trial). There is not yet an independent evidence base sufficient to recommend any specific probiotic for long COVID or post-viral syndrome management.

ICU / critically ill patients — Not supported, and one landmark trial found real harm

This is the most important "does not work — and may actively hurt" finding in the probiotics literature, and it deserves the most careful treatment.

The PROPATRIA trial (Besselink et al., published in The Lancet in 2008) was a multicenter, double-blind, placebo-controlled randomized trial in the Netherlands, run by the Dutch Acute Pancreatitis Study Group, enrolling 296 patients with predicted severe acute pancreatitis (152 randomized to a multispecies probiotic preparation — three Lactobacillus species, two Bifidobacterium species, and one Lactococcus species, at 10¹⁰ CFU twice daily — and 144 to placebo). The trial found no difference in the primary endpoint (infectious complications: 30% probiotic vs. 28% placebo, relative risk 1.06). But mortality was significantly higher in the probiotic group: 24 of 152 patients (16%) died, versus 9 of 144 (6%) in the placebo group — a relative risk of 2.53 (95% CI 1.22–5.25). Nine patients in the probiotic group developed bowel ischemia, of whom eight died, compared with zero cases of bowel ischemia in the placebo group (p=0.004). The trial's own conclusion: "probiotic prophylaxis should therefore not be administered in this category of patients" (Besselink et al., The Lancet, 2008).

This result triggered years of re-analysis and debate about mechanism (proposed explanations include bowel ischemia from a mismatch between fermentable substrate delivery and impaired gut perfusion in critically ill pancreatitis patients), but it has not been overturned by a subsequent trial in the same population, and no probiotic has since been shown safe and effective for acute pancreatitis prophylaxis. Follow-up systematic reviews and meta-analyses of probiotics across acute pancreatitis trials broadly have found no significant benefit or harm on other outcomes when PROPATRIA is pooled with smaller studies, but they also do not overturn the core PROPATRIA mortality finding in its own population, and most conclude the evidence is not sufficient to recommend probiotic use in this setting (Critical Care systematic review, 2014).

More recent large trials in other critically ill populations reinforce a consistent "no meaningful benefit" pattern, even without replicating PROPATRIA's harm signal. The PROSPECT trial (Johnstone et al., published in JAMA in 2021) randomized 2,650 mechanically ventilated ICU patients to Lactobacillus rhamnosus GG or placebo and found no significant difference in ventilator-associated pneumonia (21.9% vs. 21.3%, hazard ratio 1.03) and no significant difference on any of 20 prespecified secondary outcomes, including mortality (Johnstone et al., JAMA, 2021). A 2022 systematic review and meta-analysis of probiotics/synbiotics in critical illness concluded probiotics "probably have no effect on mortality" (moderate certainty), and while some pooled analyses suggested possible reductions in ventilator-associated pneumonia, hospital-acquired pneumonia, and length of stay, sensitivity analyses that excluded high-risk-of-bias studies negated those secondary findings (Systematic review, 2022).

Healthy people, for general prevention — Not supported

There is no independent human-trial basis for healthy, asymptomatic people taking a probiotic supplement for general disease prevention. The NIH Office of Dietary Supplements states this directly in its most recent update: "There are no official recommendations that cover the use of probiotics by healthy people" (NIH ODS Probiotics Fact Sheet, 2025). This reflects a body of expert review going back over a decade, including Guarner and colleagues' widely cited 2013 update on probiotic use and investigation status, and more recent 2024 reviews revisiting the same question for a general healthy population (Guarner et al., 2013). Every strong or moderate finding in this article is tied to a specific diagnosed condition or risk state (on antibiotics, diagnosed with IBS, undergoing H. pylori eradication, born very preterm) — not to healthy people taking a supplement "just in case."

Benefits by claim

Risks and all side effects

Per this article's evidence standard, only independent human-trial and human case-report data are used to characterize side effects. No animal toxicology data is used to support any safety conclusion in this section.

Common, mild side effects

• Gas and bloating. The most frequently reported effect in healthy people, typically transient and dose-related, per NIH ODS: "In healthy people, probiotics may cause gas, but they rarely cause infections or other health problems" (NIH ODS Probiotics Fact Sheet).
• Cramping and mild GI upset. Reported in IBS trial populations specifically, generally minor: "In people with IBS, probiotics usually have only minor side effects, if any... cramping and nausea" (NCCIH, IBS).

Rare but serious side effects

• Invasive bloodstream infections (bacteremia). A matched case-control study at an American academic hospital (Oregon Health & Science University), funded by the National Institutes of Health with no industry funding and no disclosed conflicts of interest, found the odds of probiotic use among patients with invasive infections (positive cultures from blood, cerebrospinal fluid, ascitic fluid, or pleural fluid caused by typical probiotic organisms) were 127 times higher than among matched controls (95% CI 6.21–2600). Lactobacillus was the most common organism isolated (68% of cases). Recent abdominal surgery was a significant risk factor among cases. In-hospital mortality was substantially higher among cases (14%) than controls (2%) (Tom et al., 2021). Note that this study also found severe immunosuppression, solid-organ transplantation, and chronic kidney disease were more common among controls than cases — likely reflecting that clinicians were already avoiding probiotics in the sickest, most immunosuppressed patients, which is itself informative about real-world risk perception.
• Fungemia (yeast bloodstream infection) with Saccharomyces boulardii. Multiple published case reports and case series describe fungemia associated with S. boulardii use, disproportionately in patients with central venous catheters, where contamination during capsule handling near the catheter site is a documented transmission route. A historical case series described seven cases of fungemia linked to S. boulardii (Case series, 2002), and more recent case reports continue to document central-line-associated S. boulardii fungemia (Case report; Case report, central line-related bloodstream infection). A national medicines regulator (Malaysia's NPRA) has issued a specific safety alert on S. boulardii fungemia risk.
• Invasive infection/sepsis in preterm infants. As detailed in the NEC section above, the FDA documented at least one 2023 infant death and more than two dozen other adverse events tied to probiotic products used in hospitalized preterm infants since 2018 (FDA, 2023).
• Excess mortality in a specific critical-illness population. As detailed above, the PROPATRIA RCT found a statistically significant increase in mortality (16% vs. 6%) and bowel ischemia specifically in patients with predicted severe acute pancreatitis given a multispecies probiotic (Besselink et al., 2008). This is a documented harm in a specific population, not a generalized statement that probiotics cause harm in all patients.
• Unlisted contaminant organisms. NCCIH notes: "Some probiotic products have been reported to contain microorganisms other than those listed on the label. In some instances, these contaminants may pose serious health risks" (NCCIH, Probiotics: Usefulness and Safety).
• Theoretical transfer of antibiotic resistance genes. NCCIH lists this as a possible harmful effect of probiotics, alongside infection and harmful-substance production by the organisms themselves (NCCIH, Probiotics: Usefulness and Safety). ESPGHAN's NEC position paper similarly conditionally recommends only strains verified to be free of plasmids carrying transferable antibiotic-resistance genes.

Populations at elevated risk

All interactions

Independent human evidence on probiotic-drug interactions is thinner than for most conventional drug classes, largely because probiotics act through live-organism colonization and biological competition rather than classic pharmacokinetic pathways (CYP450 metabolism, protein binding, etc.). Where independent human data is limited to case reports or is entirely absent, that gap is stated explicitly below rather than treated as evidence of "no interaction."

Who should avoid probiotics

• Immunocompromised and severely ill patients — should not take probiotics without specific medical guidance, given documented elevated infection risk (NCCIH).
• Patients with central venous catheters — documented fungemia and bacteremia risk from catheter contamination or translocation; extreme caution with capsule handling near catheter sites is required if a probiotic is used at all under medical supervision (Case report).
• Patients with predicted severe acute pancreatitis or comparable critical illness — should not receive probiotics outside a controlled trial, given the PROPATRIA mortality signal (Besselink et al. 2008).
• Preterm infants, outside hospital-supervised protocols — the FDA has specifically warned against consumer or unsupervised use of probiotic products to prevent or treat NEC (FDA, 2023).
• Patients with recent abdominal surgery — identified as a significant risk factor for probiotic-associated invasive infection in case-control data (Tom et al. 2021).
• Anyone relying on a probiotic in place of an antibiotic or antifungal for an active infection — no probiotic strain is proven to eradicate an established bacterial or fungal infection on its own; per WGO, even the best-supported adjunct use (H. pylori) requires concurrent antibiotics.

Dosage and how to take probiotics

There is no single "probiotic dose" because dose is strain- and indication-specific, and CFU count alone does not guarantee a match to trial evidence unless the exact strain also matches. The doses below are the specific amounts studied in the trials described above — not general dosing advice for any probiotic product.

Practical takeaways that apply across indications:

• Match strain to indication first, dose second. A correct CFU count on the wrong strain does not replicate trial results.
• Storage matters. Many strains require refrigeration to remain viable through the labeled shelf life; heat and humidity during shipping and storage can reduce live-organism counts substantially before the product is even opened, an issue documented across multiple probiotic stability and shelf-life studies (Viability and storage stability study).
• "Billions of CFU" on a label is not a guarantee. As detailed in the Product quality section below, independent testing has found real-world products with far fewer live organisms than labeled.
• There is no established upper safe limit for CFU dose across probiotics as a category, because dose-response and toxicity have not been systematically studied the way they are for a regulated drug; studied doses in trials for a given strain and indication represent the only dose ranges with any evidence behind them.

Product quality reality check

Because probiotics are regulated as dietary supplements (or foods) rather than drugs in most markets, there is no requirement for pre-market proof of potency, no standardized CFU testing mandate, and frequently no requirement to identify strains beyond genus and species on the label.

Independent testing findings

ConsumerLab's most recent probiotic supplement review tested 30 products (including three marketed for pets). Twenty-five were "Approved," meaning they contained their labeled amount of live probiotic organisms (or more) and were not contaminated with pathogenic organisms. Four products could not be approved, mainly because they contained substantially lower amounts of viable probiotic cells than their labels claimed — in one case, only 14 million viable cells were found in a product presumably labeled for a much higher (billions-range) count (ConsumerLab Probiotic Supplements Review). Eight of the tested products went through ConsumerLab's voluntary Quality Certification Program, which is a paid program brands can opt into — a disclosed limitation worth flagging: ConsumerLab is subscription-funded and independent in its core testing model, but the certification program specifically creates a direct financial relationship between ConsumerLab and the brands seeking certification, which should be weighed when interpreting results for certified products specifically.

Shelf-life and viability

Independent stability studies of probiotic products have documented declining viable cell counts over the labeled shelf life, with the rate of decline dependent on strain, formulation (freeze-dried powder vs. liquid), storage temperature, and packaging (Viability, storage stability, and shelf life study). A product that met its labeled CFU count on the day it was manufactured may not meet that count by the time a consumer takes the last dose in the bottle, particularly if refrigeration guidance was not followed during shipping, retail storage, or home use.

Strain-level identification is often missing

Many commercial products list only genus and species (e.g., "Lactobacillus acidophilus") without a strain designator, making it impossible for a consumer to verify whether the specific organism in the bottle matches any clinical trial. As detailed throughout this article, the WGO and ISAPP both frame strain-level specificity as essential to interpreting probiotic evidence — a label lacking that information cannot be matched to evidence at all.

FDA enforcement actions

The FDA has issued warning letters to multiple probiotic manufacturers for making disease-treatment or disease-prevention claims that require drug-level approval rather than dietary-supplement-level marketing. Examples include a 2023 warning letter to Abbott Laboratories concerning unapproved claims on an infant-related probiotic product, and earlier action against Flora Inc. in 2021 (FDA Warning Letter, Abbott Laboratories, 2023; FDA Warning Letter, Flora Inc., 2021). Separately and more seriously, the FDA's September 2023 public health communication specifically targeted two companies illegally marketing probiotic products for use in preventing or treating disease in hospitalized preterm infants (FDA, 2023).

The VSL#3 / De Simone Formulation story: why brand names do not equal strains

This case is one of the clearest illustrations in the entire supplement industry of why matching the exact formulation — not just the brand name — to the clinical trial evidence is essential.

In the early 1990s, Professor Claudio De Simone developed a high-potency, eight-strain probiotic combination now referred to as the "De Simone Formulation." It was licensed to VSL Pharmaceuticals, Inc. and sold under the brand name VSL#3 from 2002 to 2016. Multiple clinical trials on pouchitis, ulcerative colitis, and IBS during this period used this specific formulation, manufactured in the United States.

After De Simone ended his business relationship with VSL Pharmaceuticals, the company continued selling a product under the same VSL#3 brand name, but manufactured in Italy using a different production process. A federal jury found Alfasigma USA (the subsequent distributor) liable for false advertising under the Lanham Act, ruling that the manufacturing know-how for the original De Simone Formulation was De Simone's exclusive property, and that the new VSL#3 product was materially different in composition and efficacy from the original. The jury also found that, as of trial, VSL Inc. had not completed a single bridging study demonstrating the new formulation performed the same as the original (Federal court summary, Visbiome). The original De Simone Formulation is now sold in the US and Canada under the brand name Visbiome.

The practical implication for anyone reading probiotic research: pre-2016 VSL#3 clinical trial results describe the De Simone Formulation, not the post-2016 VSL#3 product sold under the same name today. A brand name surviving a manufacturing and ownership change does not mean the clinical evidence survives with it. This is the "strain and formulation matter more than the label" thesis in its most concrete, legally-adjudicated form.

Industry funding and the incentive landscape

Per this article's evidence standard, industry-funded trials are used only to explain the funding-conflict story, never as primary evidence for a benefit claim. An academic analysis of probiotic clinical trial registries across Asia, Europe, and North America found that in Asia, non-industry-funded trials made up roughly 90% of the total by 2014; in Europe, industry funding reached parity with non-industry funding at approximately 45% around 2007; and in North America, industry funding intersected with public-private-partnership funding at roughly 20% around 2021. The same analysis notes that industry- and commercially-funded clinical trials "may be at risk of bias towards more positive results," while cautioning that this presumption should be weighed alongside the possibility that some trials labeled "non-industry funded" in registries may actually have partial industry funding not fully disclosed (Wiegers, van Beek & Larsen, 2023).

Named industry actors that repeatedly surface in the probiotics research and marketing landscape, and whose funded trials this article treats with caution or excludes from primary evidence, include:

• Danone (Actimel, DanActive, Activia brands) — withdrew several EFSA-reviewed immune-health claims for its yogurt drink products after unfavorable reviews of comparable industry claims (EFSA rulings reported by The Guardian, 2010).
• Yakult — submitted 12 studies to EFSA in support of an immune/cold-defense claim for its proprietary strain; EFSA found none sufficient to substantiate the claim (EFSA rulings reported by The Guardian, 2010).
• Chr. Hansen — a major probiotic ingredient supplier that funds trials on its own proprietary strains, including studies on gut damage protection and mental health/mood biomarkers; these are commercially sponsored and are treated here as context, not primary independent evidence.
• i-Health/DSM (Culturelle brand, L. rhamnosus GG) — the commercial licensee of the LGG strain funds and promotes consumer-facing research on its branded products; independent academic replications of LGG's core AAD and other benefits (cited in the Cochrane and WGO reviews above) are preferred over company-funded consumer studies wherever both exist.
• VSL Pharmaceuticals / Alfasigma (post-2016 VSL#3) — as detailed above, a federal jury found false advertising claims tied to this product; any current VSL#3-branded marketing referencing pre-2016 clinical trial results is treated with active skepticism given the adjudicated formulation change.

This does not mean every industry-funded probiotic trial is invalid — commercial sponsors fund a large share of all clinical nutrition research globally, including much of the foundational strain-identification work. It means results from commercially sponsored trials on a company's own proprietary strain should be weighed against, and ideally confirmed by, independent academic or Cochrane-level replication before being treated as decisive — exactly the standard applied throughout the "What works and what does not" section above.

Regulatory reality: what agencies actually allow companies to say

United States (FDA)

Probiotics sold as dietary supplements in the US fall under the Dietary Supplement Health and Education Act (DSHEA) framework: manufacturers can make structure/function claims (e.g., "supports digestive health") without FDA pre-approval, but cannot make disease-treatment or disease-prevention claims without going through a drug approval pathway. No probiotic strain currently holds FDA approval for a specific disease-treatment health claim. Live biotherapeutic products intended to treat or prevent a specific disease (such as the investigational L. crispatus vaginal products discussed above) are regulated as drugs, requiring the same clinical trial pathway as any pharmaceutical (Regulatory categorization review, Clinical Infectious Diseases). FDA enforcement — including the warning letters and 2023 preterm-infant safety communication detailed above — has focused specifically on products making disease-treatment claims that exceed what dietary-supplement status permits.

European Union (EFSA)

This is the clearest regulatory debunk story in the probiotics category. Starting around 2008, EFSA's Panel on Dietetic Products, Nutrition and Allergies reviewed thousands of health claims submitted across the EU food and supplement industry, ultimately assessing more than 800 in detail. It rejected essentially every general probiotic health claim put before it, including claims about immune system support and digestive health, on the grounds that the evidence submitted was either too vague to evaluate or did not demonstrate the claimed effect. Yakult's submission of 12 studies to support an immune-defense claim for its proprietary strain was found insufficient. Danone withdrew immune-health and digestion claims for Actimel and Activia from the review process after seeing other companies' similar claims rejected (The Guardian, 2010). A notable detail: EFSA reportedly avoids using the term "probiotic" itself in approved claim language, because the agency does not consider it a sufficiently precise, scientifically defined term for regulatory purposes. As of this writing, no general "probiotic" health claim has been approved by EFSA.

Health Canada

Health Canada takes a more structured, permissive approach than EFSA through its Natural Health Products (NHP) framework. It maintains a specific Probiotics Monograph that allows companies to make defined, strain-and-dose-specific claims (for example, digestive health claims tied to a specific documented strain and CFU range) through the Natural Health Product licensing process, rather than rejecting the category outright (Health Canada Probiotics Monograph).

WGO / ISAPP consensus

The World Gastroenterology Organisation and ISAPP function as the scientific consensus bodies for the field, rather than regulators. WGO's 2023 Global Guideline lays out strain-specific, indication-specific evidence in detail, forming the backbone of the "What works and what does not" section of this article (WGO Global Guideline, 2023). ISAPP publishes the field's core consensus definitions (probiotics, prebiotics, synbiotics, postbiotics, fermented foods) through academic expert panels; ISAPP membership includes both academic scientists and industry members, which is a disclosed, structural characteristic of the organization worth noting even though its published consensus definitions are broadly accepted across independent academic literature.

Infographics with full text versions

Infographic 1: The strain-matching checklist

Infographic 2: Evidence strength by indication

Infographic 3: Who needs medical guidance before taking any probiotic

Related research

This article is the strain-and-evidence deep dive behind Pure City Research's broader Gut Health Prevention Guide, which covers dietary and lifestyle strategies for digestive health beyond supplementation. Readers researching specific overlapping conditions may also find these guides relevant:

• Gut Health Prevention Guide — broader context on diet, fiber, and lifestyle factors in digestive health, including where probiotics fit relative to whole-food strategies.
• Diabetes Prevention Guide — relevant to the weak evidence on probiotics and metabolic/glucose markers discussed above.
• Eczema Prevention Guide — relevant to the mixed evidence on probiotics and skin conditions, including the perinatal atopic dermatitis prevention signal discussed above.
• Stress, Anxiety, and Depression Prevention Guide — relevant context for the preliminary "psychobiotics" evidence discussed above.
• Heart Disease Prevention Guide — relevant to the weak evidence on probiotics and lipid/cholesterol markers discussed above.

Frequently asked questions

Sources and funding notes

Industry-funded and conflicted sources — excluded from primary evidence or downgraded

Animal and non-human evidence excluded

In-vitro evidence used

No in-vitro (non-animal, non-human) evidence was used to support any benefit or safety claim in this article. Where mechanistic plausibility is mentioned (for example, general statements about competitive exclusion or short-chain fatty acid production), it is explicitly flagged as background context derived from broader field literature, not as evidence for any specific clinical claim, and is not used to fill a gap where human-trial evidence is absent.

Remaining uncertainties

• Exact dose-response relationships (minimum effective dose, ceiling for benefit, true upper safe limit) have not been systematically established for most strains, because probiotics are not regulated or studied the way conventional drugs are.
• Long-term (multi-year) safety data for most individual strains is limited; most trials run weeks to a few months.
• Direct human RCT evidence on several theoretically plausible drug interactions (antifungals with S. boulardii, warfarin with various strains, statins, antidiabetics) is limited to case reports or mechanistic inference rather than controlled trials — these gaps are stated explicitly in the interactions table above rather than treated as proof of safety.
• Guideline bodies (ACG vs. BSG on IBS; US FDA/AAP vs. UK NHS network on preterm NEC prevention) disagree on how to weigh the same underlying evidence base, reflecting genuine, unresolved scientific and risk-tolerance differences rather than a settled consensus.
• Independent, large-scale, systrain-matched market surveillance of CFU accuracy across the full range of commercial probiotic products is limited — the ConsumerLab data used in this article covers a sample of 30 products, not the entire market.