Why probiotic research rarely applies to the bottle you bought

A trial finds that a probiotic cuts the risk of a serious antibiotic-related infection by two-thirds. You read about it, you buy a bottle, and the label says Lactobacillus rhamnosus. The trial used Lactobacillus rhamnosus GG. Those two letters are the entire study.

That gap is the most useful thing to understand about this category. Probiotic evidence attaches to individual strains, not to species and not to genera. A result for one strain says nothing reliable about a different strain carrying the same two-word Latin name, and most labels do not tell you which one is inside.

Almost everything else follows from that. It explains why colony-forming unit counts are harder to read than they look, why federal agencies and medical societies can study the same literature and land in different places, and why two infants who died after receiving probiotic products in hospital died for two completely different reasons.

The strain is the thing that was tested

The National Institutes of Health Office of Dietary Supplements describes probiotic identity in three parts: the genus, the species, and an alphanumeric strain designation. Genus and species are the Latin binomial everyone recognizes. The strain designation is the code tacked on the end, and it names one specific isolate descended from one original culture.

Think of it the way you would think of dog breeds. “Dog” tells you very little about temperament. The breed tells you a great deal. Within a probiotic species, different strains differ in what they survive, what they stick to, what they secrete and what they do to a person.

A 2015 analysis in Beneficial Microbes showed what that does to pooled research. Combining 16 different probiotic types for prevention of antibiotic-associated diarrhea in children produced a relative risk of 0.43. Restricting the same analysis to a single strain at a time produced 0.43 for one yeast strain and 0.44 for one bacterial strain, each with far wider confidence intervals. The pooled figure was an average across products that are not substitutes for one another.

The label problem is measurable. A 2019 study in PLoS One cultured organisms from 21 commercial US probiotic supplements and beverages. Sequencing matched the species on the label 82 percent of the time, and two other identification methods did worse, at 71 and 60 percent. More to the point, products listing the same species often contained strains with different genetic sequences and different behavior under salt, acid and antibiotic stress. The authors concluded that current labels do not convey the strain diversity present.

This is why a sentence of the form studies show probiotics help with X is close to meaningless on its own. The honest version names the strain, the dose, the population and the outcome, and that version is usually much narrower than the headline.

What the FDA has and has not done

Most probiotics sold in the United States are dietary supplements, and that classification decides nearly everything about how they are reviewed. The FDA’s consumer questions-and-answers page on dietary supplements says that the law does not give the agency the job of approving supplements or their labeling. Manufacturers and distributors carry initial responsibility for safety, a firm making a claim must hold substantiation that the claim is truthful and not misleading, and the agency describes itself as generally limited to postmarket enforcement. In practice that means the product reaches the shelf first and the federal interest begins afterward.

So there is no FDA approval of a probiotic supplement for preventing or treating any condition, because approval is not the pathway these products travel. A probiotic sold as a drug would have to prove safety and effectiveness in clinical trials first. One given in a hospital to treat or prevent disease requires an Investigational New Drug application, a point the FDA restated in 2023.

The agency has said the approval point most directly about babies. In its 2023 safety communication, the FDA wrote: “FDA has not approved any probiotic product for use as a drug or biological product in infants.”

The American Academy of Pediatrics made the same point from the clinical side in a 2021 report in Pediatrics. Because these products lack FDA-approved drug labeling, they cannot be marketed to treat or prevent disease in preterm infants, including necrotizing enterocolitis and late-onset sepsis.

Where the evidence holds up, and where it does not

Antibiotic-associated diarrhea

This is the strongest area, and it is narrower than it sounds. A 2012 meta-analysis in Annals of Internal Medicine pooled 20 randomized trials and 3,818 participants taking antibiotics and found probiotics reduced Clostridium difficile-associated diarrhea by 66 percent, a relative risk of 0.34, with no measurable variation between trials. Translated into people, the authors wrote that in a population with a 5 percent incidence, prophylaxis would prevent 33 episodes per 1,000. Adverse events were 9.3 percent on probiotics against 12.6 percent on control. The authors graded the evidence moderate quality, noted that 13 of the 20 trials were missing outcome data for 5 to 45 percent of patients, and reported no funding source.

For ordinary antibiotic-associated diarrhea, two large syntheses disagree about size in an instructive direction. A 2012 meta-analysis in Alimentary Pharmacology and Therapeutics found a relative risk of 0.53 across 34 trials and 4,138 patients. A 2021 meta-analysis in the Journal of Clinical Gastroenterology found 0.62 across 36 trials and 9,312 participants. The larger, newer dataset produced the smaller effect, which is the pattern a real but modest effect usually makes as evidence accumulates.

Quality is uneven underneath these numbers. A 2013 review of 16 trials in hospitalized adults rated four good, five fair and seven poor.

One affirmative guideline recommendation

In its 2024 guideline on pouchitis, the American Gastroenterological Association wrote: “In patients who experience recurrent episodes of pouchitis that respond to antibiotics, the AGA suggests using probiotics for the prevention of recurrent pouchitis.”

Pouchitis is inflammation of the internal pouch surgeons build from small intestine after removing the colon, usually for ulcerative colitis. All nine recommendations in that guideline are conditional, meaning the panel judged that reasonable patients would choose differently from one another.

Where the societies declined

The AGA’s 2020 probiotics guideline is mostly a map of what is not known. For treating C. difficile infection, for Crohn’s disease, for ulcerative colitis and for irritable bowel syndrome, the panel issued no recommendation and labeled each a knowledge gap, suggesting probiotics only inside a clinical trial for the two inflammatory bowel diseases. For acute infectious gastroenteritis in children, the guideline points the other way, with a conditional recommendation against probiotics at moderate certainty.

The National Center for Complementary and Integrative Health reaches similar conclusions for irritable bowel syndrome, citing a 2018 review of 53 studies that could not draw definite conclusions about effectiveness or identify which species or strains work best.

Two infant deaths, two different failures

These are separate events two years apart, and the difference between them is the point.

In 2014, an infant born at 29 weeks’ gestation weighing 1,400 grams received a probiotic powder for seven days starting on the first day of life and died of gastrointestinal mucormycosis, a fungal infection that invades blood vessels in the gut wall. The Centers for Disease Control and Prevention reported the case in MMWR on 20 February 2015. Investigators sequenced fungal DNA from the infant’s tissue and identified Rhizopus oryzae. They then cultured unopened bottles from the lot the infant had received and recovered the same species. The manufacturer recalled several lots on 14 November 2014. No additional cases of gastrointestinal mucormycosis in newborns were identified.

That is a contamination failure. A mold that was never meant to be in the product was in the product.

In 2023, the FDA reported that a preterm infant given a probiotic formulated with live bacteria during in-hospital care developed sepsis caused by Bifidobacterium longum and died. B. longum was the species the product was formulated to deliver. The agency issued warning letters to two manufacturers in late September and October 2023 and published a press announcement on 26 October 2023. One product was voluntarily recalled and the other manufacturer agreed to discontinue US sales.

That is not a contamination failure. The product did exactly what it was built to do, and the organism it delivered crossed from the gut into the bloodstream of a patient whose defenses were not finished developing.

The distinction matters because the fixes are different. Better manufacturing and testing address the first problem. Nothing in manufacturing addresses the second, because there is no way to quality-control a living organism out of behaving like a living organism inside a vulnerable patient.

The necrotizing enterocolitis problem has no clean answer

Necrotizing enterocolitis is a condition in which part of a premature baby’s intestine becomes inflamed and tissue dies. It is among the most feared events in a neonatal intensive care unit, and it kills.

The evidence base says probiotics reduce it. A network meta-analysis published in JAMA Pediatrics in November 2023 pooled 106 randomized trials and 25,840 preterm infants. Multiple-strain probiotics were the only intervention associated with reduced all-cause mortality against placebo, a relative risk of 0.69, and were among the most effective at reducing severe necrotizing enterocolitis, a relative risk of 0.38. The authors rated the mortality finding moderate to high certainty.

The AGA reached the same conclusion in 2020, issuing a conditional recommendation in favor of specified strains for preventing necrotizing enterocolitis in preterm, low-birth-weight infants, at moderate to high certainty. That is the strongest certainty rating anywhere in that guideline.

And the FDA warned in 2023 that preterm infants given probiotics are at risk of invasive, potentially fatal disease.

Both halves are true. The trials and the regulator are not disputing the data. They are weighing a measured average benefit across tens of thousands of infants against an individual catastrophe that cannot be predicted, in a product category with no premarket review of safety, potency or purity. The American Academy of Pediatrics landed between them in 2021, concluding that current evidence does not support routine, universal administration to preterm infants, particularly those under 1,000 grams, while also reporting that roughly 10 percent of extremely low gestational age neonates in the United States and Canada already receive one, with wide variation between units.

None of this is a pharmacy-shelf decision. It is a decision made inside a NICU by people who know the individual baby, and the useful question for a family is what their unit’s practice is and what reasoning sits behind it.

Who gets hurt, and how often

Two kinds of evidence answer two different questions here, and both belong in the picture.

Case reports establish that harm happens. A 2018 systematic review in BMC Complementary and Alternative Medicine collected 60 case reports and 7 case series covering 93 patients with infections following probiotic ingestion. Fungemia, fungus in the bloodstream, was the most common complication at 35 cases, or 37.6 percent. By organism, Saccharomyces accounted for 47 cases, Lactobacillus for 26 and Bifidobacterium for 12. The authors concluded that probiotic use cannot be considered risk-free and should be evaluated carefully for some patient groups.

A 2023 review in Medical Mycology examined 117 cases of Saccharomyces fungemia reported between 2005 and 2022 and analyzed 108. Just over two-thirds had received Saccharomyces boulardii probiotics. Among those cases, intensive care stay featured in 31.5 percent, tube or intravenous feeding in 29.6 percent and diabetes in 13 percent. All-cause mortality across the whole cohort was 36.1 percent. A separate 2023 review in Microorganisms reported that three probiotic strains have been linked by molecular identification to blood isolates taken from patients with bacteremia.

Case reports have no denominator, so they cannot tell you how often. One study can. A 2021 analysis in Mycoses reviewed 46,729 adult inpatients at a US hospital between 2016 and 2019, of whom 16,404 received S. boulardii. Eighteen of them, 0.11 percent, developed Saccharomyces cerevisiae fungemia. The incidence worked out to 1.70 cases per 10,000 patient-days and 0.26 per 1,000 central-line days. Intensive care admission raised the odds more than sixfold. The authors concluded that the risk did not appear greater than the risk of any hospital-acquired bloodstream infection.

Rare and real are both correct. The Office of Dietary Supplements states that probiotics may cause harm in people who are severely ill or immunocompromised, and the National Center for Complementary and Integrative Health says the risk of harmful effects is greater in people with severe illnesses or compromised immune systems.

That translates into specific situations rather than a general warning. If you have a central venous catheter, are receiving chemotherapy or drugs that suppress the immune system, have had an organ transplant, are in intensive care, or are being fed through a tube or a vein, whether live organisms belong in your gut is a question for the clinician managing that care. A new fever during a hospital stay in someone receiving a probiotic is information the treating team needs. Probiotics can also interact with medications, which article 677 covers in full.

What the label can and cannot tell you

Colony-forming units, abbreviated CFU, count viable cells. The Office of Dietary Supplements reports that many probiotic supplements contain 1 to 10 billion CFU per dose, while some contain 50 billion or more, and states plainly that products with higher CFU counts are not necessarily more effective than products with lower ones.

Viability is the catch. Because probiotics must be consumed alive to do anything, organisms can die during shelf life. The Office of Dietary Supplements advises readers to look for a CFU figure guaranteed at the end of shelf life rather than at time of manufacture, and to check storage conditions, since some products need refrigeration and others do not.

Whether products contain what they claim has a long research record, and it has improved. A 2002 study in the Journal of Dairy Science sequenced 58 strains from a bank of commercial probiotic cultures and found species discrepancies in 26. Among 29 strains obtained directly from manufacturers, 14 carried a species designation different from the sequencing result, and strains in six commercial products were species not listed on the label. A 2006 study in Research in Microbiology examined 213 cultures from 26 manufacturers and found more than 28 percent of commercial cultures intended for probiotic use misidentified at genus or species level, involving 34 cultures from 10 companies.

More recent work is better. The 2019 PLoS One study matched labels 82 percent of the time by sequencing. A 2025 study in Pediatric Research analyzed 17 infant probiotic products bought from a large online vendor and found that over 82 percent of the bacterial DNA corresponded to species on the label, with contaminating bacteria being closely related species rather than pathogens. Single-strain products reliably matched their labels. Multi-strain products varied more, both in the relative abundance of species and in leaving species out. Fungal DNA turned up in some samples, but cultures showed those organisms were not alive.

So here is what a label can give you, and what it cannot.

• The full strain designation. Genus, species and the alphanumeric code after it. Without the code, you cannot find the research.
• Whether the CFU figure is guaranteed through the expiration date or only at the time it was made. These are different promises.
• Storage instructions. Viability is the entire mechanism, and refrigeration requirements are a signal about it.
• How many strains are in the product, and whether each one’s count is listed separately or only as a blend total. A blend total tells you nothing about how much of any single strain you are getting.
• Whether that exact strain, with its code, appears in published human trials for the specific outcome you read about.
• What it usually cannot tell you is whether the organism in the bottle is the one on the label. That question has been answered by sequencing studies, not by packaging. Third-party verification marks are covered in article 326.

What the strain code is for

The alphanumeric code is the only thread connecting a product on a shelf to a study in a journal. A species name attached to a billion-CFU number describes a product. It does not describe evidence.

Where the evidence is strongest, it is also narrow: preventing one specific infection during a course of antibiotics, and preventing recurrence of one specific complication after one specific surgery. Both carry conditional recommendations and certainty ratings of moderate or lower.

And the population where measured benefit looks largest, premature infants, is the same population where documented harm has been worst. That is not a coincidence. An immature gut that lets a helpful organism in is the same gut that lets it into the blood.

Sources

1. U.S. Food and Drug Administration. “Risk of Invasive Disease in Preterm Infants Given Probiotics Formulated to Contain Live Bacteria or Yeast (content current as of 09/29/2023).” fda.gov
2. U.S. Food and Drug Administration. “FDA Raises Concerns About Probiotic Products Sold for Use in Hospitalized Preterm Infants (October 26, 2023).” fda.gov
3. U.S. Food and Drug Administration. “Questions and Answers on Dietary Supplements (content current as of 08/27/2026).” fda.gov
4. NIH Office of Dietary Supplements. “Probiotics: Fact Sheet for Health Professionals (updated March 25, 2025).” ods.od.nih.gov
5. National Center for Complementary and Integrative Health. “Probiotics: What You Need To Know.” nccih.nih.gov
6. Centers for Disease Control and Prevention, MMWR. “Vallabhaneni S et al. "Fatal Gastrointestinal Mucormycosis in a Premature Infant Associated with a Contaminated Dietary Supplement — Connecticut, 2014." MMWR 2015;64(6):155, February 20, 2015.” cdc.gov
7. PubMed / Gastroenterology (American Gastroenterological Association). “Su GL et al. "AGA Clinical Practice Guidelines on the Role of Probiotics in the Management of Gastrointestinal Disorders." Gastroenterology, 2020;159(2):697-705. DOI: 10.1053/j.gastro.2020.05.059.” pubmed.ncbi.nlm.nih.gov
8. PubMed / Gastroenterology (American Gastroenterological Association). “Barnes EL et al. "AGA Clinical Practice Guideline on the Management of Pouchitis and Inflammatory Pouch Disorders." Gastroenterology, 2024;166(1):59-85. DOI: 10.1053/j.gastro.2023.10.015.” pubmed.ncbi.nlm.nih.gov
9. PubMed / Pediatrics (American Academy of Pediatrics). “Poindexter B et al. "Use of Probiotics in Preterm Infants." Pediatrics, 2021;147(6). DOI: 10.1542/peds.2021-051485.” pubmed.ncbi.nlm.nih.gov
10. PubMed / Annals of Internal Medicine. “Johnston BC et al. "Probiotics for the prevention of Clostridium difficile-associated diarrhea: a systematic review and meta-analysis." Ann Intern Med, 2012;157(12):878-88. DOI: 10.7326/0003-4819-157-12-201212180-00563.” pubmed.ncbi.nlm.nih.gov
11. PubMed / Alimentary Pharmacology & Therapeutics. “Videlock EJ, Cremonini F. "Meta-analysis: probiotics in antibiotic-associated diarrhoea." Aliment Pharmacol Ther, 2012;35(12):1355-69. DOI: 10.1111/j.1365-2036.2012.05104.x.” pubmed.ncbi.nlm.nih.gov
12. PubMed / Journal of Clinical Gastroenterology. “Liao W et al. "Probiotics for the Prevention of Antibiotic-associated Diarrhea in Adults: A Meta-Analysis of Randomized Placebo-Controlled Trials." J Clin Gastroenterol, 2021;55(6):469-480. DOI: 10.1097/MCG.0000000000001464.” pubmed.ncbi.nlm.nih.gov
13. PubMed / Open Medicine. “Pattani R et al. "Probiotics for the prevention of antibiotic-associated diarrhea and Clostridium difficile infection among hospitalized patients: systematic review and meta-analysis." Open Med, 2013;7(2):e56-67.” pubmed.ncbi.nlm.nih.gov
14. PubMed / Beneficial Microbes. “McFarland LV. "Deciphering meta-analytic results: a mini-review of probiotics for the prevention of paediatric antibiotic-associated diarrhoea and Clostridium difficile infections." Benef Microbes, 2015;6(2):189-94. DOI: 10.3920/BM2014.0034.” pubmed.ncbi.nlm.nih.gov
15. PubMed / JAMA Pediatrics. “Wang Y et al. "Probiotics, Prebiotics, Lactoferrin, and Combination Products for Prevention of Mortality and Morbidity in Preterm Infants: A Systematic Review and Network Meta-Analysis." JAMA Pediatr, 2023;177(11):1158-1167. DOI: 10.1001/jamapediatrics.2023.3849.” pubmed.ncbi.nlm.nih.gov
16. PubMed / BMC Complementary and Alternative Medicine. “Costa RL et al. "Infectious complications following probiotic ingestion: a potentially underestimated problem? A systematic review of reports and case series." BMC Complement Altern Med, 2018;18(1):329. DOI: 10.1186/s12906-018-2394-3.” pubmed.ncbi.nlm.nih.gov
17. PubMed / Medical Mycology. “Vinayagamoorthy K et al. "Epidemiology of Saccharomyces fungemia: A systematic review." Med Mycol, 2023;61(2). DOI: 10.1093/mmy/myad014.” pubmed.ncbi.nlm.nih.gov
18. PubMed / Microorganisms. “Kullar R, Goldstein EJC, Johnson S, McFarland LV. "Bacteremia and Probiotics: A Review." Microorganisms, 2023;11(4):896. DOI: 10.3390/microorganisms11040896.” pubmed.ncbi.nlm.nih.gov
19. PubMed / Mycoses. “Wombwell E et al. "Incidence of Saccharomyces cerevisiae fungemia in hospitalised patients administered Saccharomyces boulardii probiotic." Mycoses, 2021;64(12):1521-1526. DOI: 10.1111/myc.13375.” pubmed.ncbi.nlm.nih.gov
20. PubMed / Journal of Dairy Science. “Yeung PSM et al. "Species-specific identification of commercial probiotic strains." J Dairy Sci, 2002;85(5):1039-51. DOI: 10.3168/jds.S0022-0302(02)74164-7.” pubmed.ncbi.nlm.nih.gov
21. PubMed / Research in Microbiology. “Huys G et al. "Accuracy of species identity of commercial bacterial cultures intended for probiotic or nutritional use." Res Microbiol, 2006;157(9):803-10. DOI: 10.1016/j.resmic.2006.06.006.” pubmed.ncbi.nlm.nih.gov
22. PubMed / PLoS One. “Ansari JM et al. "Strain-level diversity of commercial probiotic isolates of Bacillus, Lactobacillus, and Saccharomyces species illustrated by molecular identification and phenotypic profiling." PLoS One, 2019;14(3):e0213841. DOI: 10.1371/journal.pone.0213841.” pubmed.ncbi.nlm.nih.gov
23. PubMed / Pediatric Research. “DeVette CI et al. "Genomic interrogation of commercial infant probiotic products for label accuracy." Pediatr Res, 2025;98(1):250-255. DOI: 10.1038/s41390-025-03831-6.” pubmed.ncbi.nlm.nih.gov