Probiotics: What Is Proven and What Is Just Sold

"Probiotics are good for you" is a sentence with no content. Probiotics are not a compound but hundreds of different microorganisms, and what works is a particular strain, for a particular condition, at a particular dose — a result from one strain says nothing about another, even a close relative. Decent evidence exists for a short list of narrow uses, including prevention of antibiotic-associated diarrhea and prevention of necrotizing enterocolitis in preterm infants, while "immune support", weight loss, and mood are sold far more confidently than they are supported.

Probiotic, prebiotic, synbiotic, postbiotic#

Four words that packaging routinely blurs, with a simple set of distinctions.

Probiotics are live microorganisms that, when administered in adequate amounts, confer a health benefit. Three parts of that definition do real work: live, adequate amount, and demonstrated benefit. A dead culture, or a live one below the studied dose, is an ingredient rather than a probiotic.

Prebiotics are not microbes at all but substrate: food components humans cannot digest that selectively feed resident bacteria. Inulin, fructooligosaccharides, galactooligosaccharides, and resistant starch are the main ones, and their usual source is not a jar but onions, garlic, legumes, oats, bananas, and cooled cooked potatoes.

Synbiotics combine the two in one product. The logic is sound, but the combination has to be studied as a combination — pairing two separately tested ingredients does not automatically add up their effects.

Postbiotics are the newest term: preparations of inactivated microorganisms or their metabolites. Nothing is alive, which changes both the risk profile and the storage requirements, but the evidence base is far thinner than for probiotics — at present they are often sold on the novelty of the word.

Why "probiotic" means nothing without a strain name#

A probiotic's full name has three parts: genus, species, strain. Lacticaseibacillus rhamnosus GG is genus Lacticaseibacillus (reclassified from Lactobacillus in 2020), species rhamnosus, strain GG. That last piece is not decoration and not a manufacturer's catalogue number.

Strains within a species differ genetically about as much as dog breeds differ in temperament. One L. rhamnosus strain survives bile and adheres to intestinal mucosa while another dies in the stomach; one produces compounds that suppress pathogens and another does not. A trial showing an effect for strain GG therefore tells you nothing about a product whose label reads only "Lactobacillus rhamnosus".

Strain specificity has a second, less comfortable consequence: results do not add up. An eight-strain blend is not a stronger version of a single-strain product — it is a different product that needs its own trials. Some multi-strain formulations genuinely have been studied as a unit, but most blends on the shelf were assembled by marketing departments rather than by investigators.

Where the evidence is decent#

Antibiotic-associated diarrhea#

This is the most solid indication in adults and children alike. Antibiotics wipe out part of the resident flora, and 5 to 30 percent of people develop diarrhea as a result. The 2019 Cochrane review of prevention in children (33 trials, roughly 6,350 participants) found moderate-certainty evidence of reduced diarrhea incidence, with the benefit concentrated in trials using higher doses — 5 billion CFU per day and above. Two strains dominate the literature: Saccharomyces boulardii CNCM I-745, which is a yeast and therefore unaffected by antibacterial drugs, and Lacticaseibacillus rhamnosus GG.

What was actually demonstrated is worth protecting: fewer episodes of diarrhea during a course of antibiotics — not "microbiome restoration", not liver protection, not fewer yeast infections. Timing matters too, since the trials started the probiotic alongside the antibiotic rather than after the course finished.

Clostridioides difficile diarrhea is a related but harder question. A 2017 Cochrane review found reduced risk in patients whose baseline risk was high, but guidelines read the data differently: the 2020 AGA guideline recommended specific formulations for C. difficile prevention while flagging low-certainty evidence and explicitly offering "no probiotic" as a reasonable alternative.

Pediatric acute gastroenteritis: the 2018 reversal#

For years this looked like the second strong indication. Meta-analyses of small trials suggested LGG and Limosilactobacillus reuteri DSM 17938 shortened diarrhea by roughly a day. Then, in November 2018, the New England Journal of Medicine published two large, well-designed, double-blind, placebo-controlled trials on the same day.

The American PECARN trial (Schnadower and colleagues, 971 children aged 3 months to 4 years) tested L. rhamnosus GG at 10 billion CFU twice daily against placebo. The Canadian PERC PROGUT trial (Freedman and colleagues, 886 children) tested a combination of L. rhamnosus R0011 and L. helveticus R0052. Both reported the same result: no difference from placebo in gastroenteritis severity, in duration of diarrhea or vomiting, or in hospitalization rates.

This is a textbook illustration of hypothesis testing going against the marketing: the early promise came from dozens of small trials of uneven quality, and two adequately powered trials failed to confirm it. In 2020 the ESPGHAN working group updated its position on probiotics in acute gastroenteritis, and the recommendations remained weak with low certainty of evidence — a direct consequence.

Irritable bowel syndrome#

Here the data are moderate and awkward for consumers. Individual strains such as Bifidobacterium longum 35624 reduced bloating and discomfort in trials, but the studies are small, the designs heterogeneous, and head-to-head comparisons between strains are almost nonexistent. The 2020 AGA guideline concluded that in IBS probiotics should be used only in the context of a clinical trial — not because they certainly fail, but because the evidence cannot support choosing one product over another.

A reasonable practical approach in IBS: pick one product with a named strain, run it for four weeks, and judge honestly. If nothing changes, stop rather than move on to the next jar.

Necrotizing enterocolitis in preterm infants#

The strongest evidence for probiotics sits in the neonatal intensive care unit rather than the pharmacy aisle. The 2023 Cochrane review by Sharif and colleagues on preventing necrotizing enterocolitis in very preterm or very low birth weight infants pooled more than a hundred trials and found reductions in both NEC and mortality. This is a condition with high mortality, and the intervention is a hospital protocol run by neonatologists using pharmaceutical-grade preparations. It has no bearing on buying a supplement for home use.

Condition	Best-studied strains	Strength of evidence
Necrotizing enterocolitis in preterm infants	Bifidobacterium plus Lactobacillus combinations, in hospital	Strongest for the class, physician-supervised only
Antibiotic-associated diarrhea prevention	S. boulardii CNCM I-745; L. rhamnosus GG	Moderate, doses from 5 billion CFU/day
C. difficile diarrhea prevention	Multi-strain formulations	Low, guidelines disagree
Pediatric acute gastroenteritis	L. rhamnosus GG; L. reuteri DSM 17938	Low: two large 2018 RCTs found no benefit
Irritable bowel syndrome	B. longum 35624 and others	Moderate for the class, no product can be named
Pouchitis after colectomy	Multi-strain blends	Moderate, narrow clinical setting
'Immune support' in healthy adults	—	Absent
Weight loss	—	Absent
Anxiety and depression ('psychobiotics')	—	Very low, trials are small
'Gut cleansing' and detox	—	Not a medical concept

Where the evidence is weak or absent#

Immune support. The most common claim on the box and the emptiest one. Scattered trials have shown modest reductions in the duration of respiratory infections, but effect sizes are small, strains vary, and study quality is poor. No professional body recommends probiotics to healthy people for infection prevention.

Weight loss. The microbiome does differ between people with and without obesity, but causality is unresolved and supplement trials produce weight changes within statistical noise — typically under a kilogram over several months, indistinguishable from the effect of being enrolled in a study at all.

Psychobiotics. The gut–brain axis is real and genuinely interesting science. The distance from "an axis exists" to "a capsule that treats anxiety" is enormous: human trials are small, short, use different strains, and rely on soft self-reported outcomes. Selling a probiotic for anxiety today is running about a decade ahead of the data.

Gut cleansing and post-antibiotic restoration. Neither is a medical concept. The adult microbiome is fairly resilient and largely recovers on its own over weeks to months after antibiotics, and there is no good evidence that a supplement meaningfully accelerates that. There is even work suggesting that taking probiotics after antibiotics can delay the return of a person's own flora.

Читайте также: Fiber: How Much You Need and Where to Get It

CFU, viability, and what the label should say#

CFU stands for colony-forming units — the number of live microorganisms capable of reproducing. Three things about that number matter.

When it was measured. Live cultures die in storage, especially at room temperature and in humidity. "10 billion CFU at time of manufacture" says nothing about what remains after a year on a shelf. Serious manufacturers state the count at end of shelf life; look for exactly that phrasing.

Whether it matches the studied dose. Dose is part of the definition of a probiotic. If the trials used 10 billion CFU twice daily, a capsule containing 1 billion is not a gentler version of the same thing — it is a different, untested regimen.

That bigger numbers are not automatically better. "50 billion CFU" sounds impressive, but the dose–response relationship is not linear and has not been characterized for most strains. A huge CFU count next to an unnamed strain is a marketing signal, not a quality one.

Regulation and product quality#

In most countries probiotics are sold as dietary supplements rather than drugs: manufacturers do not have to demonstrate efficacy before going to market, and oversight is retrospective and selective. Shelf quality therefore varies enormously, and independent testing keeps finding the same problems — fewer live organisms than the label claims, species present that are not on the label, occasional contaminants. Missing strain identifiers compound all of it, because a product that does not name its strains cannot be verified against anything.

Fermented foods as an alternative#

Kefir, live-culture yogurt, sauerkraut, kimchi, miso, tempeh, and kombucha all contain live microorganisms, but technically none of them are probiotics: the strains are uncharacterized and the dose is not standardized, varying from batch to batch.

Even so, fermented foods look like a sensible default: they deliver protein, calcium, and B vitamins alongside the microbes, and fermented vegetables add fiber. A small randomized Stanford study published in Cell in 2021 (Wastyk and colleagues) found that a diet high in fermented foods increased microbiome diversity and lowered inflammatory markers more than a high-fiber diet did. That is one trial and not a basis for therapeutic claims, but it is an interesting direction.

The practical conclusion is unglamorous. Absent a specific goal such as preventing antibiotic-associated diarrhea, a daily serving of kefir or sauerkraut is a better investment than a jar of unnamed strains. And the single largest determinant of microbiome health is not probiotics at all — it is the quantity and variety of plants in the diet.

Читайте также: Magnesium: Forms, Doses, and Who Actually Needs It

Safety#

In healthy people probiotics are well tolerated. The usual complaints are bloating and gas in the first few days, sometimes looser stools, which resolve on their own or with a lower dose. But the assumption that probiotics are harmless by definition is wrong, and in some groups the risk is concrete.

Pregnancy and breastfeeding are a separate caveat. There is no evidence of harm, but there is also little high-quality research, so the decision belongs in a conversation with a clinician rather than with a "natural ingredients" label.

Читайте также: Vitamin D: Doses, Blood Levels, and Who Actually Needs It

What to do in practice#

1. Start with "why", not "which". Without a defined goal — diarrhea prevention during antibiotics, IBS, a physician's recommendation — there is no reason to buy a probiotic at all.
2. Match the strain to the goal, not the product to the mood. For an antibiotic course: S. boulardii CNCM I-745 or L. rhamnosus GG at 5–10 billion CFU per day, started on day one.
3. Check for the alphanumeric strain code. No code, no product.
4. Read CFU at end of shelf life and compare it against the dose used in trials.
5. Set a review date. Four weeks for IBS; for diarrhea prevention, the outcome is simply whether diarrhea happened. Do not keep "trying" indefinitely.
6. With immunocompromise, a central line, pancreatitis, serious illness, or a preterm infant, go through a physician.

Probiotics are a useful window into how the supplement market works. There is real science here, a handful of narrow indications with respectable data, and an entire industry that sells the generalization instead of the specifics. The difference between a worthwhile purchase and a pointless one usually fits on one line of the label — the line with the strain on it.