Story
Probiotics
Last updated
In one pass Most bottles on the shelf that say contains probiotics do not actually earn the name. Not this — Probiotics are universally good — any strain helps the gut — After antibiotics, a generic probiotic actually delayed the gut microbes' recovery by 3-6 months (Suez 2018, Cell). What matters is matching the strain, the condition and the dose.
Educational content, not medical advice — consult a clinician.
Story path
Chapter 1
What counts as a probiotic
Most bottles on the shelf that say contains probiotics do not actually earn the name.
A probiotic is a word with a threshold, and the threshold has four gates: the microbes must be alive, there must be enough of them, you must be able to say which strain it is, and that strain must have been tested in people for the very problem you want solved. Miss one gate and it is just a jar of ordinary bacterial powder.
The least intuitive gate is the strain. Different strains under the same species are like two strangers who share a surname: same name, completely different skills. Some can cling to the layer of mucus on the gut wall and some slide straight past; some can stand bile and some fall apart the moment they reach the small intestine. So a label that only says Lactobacillus has told you nothing. Before you buy, look on the bottle for the strain code (for example *Lactobacillus rhamnosus* GG, known as LGG). If you cannot find one, put it back.
A few groups should not buy them to take on their own: preterm infants, people who are critically ill or whose immune system is severely weakened, and anyone with a central venous line. In them, live bacteria can get into the blood. If fever or chills start while taking a probiotic, seek medical care immediately.
A probiotic is a word with a threshold, and the threshold has four gates: the microbes must be alive, there must be enough of them, you must be able to say which strain it is, and that strain must have been tested in people for the very problem you want solved. Miss one gate and it is just a jar of ordinary bacterial powder.
The least intuitive gate is the strain. Different strains under the same species are like two strangers who share a surname: same name, completely different skills. Some can cling to the layer of mucus on the gut wall and some slide straight past; some can stand bile and some fall apart the moment they reach the small intestine. So a label that only says Lactobacillus has told you nothing. Before you buy, look on the bottle for the strain code (for example *Lactobacillus rhamnosus* GG, known as LGG). If you cannot find one, put it back.
A few groups should not buy them to take on their own: preterm infants, people who are critically ill or whose immune system is severely weakened, and anyone with a central venous line. In them, live bacteria can get into the blood. If fever or chills start while taking a probiotic, seek medical care immediately.
Background · The four gates in the definition
The official definition is a single sentence, and taken apart it works as a sieve.The definition given by the World Health Organization and the Food and Agriculture Organization (WHO/FAO) in 2002, and restated by the International Scientific Association for Probiotics and Prebiotics (ISAPP) in 2014, is: live microorganisms that, when administered in adequate amounts, confer a health benefit on the host. There are four words in that sentence, and each one screens out a batch of commercial products.
① Adequate amounts: the number on a label is given in CFU, colony-forming units — the count of live cells that could each grow into a colony in a dish. Clinical trials commonly use 10⁹–10¹⁰ CFU a day. The count at manufacture is not the count at the end of shelf life: most strains lose viability over time and need refrigeration, dryness or an enteric coating to hold on. Spot checks have reported labels claiming 10 billion CFU with only 1–10% left by the end of shelf life.
② The strain, not the species: the species is *Lactobacillus rhamnosus*; the strain is *Lactobacillus rhamnosus* GG (LGG). Clinical effects follow the strain. LGG has been tested in people for preventing eczema in babies from allergy-prone families, while another strain of the same species may do nothing at all. LGG's most famous former use, shortening acute diarrhea in children, no longer holds: the pooled result in the 2020 Cochrane review crosses zero, and the American Gastroenterological Association (AGA) 2020 guideline advises against using it that way. When a seller says only contains probiotics and not which strain, it usually means that strain has no clinical data to show.
③ Confers a health benefit on the host: this needs (RCTs, trials that assign people at random to take it or not and compare the groups) in people, for that strain and that use. Not beneficial in theory, not a nicer-looking marker, but a specific disease or symptom that actually improves in a clinically meaningful way.
④ Live: dead cells are not probiotics by definition, although the idea of postbiotics (inactivated cells and the products bacteria make) is gaining ground. That is why fermented food loses its live-culture share once it is stirred into boiling water, mixed into an acidic drink or cooked at high heat.
To judge the bottle at home, check three things:
1. Does it name a specific strain (LGG, BB-12, or DSM followed by a number)?
2. Has that strain been tested in randomized trials in people for the problem you want solved?
3. Is the CFU on the label the minimum at the end of shelf life, or the number at manufacture?
Mechanism · Why only the strain carries information
Why is the species level not enough? Because the abilities that decide whether a strain can do anything are written into the genes at the strain level, and two strains of the same species can differ in having them or not.At least four abilities split this way:
Can it hold on: some proteins sticking out from the cell surface work like small hooks, and whether the strain can catch the mucus layer on the gut wall depends on whether it carries them. A strain that cannot hold on does not last long in the current that keeps pushing gut contents downstream.Can it stand bile: the bile salts in bile act like a detergent and dissolve the lipids in cell membranes. Strains that can break bile salts down first survive; the rest fall apart at the start of the small intestine.Can it take ground: some strains release small molecules that kill their neighbors, and some do not. That decides whether a strain can squeeze out a little room where every place is already taken.How the immune system reads it: the polysaccharide coat on the outside of the cell wall comes in different shapes, so the immune cells beneath the gut lining pick up different signals. Of two strains of the same species, one may tilt the immune response toward calm while the other triggers nothing.
These differences can be measured in the lab; how much each one contributes in people is mostly still unclear. The direction is clear, though: the sentence lactobacilli are beneficial carries almost no information. It names a surname, not a person.
This also explains two common facts. First, a clinical trial has to start over when the strain changes, because the last strain's hooks, bile handling and ways of taking ground may be missing from the next. Second, moving one strain's evidence onto another strain of the same species (the most common swap on the shelf) does not hold up: what you are buying is that code, not the genus name.
One more practical inference follows. A product that will not print a strain code is not short of room on the label; once printed, the code has to answer has this strain ever been tested. Being unable to name the strain is itself the answer.
Evidence · Strains with human trials behind them
The strains below have been tested in people for specific uses. The list is not complete; it covers the ones people ask about most, and each entry says how much evidence there is.Lactobacillus:
***L. rhamnosus* GG (LGG)**: in a randomized trial in Finland (Kalliomäki 2001, published in *The Lancet*), pregnant women with a family history of allergy took LGG before birth and their babies took it for 6 months after, and by age two the rate of eczema was roughly halved (23% vs 46%). It is a single trial and still needs to be replicated by others. LGG also used to be the poster strain for treating acute diarrhea in children, and that use no longer holds: the 2020 Cochrane review pooled 6 trials and 3,058 participants and found diarrhea only 8.6 hours shorter on average, with a 95% (the range the true effect most likely falls in) running from 29 hours shorter all the way to 12 hours longer, at very low certainty. An interval that crosses zero has not shown an effect. The American Gastroenterological Association (AGA) 2020 guideline therefore suggests against probiotics for children with acute infectious gastroenteritis (a conditional recommendation)The GR-1 plus RC-14 combination (*L. rhamnosus* GR-1 and *L. reuteri* RC-14): an add-on for women with recurrent urinary tract or vaginal infections; the evidence comes from a few trials and the results are not fully consistent***L. reuteri* DSM 17938**: infant colic (long bouts of crying in a baby who is hard to settle); some randomized trials are positive***L. casei* Shirota (the strain in Yakult)**: general gut function; the marketing goes far beyond the clinical evidence
Bifidobacterium:
***B. lactis* BB-12 and DN-173 010**: general gut motility, with small improvements in some symptoms of irritable bowel syndrome ()***B. infantis* 35624 (Align)**: has positive results from small randomized trials in IBS, at low certainty. Ford 2018, a published in *Alimentary Pharmacology & Therapeutics*, pooled more than fifty probiotic trials and concluded that certain combinations or strains appeared to help overall IBS symptoms and abdominal pain, but that which ones work remains, for the most part, unclear. Neither of the two US societies recommends them: the AGA 2020 guideline advises using probiotics for IBS only within a clinical trial, and the American College of Gastroenterology (ACG) 2021 IBS guideline suggests against probiotics for overall symptoms (a conditional recommendation, very low quality of evidence). So do not describe it as "guideline-recommended"
Saccharomyces (a yeast, not a bacterium):
***S. boulardii* CNCM I-745: for preventing antibiotic-associated diarrhea (AAD, diarrhea that starts during or after a course of antibiotics) in children**. Guo 2019, a Cochrane review, pooled trials in children aged 0-18: 8% of the probiotic groups developed diarrhea versus 19% of controls (, RR, 0.45, meaning the risk fell to about half that of controls; roughly one case avoided for every 9 children treated; moderate certainty). The result pools many different probiotics; the review's authors single out high doses of *L. rhamnosus* or *S. boulardii* as the ones most worth confirming in a large trial. The adult case rests on something else: the AGA 2020 guideline gives a conditional recommendation for **preventing *C. difficile* infection in adults and children taking antibiotics**, and *S. boulardii* is among the strains it names. The evidence for travelers' diarrhea is not strong enough to put it on this listBecause it is a yeast, antibiotics cannot kill it, so it can be taken at the same time as the antibiotic
Multi-strain products:
VSL#3 (now sold as Visbiome or Vivomixx): an 8-strain combination. The AGA 2020 guideline gives it a conditional recommendation for pouchitis (inflammation of the pouch built from the small intestine after the colon is removed); for ulcerative colitis and Crohn's disease it advises use only within a clinical trialAssorted children's multi-strain products: mixed data
Categories without real evidence:
Most digestive health probiotic blends on supermarket shelves, labeled only as a mix of lactobacilli and bifidobacteria, with no specific strain namedThe marketing line "yogurt with active probiotics": ordinary yogurt is fermented with *Lactobacillus bulgaricus* and *Streptococcus thermophilus*, which mostly do not reach the colon. They help people with lactose intolerance digest lactose, but they are not probiotics validated by the definition. Shelf-stable yogurt that is heat-treated after fermentation contains no live cultures at allPrebiotics (the fibers that feed bacteria) feed the bacteria you already have. That is a different route, and it does not have to run the gauntlet of stomach acid
Before you buy, ask yourself three things: which strain is it (only consider it if the strain code is written); has that strain been tested in randomized trials in people for your use (5 minutes on PubMed); and is the label's CFU a refrigerated, end-of-shelf-life figure or the number at manufacture.
kalliomaki-2001-lancetlacy-2021-acg-ibs
Chapter 2
How many swallowed bacteria survive
The number of live bacteria you swallow and the number that reach the colon still able to do anything are two different numbers.
The bacteria first have to pass two gates: the strong acid in the stomach, and then bile in the small intestine, which acts like a detergent. How many get through depends on the strain and the formulation. For strains specifically selected for acid resistance, an estimated 20 to 40 percent survive the stomach; ordinary strains that were never selected do far worse. Those that make it through almost never settle in the colon: the places on the wall and the food in the gut are already taken by your own resident bacteria, so the newcomers cannot get a hold and are carried out with the gut contents. A few weeks after you stop taking them, they usually can no longer be found in stool.
So a probiotic is more like a passing temp worker than a tenant moving in.
The odd thing is that a passer-by that cannot make a home can still spare children on antibiotics some diarrhea: in the pooled trials, diarrhea fell from 19% to 8%. What it does while passing through starts with what antibiotics do to the gut.
The bacteria first have to pass two gates: the strong acid in the stomach, and then bile in the small intestine, which acts like a detergent. How many get through depends on the strain and the formulation. For strains specifically selected for acid resistance, an estimated 20 to 40 percent survive the stomach; ordinary strains that were never selected do far worse. Those that make it through almost never settle in the colon: the places on the wall and the food in the gut are already taken by your own resident bacteria, so the newcomers cannot get a hold and are carried out with the gut contents. A few weeks after you stop taking them, they usually can no longer be found in stool.
So a probiotic is more like a passing temp worker than a tenant moving in.
The odd thing is that a passer-by that cannot make a home can still spare children on antibiotics some diarrhea: in the pooled trials, diarrhea fell from 19% to 8%. What it does while passing through starts with what antibiotics do to the gut.
Mechanism · How a passer-by can still work
Start with what antibiotics do to the gut, and the line diarrhea risk falls to about half (antibiotic-associated diarrhea in children, from Guo 2019, a Cochrane review that pools many different probiotics) has somewhere to land.Antibiotics do not tell friend from foe. Aimed at one infection, they also wipe out a large share of the resident bacteria in your colon. Two things then happen at once:
Seats empty out: resident bacteria normally occupy the attachment sites on the wall and the sugars in the gut, and the opportunistic pathogens that normally cannot squeeze in (*C. difficile* is the best known) wait at the door. Once the residents are gone, the queue is let through.Nobody is cooking: with fewer fiber-fermenting bacteria, the colon wall loses its supply of the short-chain fatty acid its cells like best, butyrate. The wall itself goes hungry, water that should have been absorbed stays in the gut, and stool turns loose.
Now look back at the passer-by. It cannot settle, but as long as it is present, the prevailing explanation holds that three things happen:
It takes up seats: it also wants attachment sites and sugar. For as long as it holds them, the opportunistic pathogens cannot. This is a substitute standing in an empty seat, not a new resident moving in.It makes acid on the spot: it ferments sugar and releases acid right there, pushing the pH of that small stretch of gut down. Many opportunistic pathogens grow poorly in a more acidic setting, so they are held back.It lets the immune system see it: fragments of its cell wall are read by immune cells beneath the gut lining. When the immune system reads there are still normal bacteria out there, it keeps the production lines for mucus and antimicrobial peptides running, and that mucus is the first layer of the barrier.
These three come mostly from laboratory and animal studies, and how much each contributes in people is still unclear. But they share one feature: they hold only while it is present. That single fact explains nearly every practical rule later on: why you take it during the antibiotic course rather than waiting until the course ends; why you take it continuously; why the effect leaves as soon as you stop; and why tune it once, benefit for life is impossible. It is not someone moving in to remodel the house. It is a temp worker you hired, and when the pay stops, the worker leaves.
*S. boulardii* deserves a note of its own: it is a yeast, not a bacterium, and the antibiotic's sights cannot find it. Other strains have no chance until the antibiotic has cleared, but it can occupy those empty seats while the antibiotic is still in your body. That is one reason it has been studied more than most for antibiotic-associated diarrhea, and the reason it can be taken at the same time as the antibiotic without spacing the doses.
Read the three the other way round and you get a test: any probiotic pitch claiming a clear effect in healthy people with an intact microbiome starts with a low prior. The seats are already full, so the first two jobs have nowhere to happen, and only the immune route is left, whose effect is small to begin with.
Numbers · How many get through the stomach
Being able to work depends first on being able to arrive. The numbers at this gate are blunt, and they are uncertain.Swallowing 10 billion live bacteria and 10 billion live bacteria reaching the colon can be several orders of magnitude apart.
Stomach acid sits at a pH of 1.5-2.0, which is hostile to most lactic-acid bacteria and bifidobacteria. The orders of magnitude commonly given in reviews and textbooks are roughly:
Typical *Lactobacillus* and *Bifidobacterium* strains: survival after the stomach in the range of 0.01-1%*S. boulardii* (a yeast): far more acid-resistant, with about 10-30% surviving the stomachSpore-forming bacteria (*Bacillus coagulans*, *Bacillus subtilis*): nearly 100% get through the stomach, but the clinical evidence is limited
Ways to protect them:
Enteric coating: it does not dissolve in stomach acid and releases the bacteria only in the small intestine, which can raise the number actually delivered 10-100 timesTaking them with a meal: food buffers stomach acid, but it also dilutesDead-cell preparations (postbiotics, heat-killed bacteria): some studies suggest dead cells also have effects, because cell-wall components (peptidoglycan, for example) can themselves modulate immunity, though the idea is still evolving
⚠️ Read these percentages as orders of magnitude, not exact values. Survival can differ by several orders of magnitude depending on the strain, the formulation and how it is measured. A review of probiotic survival (Bezkorovainy 2001) estimates 20-40% for strains specifically screened for acid resistance; ordinary shelf strains that never went through that screening sit in the very low range above. On the other hand, a study in people that sampled the gut lining directly by endoscopy (Zmora 2018) found that the 11-strain combination it used was still alive after passing through the whole digestive tract. So do not memorize any single number. Remember the rule instead: acid resistance is a trait of the strain, not of probiotics as a class. And what stops most bacteria is often not the stomach but the next gate: they cannot settle.
bezkorovainy-2001-ajcn-survival
Mechanism · Why a probiotic does not settle
Colonization (settling in the gut long term and multiplying there) is a separate matter. Most probiotics cannot do it, and two Israeli studies that sampled the gut directly by endoscopy (published in 2018 in the same issue of *Cell*) showed this clearly:Zmora 2018: healthy volunteers took an 11-strain combination, and the researchers sampled the gut lining directly. The bacteria survived the trip through the digestive tract, but whether they held on to the lining varied by person, by gut region and by strain; some people let them in and some barely did at all. Whether the bacteria turned up in stool did not tell you whether they had stayed on the lining. The effect on the lining's community was temporary and differed from person to person.Suez 2018: people first took antibiotics, and three ways of recovering were compared. After antibiotics had disturbed the microbiome, probiotics actually held on to the human gut lining more easily, which fits the reasoning that only an empty seat gives leverage. The cost was that people taking the probiotic regrew their own original microbiome more slowly and less completely. People given a transplant of their own stool, banked before the antibiotics (autologous fecal transplant), recovered nearly completely within days.
None of this contradicts probiotics reduce antibiotic-associated diarrhea, because they measure different things: the Guo 2019 pooled analysis measured whether diarrhea happened, while Suez measured how fast your own microbiome grew back. For someone on antibiotics who is worried about diarrhea, the first is the more direct answer. But for the common claim that probiotics help your microbiome recover, the human data available so far point the other way (one small study, still to be replicated).
Three practical inferences follow:
1. Probiotics generally have to be taken continuously to keep an effect; it is not tune it once, benefit for life
2. Individual differences are large: some people's guts let a given strain stay for a while, and others reject it entirely
3. Fecal microbiota transplant (FMT) is not a bigger dose of probiotic. It is a different kind of intervention: it delivers a complete community and its metabolites, along with the environment the bacteria cling to, which is why it can colonize where a single strain cannot
So claims like yogurt cures everything or one probiotic capsule rewrites your microbiome miss where probiotics really stand: passers-by, with a temporary effect.
Why they cannot settle is quite ordinary: the colon is not an empty room, it is a full one. To stay, a strain has to find two things at once: an unoccupied attachment site on the wall, and a meal nobody else has eaten. The residents already hold both, and they release molecules that suppress their neighbors. A newcomer with no seat can only drift in the current inside the gut, and because gut contents keep moving down, whatever drifts is carried out.
So disappearing soon after you stop is not being killed; it is being washed out. The difference matters: the first suggests you could push through with a bigger dose, while the second tells you a bigger dose only delays the wash-out a little.
Two more things follow from this chain:
Why a prebiotic (put plainly, fiber) is often the better buy: fiber feeds the residents who already hold their seats and already live there, so it never has to pass the colonization gate, and it is not bothered by stomach acid. You are not squeezing one more person into a full room; you are bringing food to the people already inside.Why transplanting a whole community is a different order of thing: it does not squeeze one more person into a full room, it replaces the whole household and moves in their division of labor with them. The chapter What a fecal transplant does covers it.
Evidence · A trial of fermented food versus fiber
A randomized trial from the Sonnenburg lab at Stanford (Wastyk and colleagues, published in *Cell* in 2021) is the most cited in this area. It enrolled 36 healthy adults, 18 per group, for a 10-week dietary intervention:One group stepped up to a high-fiber dietThe other stepped up to a lot of fermented food (yogurt, kefir, kimchi, sourdough, kombucha, pickled vegetables and so on), reaching an average of 6.3 servings a day by the end of the maintenance phase
Results:
The trial's prespecified primary outcome (a cytokine response score) did not change in either groupAmong the secondary results, microbiome diversity rose steadily in the fermented-food group, and of the 93 cytokines, chemokines and other inflammation-related serum proteins measured, 19 went down (named ones include , IL-10 and IL-12b). That is 19 out of 93, not a drop across the boardIn the high-fiber group microbiome diversity did not change, but the microbiome's enzymes for breaking down polysaccharides increased, and the immune measures split into three different paths according to how diverse each person's microbiome was at the start. The researchers suggest that some people's microbiomes could not yet ferment that much fiber and needed longer to adapt
In principle fiber should have helped more (it feeds bacteria directly and they turn it into short-chain fatty acids), yet on this trial's measures of diversity and inflammation, the fermented-food group changed more.
Remember its limits when you read it: the sample was small, the participants were healthy, and the primary outcome was negative. Whether these shifts in markers turn into less illness is something it did not measure.
One explanation (not yet tested directly): fermented foods deliver several live cultures at once, the metabolites they already made in the jar (short-chain fatty acids, amino acids, vitamin , polyphenols reshaped by bacteria), and fragments of cell wall (the postbiotic part). That is what whole-food matrix means: not one strain, but bacteria, metabolites and the food itself acting together. A probiotic capsule delivers one high-purity strain, and the matrix is gone.
Practical suggestions:
Make rotating fermented foods your first move: plain unsweetened yogurt, kefir, kimchi, sauerkraut, natto, miso, kombucha, vinegared vegetables1-2 servings a day, varying the kinds, is closer to what the trial did than the same cup of yogurt every day. But know that this is far less than the trial's amount, and how much 1-2 servings a day changes has not been measuredPair it with fiber to feed the bacteria you already have: oats, beans, whole grains, vegetables and fruitUse supplements only for specific uses with evidence, such as preventing antibiotic-associated diarrhea or preventing *C. difficile* infection while on antibiotics (the chapter Which conditions have evidence gives the full grading)For vague gut health, there is no need to take probiotic capsules long term
What actually reshapes the microbiome is giving the bacteria a good working environment day after day (fiber, a varied diet, avoiding antibiotics you do not need), not tossing in a few hundred million live bacteria in one go.
Chapter 3
Which conditions have evidence
Do probiotics work? That is the wrong question. The only form of it that can be answered is: which strain, used for what.
The same bottle can have very different evidence behind it for two different uses: preventing antibiotic-associated diarrhea in children is backed by many randomized trials, while weight loss is essentially marketing. So the evidence has to be graded use by use rather than rolled into one overall score.
Once it is graded, a handy pattern shows up: the better-supported uses almost all fall where the gut microbiome has just been disrupted, or has not been built yet, such as while taking antibiotics, or in a preterm infant whose microbiome is only starting out. A passer-by has leverage only when there are empty seats.
Adding one more strain to a healthy adult with an intact microbiome is like squeezing one more person into a full train carriage. Any pitch that promises a big effect in people like that deserves a discount before you look any further.
The same bottle can have very different evidence behind it for two different uses: preventing antibiotic-associated diarrhea in children is backed by many randomized trials, while weight loss is essentially marketing. So the evidence has to be graded use by use rather than rolled into one overall score.
Once it is graded, a handy pattern shows up: the better-supported uses almost all fall where the gut microbiome has just been disrupted, or has not been built yet, such as while taking antibiotics, or in a preterm infant whose microbiome is only starting out. A passer-by has leverage only when there are empty seats.
Adding one more strain to a healthy adult with an intact microbiome is like squeezing one more person into a full train carriage. Any pitch that promises a big effect in people like that deserves a discount before you look any further.
Evidence · How strong the evidence is, use by use
Here they are, ranked by how much evidence exists. The order itself is information: the further down, the more you should doubt the sentence printed on the front of the bottle.Tier one: pooled randomized trials, or a conditional recommendation from a guideline
Preventing antibiotic-associated diarrhea (AAD, diarrhea that starts during or after a course of antibiotics) in children: Guo 2019, a Cochrane review, pooled more than thirty trials; 8% of the probiotic groups developed diarrhea versus 19% of controls (, RR, 0.45; roughly one case avoided for every 9 children treated; moderate certainty). The pooled products were many different probiotics, and the authors single out high doses of *L. rhamnosus* or *S. boulardii* as most worth confirming in a large trial. These are children's data; do not quietly carry them over to adultsPreventing ***C. difficile* infection in adults and children while they take antibiotics: the American Gastroenterological Association (AGA) 2020 guideline gives a conditional recommendation for a few specific strains or combinations (low quality of evidence). That is a different question from an established *C. difficile* infection**: for that, the AGA advises probiotics only within a clinical trial, and for infections that keep coming back the strongest evidence is for fecal microbiota transplant (FMT, with cure rates of 80-90%)Preventing necrotizing enterocolitis (NEC, an emergency in which a preterm baby's gut wall becomes inflamed and dies) in preterm infants: the AGA 2020 guideline gives a conditional recommendation for specific *Lactobacillus* plus *Bifidobacterium* combinations. ⚠️ This benefit is tied to a real harm: in September 2023 the US FDA warned that preterm infants given probiotics are at risk of invasive, potentially fatal infection. A preterm infant with a birth weight under 1000 g died of *Bifidobacterium longum* sepsis, and genetic sequencing showed that the organism matched exactly the bacteria in the product given. Both halves are true, so this is a decision for the neonatal team, not a choice made in front of a shelfPouchitis (inflammation of the pouch built from the small intestine after the colon is removed): the AGA 2020 guideline gives a conditional recommendation for an 8-strain combination
Tier two: randomized trials exist, but results are inconsistent or effects small
Irritable bowel syndrome (): specific strains (*B. infantis* 35624 and some multi-strain combinations) have trials showing small improvements. Ford 2018, a , pooled more than fifty trials and concluded that certain combinations or strains appeared to help overall symptoms and abdominal pain, but that which ones work remains, for the most part, unclear. Neither of the two US societies recommends them: the AGA 2020 guideline advises use only within a clinical trial, and the American College of Gastroenterology (ACG) 2021 guideline suggests against probiotics for overall IBS symptoms (a conditional recommendation, very low quality of evidence). That societies and trials do not line up is itself the honest state of the evidenceInfant colic: *L. reuteri* DSM 17938, with some positive randomized trialsPrimary prevention of allergy and eczema (preventing it before it starts): in a randomized trial in Finland (Kalliomäki 2001), pregnant women with a family history of allergy took LGG late in pregnancy and their babies took it for six months after birth, and by age two the rate of eczema was roughly halved (23% vs 46%). It is a single trial and still needs replicationBacterial vaginosis, as an add-on: combined with antibiotics, some trials saw fewer recurrences
Studied, and the results do not support it
Acute infectious diarrhea in children: this one has dropped out of tier one. The 2020 Cochrane review (Collinson) pooled 6 trials and 3,058 participants and found illness only 8.6 hours shorter on average, with a 95% (the range the true effect most likely falls in) running from 29 hours shorter to 12 hours longer, at very low certainty. The AGA 2020 guideline suggests against probiotics for children with acute infectious gastroenteritis (a conditional recommendation). The treatment for acute diarrhea in a child is rehydration
Tier three: very weak evidence, little practical meaning
Boosting immunity, preventing the common cold: some randomized trials are weakly positive, with small effectsAllergic rhinitis: weak evidenceTreating atopic dermatitis once it is diagnosed: works less well than primary preventionDepression and anxiety (so-called psychobiotics): the early data are interesting, but randomized trials are few and the effects smallAthletic performance: only 1-2 small randomized trials
No real evidence, yet widely marketed: weight loss (no strain has been shown in trials to reliably lower body weight); anti-aging and longevity (no randomized trials in people); the autism spectrum (animal studies exist, human evidence is scant and inconsistent); skin whitening and glow (marketing language); preventing diabetes or controlling blood glucose (a few small studies, a very weak signal).
To decide whether a bottle of probiotics is worth buying, ask yourself three things:
1. Is the problem I want solved in tier one or tier two?
2. Does it contain the strain that was tested in randomized trials, or a vague broad-spectrum formula?
3. Could I use fermented food plus fiber instead?
If none of the three apply, the money is better spent at the produce market.
fda-2023-probiotics-preterm-warningkalliomaki-2001-lancetlacy-2021-acg-ibs
Mechanism · Why strong evidence clusters in few places
Line up the tier-one uses and they share one shape: the gut microbiome is empty right now, or has not been built yet.Someone taking antibiotics: a large share of the residents has been knocked out, the wall and the food supply are open, and opportunistic pathogens queue at the door.A preterm infant: the microbiome is still being built from nothing, and whoever arrives first takes the seat.Pouchitis: the pouch built after the colon is removed is new ground where a microbiome has to grow back.
Only in settings like these does what a passer-by brings (taking up seats, making acid, letting the immune system see it) have leverage, because there are seats to take. A study in people saw this too (Suez 2018): after antibiotics had disturbed the microbiome, probiotics held on to the gut lining more easily than usual. Turn it around: in a healthy adult with an intact microbiome, the seats are already full, and one more strain cannot even find a place to stand before it is washed out.
There is a counter-example worth taking seriously: acute diarrhea in children. The gut is flushed through several times a day there too, so by this reasoning there should be leverage, yet the pooled trials did not find a clear benefit. So an empty seat looks more like a necessary condition than a sufficient one: the strain, the dose and the timing also have to be right. That step is reasoning, not a conclusion a trial has tested directly.
Even so, this reasoning is the most useful tool in the whole story, because it lets you judge a claim you have never seen without memorizing the full table:
A pitch claiming a clear effect in ordinary healthy people: a low prior, because it asks a passer-by to do work where there are no empty seats.A pitch claiming to shift a long-term measure (weight, lifespan, the state of your skin): a lower prior still, because a passer-by's effect disappears once you stop, and long-term measures need something that stays around to push them.A pitch that falls in a window when the microbiome has just been disrupted: worth looking at closely, including exactly which strain was used and in which people it was tested.
The other side of the same fact also explains why probiotic effects differ so much from person to person. Whether your gut is empty or full right now, and which mix of residents you carry, decides whether the same capsule has anywhere to land in you. That is not a vague constitution; it is a very concrete question: is there an empty seat.
Safety · When you should not take them
A few important warnings:People who are critically ill, immunosuppressed or have a central venous line: probiotics can cause bacteremia (bacteria getting into the blood; rare but real), so they should be used with caution in the ICUAcute pancreatitis predicted to be severe: in the Dutch PROPATRIA trial (published in *The Lancet* in 2008), patients like this were given a multi-strain probiotic, and deaths were 16% in the probiotic group versus 6% on placebo, more than twice the placebo rate; some in the probiotic group also developed bowel ischemia. Serious illness is not a just add more bacteria situationPreterm infants in hospital: in September 2023 the US FDA warned that preterm infants given probiotics are at risk of invasive, potentially fatal infection. A preterm infant with a birth weight under 1000 g died of *Bifidobacterium longum* sepsis after being given a probiotic in hospital, and genetic sequencing showed the organism was the same strain as the one in the product. The FDA also noted that in the US no probiotic is approved for use in infantsPregnancy and infancy: most of the evidence comes from strains with long use and fuller safety records, such as LGG and BB-12; do not casually switch to others
⚠️ Read the preterm line together with the one in tier one of the evidence, not as half a memory. In the same group of preterm infants, probiotics really do lower the risk of necrotizing enterocolitis (a conditional recommendation from the AGA 2020 guideline), and a child really did die of the strain in the product. These are not two contradictory statements; they are one event in which benefit and harm sit side by side. That is exactly why a neonatal team has to weigh it one infant at a time, rather than a parent deciding in front of a shelf.
These warnings are not just in case boilerplate. They rest on a very concrete chain: probiotics are alive, and the gut wall is only one cell thick.
In a healthy person, that layer of cells plus the mucus covering it keeps bacteria firmly on the inside of the gut. When someone is critically ill, immunosuppressed, has a damaged gut wall, or has a line running straight into a blood vessel, that gate is no longer tight, and live bacteria that should stay inside the gut can cross into the bloodstream. That is bacteremia.
So the instinct to add more good bacteria turns completely around in serious illness: what you are adding is not a nutrient but a living organism. A nutrient in the wrong place is at worst wasted; a live organism in the wrong place is an infection. That is why this call always belongs to a doctor, never to you in front of a shelf.
besselink-2008-propatria-lancetfda-2023-probiotics-preterm-warning
In practice · How to read a probiotic label
Some traps that often appear on labels.10 billion live bacteria: is that the count at manufacture or at the end of shelf life? Cheap products often print the manufacture figure, and after six months only 1-10% may be left. A good label states no less than so many CFU at the end of shelf life (English labels say "At expiration"), or gives refrigeration instructions.
16-strain blend: it sounds more complete, but strains can compete with and suppress one another, and each strain's dose is diluted. Clinical randomized trials are almost all one strain for one use, not the full set. VSL#3 (Visbiome) is one of the few multi-strain combinations tested in randomized trials.
Upgraded with added prebiotics: prebiotics (fiber, fructo-oligosaccharides, inulin) are useful on their own, but whether they really work together with the strain you are buying is something ordinary products have no data on. Simply eating more fiber is much cheaper than buying a prebiotic supplement.
Clinically proven, doctor recommended: with no assessment by a regulator or a major medical society behind it, this is marketing language. What real medical consensus there is sits in the American Gastroenterological Association (AGA) 2020 guideline, and its three kinds of verdict have to be kept apart rather than blurred into one sentence:
Conditionally recommended: pouchitis; **preventing *C. difficile* infection in adults and children taking antibiotics**; preventing necrotizing enterocolitis in preterm, low-birth-weight infantsSuggested against: children with acute infectious gastroenteritis (a conditional recommendation, moderate quality of evidence)No recommendation; use only within a clinical trial: Crohn's disease, ulcerative colitis, irritable bowel syndrome, and ***C. difficile* infection that has already developed**
⚠️ The one most often reversed is *C. difficile*: preventing it while taking antibiotics is a conditional yes; treating it once you have it is where the guideline makes no recommendation. As for generic gut health, the guideline has no such row at all. Never assessed is not the same as recommended.
Concept upgrades such as live bacteria, bacteriocins and postbiotics: dead-cell preparations or bacterial products are interesting in theory, but randomized trials are very scarce, so do not pay a premium for a concept upgrade.
Refrigerated or shelf-stable: refrigerated products usually keep more activity but are harder to transport and carry; shelf-stable products rely on spore-formers or freeze-drying, which is convenient, but some strains may lose potency compared with the refrigerated version. Picking a form you will keep taking matters more than picking one that is theoretically better but that you find a hassle.
Turning this into action:
If you have a specific use from tier one or tier two: buy the specific strain tested in randomized trials for that use (LGG, *S. boulardii*, *B. infantis* 35624 and so on)If you want to tune your gut: fermented food, plenty of fiber and a varied diet are worth doing before any probiotic supplementIf you are seriously ill or immunosuppressed: talk to your doctor firstIf you are healthy and want a daily capsule as insurance: no need; spend the money on vegetables and fruit
Chapter 4
What fermented foods add
Fermented food and a probiotic capsule do not travel the same road.
A capsule gives you only one thing, live bacteria, and even that one thing first has to run the gauntlet of stomach acid. Fermented food gives you three: live bacteria, the metabolites the bacteria already made in the jar, and the food itself, reworked once by the bacteria: some of the lactose has been eaten, and some of the phytate that blocks iron and zinc absorption has been broken down.
What matters is that the last two are not bothered by stomach acid or by heat. They do not need to reach the colon alive; once you have eaten them, you have them.
So by this reasoning, eating a rotation of fermented foods each day gets you a whole set rather than one strain. That follows from what they contain; it is not a result from a trial that compared the two head to head.
A capsule gives you only one thing, live bacteria, and even that one thing first has to run the gauntlet of stomach acid. Fermented food gives you three: live bacteria, the metabolites the bacteria already made in the jar, and the food itself, reworked once by the bacteria: some of the lactose has been eaten, and some of the phytate that blocks iron and zinc absorption has been broken down.
What matters is that the last two are not bothered by stomach acid or by heat. They do not need to reach the colon alive; once you have eaten them, you have them.
So by this reasoning, eating a rotation of fermented foods each day gets you a whole set rather than one strain. That follows from what they contain; it is not a result from a trial that compared the two head to head.
Mechanism · What fermentation already did in the jar
Fermentation's original purpose was not health; it was preservation. And the principle of preservation is itself a clean mechanism chain.Step one: the bacteria eat the sugar and release acid. Lactic-acid bacteria ferment the sugar in the food into lactic acid, and the pH in the jar keeps falling. Most of the microbes that spoil food cannot grow at that acidity, so the food keeps. The sourness you taste is the taste of the preservation mechanism itself: acid is not a by-product of fermentation; acid is the work fermentation is doing.
Step two: acid and enzymes cut big molecules into smaller ones. Along the way the bacteria do a round of pre-digestion for you. Some of the lactose is broken down (which is why many people with lactose intolerance can eat yogurt but cannot drink milk). Some of the phytate in grains and beans, which holds iron and zinc tightly, is broken down, so those minerals are easier for the small intestine to take up. And some protein is cut into small peptides and free amino acids, which are easier to absorb and bring savory flavor along.
Step three: the bacteria leave their products in the food. Short-chain fatty acids, some vitamins, polyphenols the bacteria have reshaped. These are not live cells; they are what live cells left behind, known collectively as postbiotics. They are not bothered by stomach acid, do not need to colonize, and certainly do not need a home in your gut.
After the three steps, a jar of fermented food holds three things at once:
Live bacteria: fragile; a good share fail the two gates of stomach acid and bileMetabolites: not fragile; all of them arriveThe reworked food matrix: not fragile; all of it arrives
A capsule holds only the first, and it happens to be the most easily lost of the three. That is the concrete meaning of the phrase whole-food matrix, and it is one explanation researchers offer for the result of the fermented-food trial (covered in the chapter How many swallowed bacteria survive): the group that ate a rotation of fermented foods changed more on microbiome diversity and inflammation markers than the group that only added fiber.
Two common questions fall out of this:
Why does boiling water, a hot soup or canning for the shelf take something away? Live bacteria are organisms held together by a whole set of enzymes, and once high heat unfolds those proteins they do not fold back. But only the first of the three things is lost; the other two remain. So heated fermented food has not turned into rubbish; it has gone from three things to two.Why are miso and soy sauce, which are salted and heated, low in live bacteria yet still worth eating? Same reason: the two things you wanted were there all along.
Background · Which foods are really fermented
People have used fermentation to preserve food for about 8000 years. Fermented food is also the most common source of live bacteria in an everyday diet, and it is cheap. One distinction matters, though: by definition, the microbes in fermented food are not called probiotics unless a particular strain has been singled out and tested in people.Main types (by region and by microbe):
Yogurt: standard fermentation with *Lactobacillus bulgaricus* and *Streptococcus thermophilus*. A real yogurt's ingredient list has only milk and live cultures: no sugar, no flavoring, no thickenersKefir: made on a milk or water base and fermented by a mix of 10-30 kinds of bacteria and yeasts, far more varied than yogurtPickled vegetables and kimchi: vegetables fermented by lactic-acid bacteria, with polyphenols from chili and other ingredientsSauerkraut: cabbage and salt, fermented spontaneouslyNatto: fermented with *Bacillus subtilis* var. *natto*, and the common food richest in vitamin (the form)Miso and soy sauce: a koji made with *Aspergillus oryzae*, with bacteria joining later; salting and partial pasteurization leave few live cellsKombucha: tea fermented with a symbiotic film of bacteria and yeast (a SCOBY), containing acetic-acid bacteria and yeastsSourdough: wild yeast plus lactic-acid bacteria; most of the microbes die in baking, but phytate goes down and the blood-glucose response may be gentler
Real fermentation and fake fermentation:
Real: fermented by cultures that arrived naturally or were added, kept refrigerated, with an ingredient list that says live culturesFake (a short commercial ferment followed by pasteurization, or simply pickling in vinegar): keeps at room temperature, lists vinegar rather than cultures, and has almost no live cellsMost shelf-stable yogurt is heat-treated after fermentation, and its bacteria have been killed
Most pickled cucumbers, vinegar pickles and heat-treated sauerkraut on supermarket shelves are not really fermented. In US supermarkets, refrigerated brands such as Bubbies and Wildbrine are, and so is kimchi you make yourself. In China, check whether the ingredient list names cultures or vinegar, and whether the product sits in the refrigerated case or on a room-temperature shelf.
In practice · How to eat fermented food each day
The health evidence for fermented food:The Sonnenburg lab's randomized trial, published in *Cell* in 2021: in 36 healthy adults, after 10 weeks of dietary intervention, 19 of the 93 inflammation-related proteins measured went down in the fermented-food group and microbiome diversity rose; but the trial's primary outcome did not change, and the sample was smallKorean cohort studies (Park 2017 and others): people who ate more kimchi had a lower body mass index () and less metabolic syndrome. This is an observed association, with , and cannot show that kimchi caused itYogurt and the heart: in of observational studies, people who ate yogurt regularly had a lower risk of diabetes and cardiovascular disease. Again, an association, not cause and effect
In practice:
1-2 servings of different fermented foods a day, rotating the kinds200 g of yogurt or kefir at breakfast, plus a little kimchi or sauerkraut (50-100 g) with a main meal, is a good combinationIf you are lactose intolerant, try plant-based yogurt, coconut kefir, vinegared vegetables or kombuchaWatch the salt: kimchi and sauerkraut are high in salt, so people with high blood pressure should count them in their daily saltFor children, older adults and people with weakened immunity, choose commercially made products that are pasteurized first and then given cultures, and avoid home spontaneous ferments (there is a contamination risk)
In practice · How to read a yogurt label
Real yogurt versus a yogurt drink: one minute with the ingredient list tells them apart.Real yogurt's ingredients: **raw milk, *Streptococcus thermophilus*, *Lactobacillus bulgaricus***, and that is it: no more than 5 items.
Common ingredients in yogurt drinks and flavored yogurt (the ones to avoid):
Raw milk, reconstituted milk, waterLarge amounts of white sugar, high-fructose corn syrup, sucroseFlavoringsThickeners (pectin, carrageenan, xanthan gum)Food coloringPreservatives (potassium sorbate)
Also, most yogurt sold on a room-temperature shelf has been heat-treated after fermentation and no longer contains live cultures.
A few common misconceptions:
Sugar-free does not mean healthy; check the grams of carbohydrate. The carbohydrate in real yogurt is mainly lactose, about 4 g/100 g; a sugar-free flavored yogurt may still contain sweetenersGreek yogurt is strained to remove whey and so concentrated, high in protein (about 10 g/100g, against 3-4 g in regular yogurt) and low in carbohydrate, which makes it a good choiceWhole-fat or fat-free: in of observational studies, people who ate whole-fat yogurt did not have a higher risk of cardiovascular disease, and if anything a lower one. That is an association and cannot show the fat caused it, but it does mean you need not be led by a low-fat label
For children: for under-3s, the usual advice is plain whole-fat yogurt with no added sugar, which provides protein and calcium, and some products are also fortified with vitamin D. With sweetened children's yogurts, check the sugar on the nutrition panel; many contain plenty, and made for kids is mostly a marketing idea.
Lactose intolerance: fermentation partly breaks down the lactose in yogurt and kefir, and most people with lactose intolerance can handle 100-200 g. People who cannot tolerate lactose at all, or who are allergic to milk protein, can choose plant-based ferments (soy, coconut, almond); plant-based yogurts carry less protein and calcium than dairy ones, so prefer fortified versions.
Storage details: keep it strictly below 6°C in the fridge; finish it within 3-5 days of opening; a layer of yellowish liquid on top (whey) is normal and contains protein, so there is no need to pour it off; if it forms clumps, smells off or the lid bulges, throw it out.
The yogurt-cures-everything marketing versus reality:
Tuning the gut: real yogurt plus a varied diet has some evidence; do not expect miraclesBoosting immunity: a vague claimSkin whitening and weight loss: no evidence
Yogurt fits best as a natural breakfast or snack that is high in protein, moderate in calories, contains calcium (sometimes with added vitamin D) and has a low glycemic index (). Treat it as food, not medicine.
Chapter 5
What a fecal transplant does
Adding one strain is squeezing one more person into a full room; a fecal microbiota transplant (FMT) replaces the whole household.
What comes over from the donor is not a single strain but a community that has already balanced itself: who eats what, and whose waste is whose meal, is settled. It arrives at a density high enough to fill the attachment sites, and it brings the mucus and metabolites the bacteria depend on to hold on. So it does not have to fight for a crack on someone else's ground; it is the new ground.
That is why it can do what adding a strain cannot: it does not add a temp worker, it swaps in a whole ecosystem.
And precisely because it swaps the whole set, it is a medical procedure, not wellness. Donors have to be screened, the process has to be regulated, and the long-term effects are still being watched.
What comes over from the donor is not a single strain but a community that has already balanced itself: who eats what, and whose waste is whose meal, is settled. It arrives at a density high enough to fill the attachment sites, and it brings the mucus and metabolites the bacteria depend on to hold on. So it does not have to fight for a crack on someone else's ground; it is the new ground.
That is why it can do what adding a strain cannot: it does not add a temp worker, it swaps in a whole ecosystem.
And precisely because it swaps the whole set, it is a medical procedure, not wellness. Donors have to be screened, the process has to be regulated, and the long-term effects are still being watched.
Evidence · Where fecal transplants are used
Fecal microbiota transplantation (FMT) sounds unpleasant, but it is one of the rare cases in modern medicine of transplanting a whole ecosystem, and for one indication it reaches cure rates of 80-90%.The use with the strongest evidence, and the only one with approved products: recurrent *C. difficile* infection, meaning *C. difficile* infection that keeps coming back after antibiotic treatment. The classic randomized trial is van Nood 2013 (published in the *NEJM*): after a first infusion of donor stool, 81% of patients with recurrent infection were cured, against only 31% in the control group given vancomycin alone; the trial was stopped early after an interim analysis. In 2022 the US FDA approved the first stool-derived live biotherapeutic, RBX2660 (Rebyota), and note that it is a suspension given through the rectum, not a capsule; SER-109 (Vowst), approved in 2023, is the first oral capsule. Both are used to prevent *C. difficile* infection from coming back again.
Uses still under study (low to very low evidence):
Ulcerative colitis, a form of inflammatory bowel disease: some randomized trials saw higher remission rates, but it usually takes several, closely spaced sessionsDiarrhea-predominant irritable bowel syndrome: small trials with inconsistent resultsMetabolic syndrome and insulin resistance: in a small trial (Vrieze 2012, published in *Gastroenterology*), stool from lean donors was transferred to people with metabolic syndrome, and insulin sensitivity improved at 6 weeks, but the effect did not lastAutism spectrum disorder: Kang 2017 (published in *Microbiome*) reported symptom improvement in 18 children, but it was an open-label study with no control group, exactly the kind of design that cannot separate a real effect from expectation plus natural ups and downs. If even the formal research version is unsettled, a do-it-yourself version at home is out of the questionDepression and anxiety: early stageMultiple sclerosis: early research
kang-2017-microbiome-mtt-autism
Mechanism · Why a whole community can settle
Why FMT can do what a single-strain probiotic cannot: it is not a bigger dose of probiotic; it is a different kind of intervention:1. It delivers a complete community: over a thousand species, including ones that cannot be grown in the lab; a probiotic capsule holds 1-16 strains
2. It delivers the bacteria's metabolic output as well: short-chain fatty acids, vitamins and bile-acid metabolism, all already running
3. It delivers the bacteria's home: mucus, peptidoglycan, fungal components, phages
4. It delivers a combination already selected: one that has worked inside another healthy person
Those four are really one point: it brings over a system that is already running, not a spare part.
Think of a shared flat. Adding one strain is a stranger standing in the hallway with a suitcase: no empty bed, no meal set aside for them, and residents who close ranks against them, so they leave after a few days. That is the passer-by described in the chapter How many swallowed bacteria survive. FMT does not squeeze that person in; it moves the whole household in along with its order: who sleeps in which bed, who shops and who cooks, who takes out the rubbish. The division of labor arrives with the people. That is why it can gain a foothold. In the van Nood trial, after the transplant, the diversity of bacteria in patients' stool rose to a level similar to that of the healthy donors.
Follow this reasoning and two judgments come straight out:
Why its strongest indication is exactly that one. After repeated rounds of powerful antibiotics, few residents are left and *C. difficile* has taken the ground. The house is empty (more precisely, taken over by one bad tenant), and only a whole household of healthy residents moving in can push it out.Why it has not become a wellness product. Doing the same thing in someone with an intact microbiome has no such leverage: the house is full, replacing the residents is unnecessary, and the risk is out of proportion.
So the difference between FMT and probiotics is mainly not about stronger or weaker but about whether it can stay. Once that is clear, you also know why there is no home FMT kit on the shelf, and why there should not be.
Safety · Why it must be done under regulation
How FMT is done: by colonoscopy or enema (the traditional routes, delivering straight to the colon); through a nasojejunal tube (entering from the upper digestive tract); or as frozen capsules (known informally as poop pills, swallowed, and now used in more and more hospitals). The source can be a commercial stool bank (OpenBiome in the US, with similar organizations in other countries).Risks: pathogens can be passed along with the transplant, so donors have to be strictly screened (HIV, hepatitis B, hepatitis C, parasites, multidrug-resistant bacteria and more). In 2019 in the US, **2 immunosuppressed patients who received stool from the same donor developed bloodstream infections with a drug-resistant *E. coli* that produces extended-spectrum beta-lactamase (ESBL), and 1 of them died; the US FDA then tightened its screening standards. ⚠️ This is often retold as 2 deaths**, but both the original notice and the report in the *NEJM* describe 2 infections and 1 death. The long-term effects are still unknown: could a transplant carry over the donor's metabolic tendencies, behavior or immune traits? That is still being studied.
The current consensus is that FMT is not general wellness but a medical treatment for specific indications, and it has to be done under regulation. Do not trust any do-it-yourself FMT or home stool transplant clinic.
Background · Where microbiome medicine is heading
How the microbiome field changed from 2010 to 2025, and the limits of the probiotic era.First-generation probiotics (1900-2000s): the idea was to add a single good bacterium to repair the gut; in practice most strains did not colonize, individual differences were large, and clinical effects were small. Yogurt cultures and ordinary commercial probiotics were typical.
Second-generation probiotics (2010-2020s): the idea shifted to strain-specific products validated by randomized trials for specific uses. In practice, LGG, *S. boulardii*, *B. infantis* 35624 and a few others became options with evidence behind them; what actually took hold were the settings where guidelines give a conditional recommendation (pouchitis, preventing *C. difficile* infection while on antibiotics, preventing necrotizing enterocolitis in preterm low-birth-weight infants), not the generic blends on the shelf.
Third generation: microbiome drugs (2022 onward):
Standardized stool-derived products reached the market: Rebyota (2022, a suspension given through the rectum) and Vowst (2023, oral capsules)Next-generation probiotics: beyond lactic-acid bacteria and bifidobacteria, including *Akkermansia muciniphila* (studied by Cani and others) and *Faecalibacterium prausnitzii* (which has shown anti-inflammatory effects in animal and cell studies)Synthetic communities: dozens of strains designed togetherBacteria plus metabolite combinations (the postbiotic idea): a probiotic plus its key products
Personalized microbiome medicine (under study): testing your own microbiome (among commercial services, uBiome has gone out of business while Viome and others continue), but its clinical use is very limited: knowing which bacteria you have does not tell you what to take. On predicting disease from the microbiome, part of the response to anti-PD-1 immunotherapy can be predicted from gut bacteria (Routy 2018, published in *Science*, an association finding); on drug response, gut bacteria affect how digoxin and chemotherapy drugs are metabolized (Haiser 2013, published in *Science*).
The most likely directions over the next 5-10 years:
More FDA-approved microbiome drugs (live biotherapeutic products, LBPs)Next-generation probiotics going mainstream (*Akkermansia* is already sold commercially)Combination therapy of microbiome products with drugs (alongside immunotherapy or chemotherapy)Precision probiotics: chosen and tracked according to a person's microbiome and specific use
At the level of daily life, though, big changes are unlikely within 5 years. The best choice for healthy people remains a varied diet, fermented food and plenty of fiber. That is not outdated advice: the link between a high-fiber, varied diet and better health has shown up again and again across decades of research, and it is cheap and doable.
One last point about consumer psychology: by industry estimates the probiotic market is worth more than $70B a year, and that much commercial interest inflates the idea of taking one every day as insurance without limit. The gap between the facts and the marketing is wide, and time spent getting the information straight usually pays off far better than buying more supplements.
Chapter 6
Should I take one, and how
Start by asking yourself: does the thing I want solved have a name?
If it does, check whether that name is on the short list that has evidence. Taking antibiotics and wanting less diarrhea is the use with the most evidence: pick the strain that was used in the trials and take it throughout the antibiotic course. A child's acute diarrhea, flare-ups of irritable bowel syndrome and travel also have names, but for them the evidence is against it or weak; the treatment for acute diarrhea in a child is rehydration, not probiotics.
If it has no name and you just want to tune your gut, capsules will not do much. What you need to change is the resident bacteria's working environment, not how many temp workers you toss in. That environment comes down to three things: give them food (fiber and a variety of plants), throw in fewer bombs (antibiotics you do not need, ultra-processed food), and bring in fresh blood often (fermented food).
All three act on the microbiome you already have. None of them has to get past stomach acid, and nothing has to colonize.
If it does, check whether that name is on the short list that has evidence. Taking antibiotics and wanting less diarrhea is the use with the most evidence: pick the strain that was used in the trials and take it throughout the antibiotic course. A child's acute diarrhea, flare-ups of irritable bowel syndrome and travel also have names, but for them the evidence is against it or weak; the treatment for acute diarrhea in a child is rehydration, not probiotics.
If it has no name and you just want to tune your gut, capsules will not do much. What you need to change is the resident bacteria's working environment, not how many temp workers you toss in. That environment comes down to three things: give them food (fiber and a variety of plants), throw in fewer bombs (antibiotics you do not need, ultra-processed food), and bring in fresh blood often (fermented food).
All three act on the microbiome you already have. None of them has to get past stomach acid, and nothing has to colonize.
Clinical · The situations you can name
Question 1: Do I have a specific clinical use?Taking antibiotics or just finished: *S. boulardii* 250 mg twice a day, or LGG at 10¹⁰ CFU a day. Bacteria such as LGG should be taken 2 hours apart from the antibiotic; *S. boulardii* is a yeast that antibiotics cannot kill, so it can be taken at the same time. The evidence for this comes mainly from trials in children (the Guo 2019 Cochrane review)Traveling somewhere with poor sanitation: the evidence for probiotics preventing travelers' diarrhea is not strong enough to put them on this list. What actually helps is care with drinking water and food hygiene, and carrying oral rehydration saltsIrritable bowel syndrome that keeps flaring: if you want to try, you can try *B. infantis* 35624 (Align) for 4 weeks and stop if it does not work. But know that this is trying it yourself, not following a guideline: the American Gastroenterological Association (AGA) 2020 guideline advises probiotics for only within a clinical trial, and the American College of Gastroenterology (ACG) 2021 guideline suggests against themA child's acute infectious diarrhea: rehydration is the treatment; probiotics are not. Pooled in the 2020 Cochrane review, the difference in duration crosses zero (8.6 hours shorter on average, with a 95% from 29 hours shorter to 12 hours longer, at very low certainty), and the AGA 2020 guideline suggests against it (a conditional recommendation). Put the money and the attention into oral rehydration saltsWomen with recurrent vaginal or urinary tract infections: the GR-1 plus RC-14 combination; the evidence comes from a few trials, so discuss it with a doctorA family history of allergy, and wanting to prevent eczema during pregnancy and infancy: LGG has been used late in pregnancy and in infancy, but the support rests mainly on one Finnish trial, so talk it over with a pediatrician first
Question 2: I just want to tune my gut.
You do not need a probiotic capsule; these three are more worthwhile:
1. 1-2 servings of different fermented foods a day (yogurt, kefir, kimchi, natto)
2. 25-40 g of fiber a day; eating 30 or more different plants a week is an often-quoted target (it comes from observational research)
3. Fewer antibiotics you do not need, and less ultra-processed food
Money spent at the produce market goes further than money spent at the pharmacy.
Question 3: I have heard probiotics boost immunity and slow aging.
The evidence for these uses is weak, and if you keep taking them it is marketing that has persuaded you, not science. Your effort goes further on sleep, exercise, trimming your waist and managing stress, where the evidence really is strong.
lacy-2021-acg-ibs
In practice · How to take them, and why
How to take them, point by point:With or after a meal: food buffers stomach acid and brings the bacteria a little nourishmentTake bacterial probiotics 2-4 hours apart from antibiotics, or the antibiotic will kill them too; *S. boulardii* is the exceptionKeep refrigerated products refrigerated and shelf-stable ones somewhere cool; do not leave them baking in a carFollow the course used in the trials: for preventing antibiotic-associated diarrhea, take it throughout the antibiotic course; for a chronic problem such as irritable bowel syndrome, give it at least 4 weeks, and switch only if it does not workDo not mix several brands at once: it is wasteful, and you cannot tell which one is doing anythingOnce symptoms improve, you can taper off gradually, or switch to fermented food to keep things going
These are mostly rules of experience, but each one can be explained by the mechanisms covered earlier:
With or after a meal: food dilutes the strong acid in the stomach and nudges the pH up a little, so the stomach takes a smaller cut. The same meal also brings the bacteria some food.A few hours apart from the antibiotic: antibiotics do not tell friend from foe, and the strain you added looks the same to them as the one they are hunting. Spacing the doses lets the drug concentration inside the gut fall first, so the two do not collide head-on. *S. boulardii* is the exception that does not need spacing, because it is a yeast and the antibiotic's sights cannot find it.Refrigerated products have to stay cold: live bacteria are organisms kept alive by a set of enzymes, and the higher the temperature, the faster those enzymes break down and the fewer live cells are left by the time they reach your mouth. Leave them in a car in the sun and the number you paid for evaporates.Keep taking them for a stretch: a passer-by does not colonize, so the whole effect depends on being present. Stop and the worker has left, so you keep supplying them; and that is exactly why there is no such thing as one course and you are set for life.Do not mix brands: the clinical evidence was built strain by strain. Mix them and you get the full dose of no single strain, and you cannot tell which one worked, so next time you still will not know what to buy.After improvement, switch to fermented food to keep it going: once symptoms settle, what you need is no longer a temp worker holding empty seats but a good working environment for the residents. Fermented food and fiber do exactly that, without having to get past stomach acid every day.
Put side by side, these rules say only one thing: what you are buying is a stretch of presence, not a permanent makeover. So every detail of how you take them is about making that stretch longer and more complete, and easier for you to observe yourself.
Red flag · When to stop and see a doctor
Red flags (stop taking them and see a doctor):Bloating or diarrhea that gets worse after starting and does not settle: it is not right for youFever, chills or feeling persistently unwell: get an emergency assessment (rare bacteremia)Critical illness, immunosuppression, a central venous line, severe pancreatitis: stop, and do not take them on your ownPreterm infants: do not give them yourself. The US FDA issued a warning about this in September 2023, and in the US no probiotic is approved for use in infants; whether a preterm infant in hospital gets one is for the neonatal team to decide
Wellness-grade and pharmaceutical-grade probiotics: products sold as food supplements are only loosely regulated by both the US FDA and China's National Medical Products Administration (NMPA), and the gap between the label and the actual content can be small or large, so prefer products with third-party certification (USP, NSF, ConsumerLab). Pharmaceutical-grade (prescription) products come with approved indications, regulation and randomized-trial data. For most everyday uses a wellness-grade product is enough; for critically ill patients or treatment, look for pharmaceutical grade.
fda-2023-probiotics-preterm-warning
In practice · The whole story in a few lines
It comes down to these points:A probiotic is not a good-bacteria supplement but a precision tool that only works when the strain, the use and the time window all matchFor most people most of the time, food (fermented food, fiber and a varied diet) is worth doing before capsulesFecal microbiota transplant is real microbiome therapy, but only for specific indications; do not be taken in by talk of home stool transplantsMost people do not need a daily probiotic capsule as insurance; spend the money on things that pay back more (vegetables and fruit, exercise, sleep)
A healthy gut is not the result of eating more good bacteria but of giving the microbiome in your body a good working environment.
That good working environment is enough fiber, a variety of plants, fewer unnecessary antibiotics and less ultra-processed food, plus enough sleep and well-managed stress.
These things are cheap, can be kept up for the long term, and have almost no side effects. They are the real alternative to probiotics.
References · 11
- Hill, C., Guarner, F., Reid, G., et al. (2014). The ISAPP consensus statement on the scope and appropriate use of the term probiotic. Nature Reviews Gastroenterology & Hepatology, 11, 506-514. A probiotic benefit is strain- and dose-specific, not a property of all fermented foods. 10.1038/nrgastro.2014.66
- Zmora, N., Zilberman-Schapira, G., Suez, J., et al. (2018). Personalized gut mucosal colonization resistance to empiric probiotics is associated with unique host and microbiome features. Cell, 174(6), 1388-1405.e21. Invasive sampling in mice and healthy humans during an 11-strain probiotic or placebo. Probiotics remained viable after passing through the gut, but human mucosal colonization was person-, region- and strain-specific, predicted by baseline host and microbiome features, and indistinguishable by probiotic presence in stool; the mucosal microbiome only partially correlated with stool. The effects on the mucosal community and gut transcriptome were transient and individualized (abstract, PMID 30193112). 10.1016/j.cell.2018.08.041
- Suez, J., Zmora, N., Zilberman-Schapira, G., et al. (2018). Post-antibiotic gut mucosal microbiome reconstitution is impaired by probiotics and improved by autologous FMT. Cell, 174(6), 1406-1423.e16. After antibiotics, a multi-strain probiotic colonized the human mucosa more readily but induced a markedly delayed and persistently incomplete reconstitution of the native stool and mucosal microbiome and host transcriptome compared with spontaneous recovery, whereas autologous fecal microbiome transplantation gave a rapid, near-complete recovery within days (abstract, PMID 30193113). 10.1016/j.cell.2018.08.047
- Guo, Q., Goldenberg, J. Z., Humphrey, C., & El Dib, R. (2019). Probiotics for the prevention of pediatric antibiotic-associated diarrhea. Cochrane Database of Systematic Reviews, 4, CD004827. 10.1002/14651858.CD004827.pub5
- Su, G. L., Ko, C. W., Bercik, P., Falck-Ytter, Y., Sultan, S., Weizman, A. V., & Morgan, R. L. (2020). AGA Clinical Practice Guidelines on the Role of Probiotics in the Management of Gastrointestinal Disorders. Gastroenterology, 159(2), 697-705. 10.1053/j.gastro.2020.05.059
- Collinson, S., Deans, A., Padua-Zamora, A., Gregorio, G. V., Li, C., Dans, L. F., & Allen, S. J. (2020). Probiotics for treating acute infectious diarrhoea. Cochrane Database of Systematic Reviews, 2020(12), CD003048. Duration of diarrhoea: mean difference 8.64 hours shorter (95% CI 29.4 hours shorter to 12.1 hours longer; 6 trials, 3058 participants), certainty very low. 10.1002/14651858.CD003048.pub4
- Ford, A. C., et al. (2018). Systematic review with meta-analysis: the efficacy of prebiotics, probiotics, synbiotics, and antibiotics in IBS. Alimentary Pharmacology & Therapeutics, 48(10), 1044–1060. Note: despite the id, published in Alimentary Pharmacology & Therapeutics. 53 RCTs of probiotics (5,545 patients) up to July 2017. Particular combinations, species or strains appeared to benefit global symptoms and abdominal pain, but which ones are effective remains, for the most part, unclear; prebiotic and synbiotic data were sparse; rifaximin had modest efficacy in non-constipated IBS (RR of persisting symptoms 0.84) (abstract, PMID 30294792). 10.1111/apt.15001
- Wastyk, H. C., Fragiadakis, G. K., Perelman, D., Dahan, D., Merrill, B. D., Yu, F. B., Topf, M., Gonzalez, C. G., Van Treuren, W., Han, S., Robinson, J. L., Elias, J. E., Sonnenburg, E. D., Gardner, C. D., & Sonnenburg, J. L. (2021). Gut-microbiota-targeted diets modulate human immune status. Cell, 184(16), 4137–4153.e14. 10-week diet intervention within a 17-week protocol: generally healthy adults randomized to a high-fiber or a high-fermented-food diet (n = 18 per arm per the summary; 39 assigned in all). The primary outcome, the cytokine response score, did not change significantly in either arm. The fermented-food arm rose from 0.4 to 6.3 ± 2.9 servings/day and showed increased microbiota diversity and decreased inflammatory markers; the high-fiber arm increased glycan-degrading CAZymes while community diversity stayed stable (full text, PMC9020749). 10.1016/j.cell.2021.06.019
- van Nood, E., Vrieze, A., Nieuwdorp, M., et al. (2013). Duodenal infusion of donor feces for recurrent Clostridium difficile. The New England Journal of Medicine, 368(5), 407-415. Stopped after an interim analysis. Duodenal infusion of donor feces (after 4 days of vancomycin and bowel lavage) resolved recurrent C. difficile diarrhea in 13/16 (81%) after the first infusion, and in 2 of the remaining 3 after a second donor; vancomycin alone 4/13 (31%), vancomycin with bowel lavage 3/13 (23%) (abstract, PMID 23323867). 10.1056/NEJMoa1205037
- DeFilipp, Z., Bloom, P. P., Torres Soto, M., et al. (2019). Drug-resistant E. coli bacteremia transmitted by fecal microbiota transplant. The New England Journal of Medicine, 381(21), 2043-2050. Two immunocompromised recipients of stool from one donor developed ESBL-producing E. coli bacteremia; one of the two died. 10.1056/NEJMoa1910437
- U.S. Food and Drug Administration. (2023). FDA approves first orally administered fecal microbiota product for the prevention of recurrence of Clostridioides difficile infection [Press release, April 26, 2023]. Vowst is the first oral fecal microbiota product (four capsules once daily for three days); Rebyota, approved 2022, is a rectal suspension. www.fda.gov/news-events/press-announcements/fda-approves-first-orally-administered-fecal-microbiota-product-prevention-recurrence-clostridioides