The gut microbiome gets invoked for almost everything: immunity, mood, weight, autoimmune disease, leaky gut, candida. The evidence tells a narrower story than the marketing, and a more useful one. The microbiome is genuinely modifiable and genuinely matters in a handful of well-defined places. In most of the places it is sold for, the evidence stops at a bacterial census — a readout of which microbes are present — that no one ever follows through to an outcome a person can feel.
The bottom line
- The biggest dietary lever on your microbiome is fermentable fibre — and that is already the fibre recommendation. Eat plenty of fibre from a plant-diverse diet — the general fibre intake target (Dietary Fibre and Health, where the floor and its studied-range caveat are stated); feed your microbiome and eat enough fibre are the same instruction. (Diet is one modulator among several the evidence names, medication being another; the ranking of fibre as the top dietary lever is the wiki’s synthesis.)
- It altered the microbiome is a surrogate, not a health benefit. Almost every microbiome study reports that an intervention changed which bacteria are present, and stops there. That is a change in a readout, not in your health, until someone follows it through to an outcome and shows the chain transmits (Surrogate Outcomes).
- Probiotics work strain-by-strain and indication-by-indication — not as a general gut-health tonic. There are real, specific wins (below); take a probiotic for immunity is not one of them, though for general immunity and mood the honest state is weak-to-suggestive benefit, heavily discounted — not a blank.
- Prebiotics are mostly fibre by another name, so their outcome case largely reduces to the fibre evidence.
- Faecal microbiota transplant (FMT — transferring a healthy donor’s stool microbes) is real medicine, but so far for one thing: recurrent C. difficile infection. Everything else is experimental.
- Leaky gut syndrome and chronic candida overgrowth as general, diet-treatable causes of systemic disease fail the basic test — is there a measurable claim on a real outcome? — and must be kept separate from the real medical conditions that share their names.
- The microbiome is discussed far out of proportion to its established effect sizes. Here, attention is an anti-signal: rank the topic low, and put your effort into the levers (fibre, diet quality) that were already ranked without the word microbiome attached (Layer 1 - Ranking Interventions for a Stratum).
Confidence is medium. The probiotics-by-indication claims rest on a gold-standard guideline and the FMT-for-C. difficile claim on a held landmark RCT, but the broad microbiome-and-health story is mostly observational, and the field’s central confound — which way the causal arrow points — is usually unresolved.
§2 — The one move that decides most of it: composition-shift is a surrogate
Almost every microbiome study reports a composition change — which bacteria are present, and how diverse the community is — and stops there. That is a surrogate, not a health outcome. Valdes states the limit plainly: «observational studies can show cross-sectional associations between microbes and health traits but are limited by the inability to measure causal relations. The strongest level of evidence is obtained from interventional clinical studies — in particular, randomised controlled trials.» (Valdes et al., 2018) So low diversity is a marker of dysbiosis — a disturbed microbial community — not a demonstrated cause of disease. A shift in the bacterial census earns belief only when it is followed through to something a patient feels: an infection cured, a disease averted (Surrogate Outcomes).
The arrow’s direction is usually unresolved, and it is the field’s central confound. Does the disturbed community cause the illness, or does the illness — and the diet behind it — reshape the community? Flag this; do not pretend to resolve it.
The template done right is Suez 2022 (Non-Sugar Sweeteners). Its 120 healthy, sweetener-avoiding adults took saccharin, sucralose, aspartame or stevia for two weeks; each «distinctly altered stool and oral microbiome and plasma metabolome, whereas saccharin and sucralose significantly impaired glycemic responses» (Suez et al., 2022). The trial did not stop at the shift: transplanting the human microbiota into germ-free mice — animals raised with no microbes of their own — reproduced the donors’ glucose responses, pinning the effect as causal rather than reverse causation.
Even this exemplar is narrow. The endpoint is a two-week glycemic response, not diabetes; the effect is personalized, with responders and non-responders in each arm; and causal-on-the-marker is not causal-on-the-disease. A microbiome study that clears the surrogate bar this cleanly is the exception, and it still cannot license a claim about a hard outcome.
§3 — What actually moves the microbiome: diet, and mostly fibre
The dominant modifiable lever on the microbiome is diet, and within diet, fermentable fibre — the fraction gut bacteria ferment into short-chain fatty acids (SCFA: acetate, propionate, butyrate). But the popular shorthand “more fibre, more diversity” is wrong as stated. Valdes is explicitly two-sided: «Many short term feeding trials with purified dietary fibres or even whole plant based diets either have no effect on microbiota diversity or reduce it, but can still have clinical benefits.» (Valdes et al., 2018) The diversity link in fact runs the other way round: long-term weight gain «correlates with low microbiota diversity, and this association is exacerbated by low dietary fibre intake» (Valdes et al., 2018) — low fibre worsens that association, rather than fibre lifting diversity directly.
So the microbiome case for a fibre-rich, plant-diverse diet still holds, but it travels through SCFA and clinical benefit, not through a simple diversity gradient (Dietary Fibre and Health). The outcome numbers behind fibre are real but split by evidence grade. The large ones are observational, from self-reported intake: highest-versus-lowest fibre tracks roughly 15-30% lower risk — all-cause mortality RR 0.85, CHD 0.76, type 2 diabetes 0.84, colorectal cancer 0.84 (Reynolds et al., 2019) — so they carry the healthy-user confound and dietary measurement error (Measurement Error in Dietary Assessment). The trial-grade effects are causal and small: viscous fibre lowers LDL by about -0.057 mmol/L per gram over the practical <=10 g/day range (a roughly linear slope that plateaus above ~10 g), and Brown’s 67-trial meta-analysis concludes soluble fibre «can make only a small contribution to dietary therapy to lower cholesterol» (Brown et al., 1999).
A 10-week head-to-head sharpens the timescale. In Wastyk 2021 (Fermented Foods and Health), the high-fermented-foods arm steadily raised alpha diversity — the richness of species within one person’s gut — while in the high-fibre arm «alpha diversity … did not change cohort-wide over the course of the intervention» (Wastyk et al., 2021). This is a distinction on horizon — a 10-week intervention versus a long-term association — not a contradiction of Valdes; fibre’s diversity effect plausibly needs longer than a few weeks. And the fermented arm’s rise «was not primarily due to consumed microbes but rather a result of shifts in or new acquisitions to the resident community» (Wastyk et al., 2021) — indirect remodeling of the community already present, not colonization by the microbes eaten.
The two routes are genuinely different. Fibre feeds the resident community (the prebiotic route); fermented foods add microbes and their metabolites (the probiotic route). Both are measured here only at the surrogate level — a diversity shift, not an outcome.
§4 — Probiotics: strain- and indication-specific, not a general tonic
“Take a probiotic for gut health” is not a claim the evidence supports. The WGO guideline’s governing discipline is that «the effects of probiotics are strain-specific and dose-specific» (World Gastroenterology Organisation et al., 2023); the evidence attaches to specific strain x indication x dose combinations, not to the word “probiotic” — live microbes taken to confer a benefit. Where the evidence is real:
| Indication | WGO evidence status | Note |
|---|---|---|
| Necrotizing enterocolitis (preterm neonates) | benefit — less NEC and reduced death, NNT 20 | «not all probiotic preparations tested are effective»; moderate certainty |
| Antibiotic-associated diarrhoea | benefit (specific strains) | a real prevention win |
| C. difficile-associated diarrhoea (prevention) | benefit (specific strains) | prevention, distinct from FMT treatment |
| Acute infectious diarrhoea | benefit (specific strains) | modest, strain-specific |
| IBS | symptom benefit for certain strains | strain-specific effects |
| Pouchitis | benefit (specific preparations) | — |
| H. pylori eradication (adjunct) | reduces therapy side-effects | «quality of the evidence was weak» |
(World Gastroenterology Organisation et al., 2023)
The standout patient-important outcome is in preterm infants: «The number needed to treat to prevent one death from all causes by treatment with probiotics is 20» (World Gastroenterology Organisation et al., 2023). But the same guideline warns «not all probiotic preparations tested are effective», so this is a strain-quality claim, not a blanket endorsement of the category.
Two caveats travel with every row. First, pooling trials across different strains without a shared-mechanism rationale «should be avoided» — a strain-blind meta-analysis can manufacture or wash out an effect, the Is the Food Category Doing Any Work problem at the level of the organism: the strain is the exposure, not the label. Second, swallowed probiotics are typically transient colonizers — they pass through rather than take up permanent residence, so an effect generally lasts only as long as dosing does.
For general immunity and mood the honest state is not “no evidence” — it is weak-to-suggestive benefit, heavily discounted. WGO reports «suggestive evidence that several probiotic strains and the prebiotic oligofructose are useful in improving the immune response» (World Gastroenterology Organisation et al., 2023), and Valdes reports meta-analytic benefit for upper-respiratory infection, eczema prevention and depressive symptoms. Both are undercut by the same problem — strain-blind pooling and heterogeneity, with Valdes noting the studies «were not homogeneous … which limits precise recommendations». The demotion to weak-to-suggestive is the wiki’s own reading of these two sources — a discounted finding of benefit, not the insufficient-evidence state.
§5 — Prebiotics: mostly fibre by another name
WGO defines a prebiotic as «Prebiotic A selectively fermented ingredient that results in specific changes in the composition and/or activity of the gastrointestinal microbiota, thus conferring benefit(s) upon host health» (World Gastroenterology Organisation et al., 2023). Read against the food, that is largely fermentable fibre under another name — inulin, oligofructose and the like are the same substrate the fibre evidence already covers. So a prebiotic’s outcome evidence mostly is the fibre evidence (Dietary Fibre and Health), and a standalone prebiotic supplement, taken on top of an already-adequate fibre intake, is a marginal and mostly-unevidenced increment.
§6 — FMT: the field’s strongest win, and only for one thing so far
Faecal microbiota transplant (FMT — transferring a donor’s stool microbes into a patient’s gut) is the clearest proof that the microbiome can be causal in humans. In the van Nood 2013 open-label RCT, in patients with recurrent C. difficile infection and mostly elderly, a donor-feces infusion cured 13 of 16 (81%) on the first infusion and 15 of 16 (94%) overall, against 4 of 13 (31%) for a standard 14-day course of vancomycin and 3 of 13 (23%) for vancomycin plus bowel lavage. (van Nood et al., 2013) That is roughly a 63 percentage-point absolute gain, a number needed to treat of about 2.
The effect was so large that «The study was stopped after an interim analysis». (van Nood et al., 2013) Early stopping inflates the estimated magnitude, so read the exact size with caution — but the direction of a three-to-one cure advantage is not in doubt.
This is the surrogate done right, not a bare composition claim. The same trial measured the microbial shift and followed it to the clinical cure: Simpson’s diversity climbed from about 57 to about 179, into the donor range, as patients recovered, and the authors read the mechanism as «the reestablishment of the normal microbiota as a host defense against C. difficile». (van Nood et al., 2013) A change in the bacterial census (Surrogate Outcomes) earns belief here because it was carried through to whether the infection cleared.
The scope boundary is the decision-relevant part. Recurrent CDI is the one case where the pathology is the dysbiosis (a disrupted, low-diversity microbial community): antibiotics collapse gut diversity, and FMT restores it directly. That is exactly why it is both the cleanest causal win and a poor template for anything else. The CDI win is no evidence that manipulating the microbiome improves general health, weight, or metabolic disease, where no established dysbiosis-as-cause exists for a transplant to reverse. Valdes states the field bound plainly: «For other pathologies, faecal transplants are not yet clinical practice but have been explored.» (Valdes et al., 2018)
§7 — Antibiotics: a real perturbation, but the outcome question is a gap
Antibiotics genuinely disrupt the gut community — that much is not in question. What is missing is the outcome evidence: whether a course of antibiotics leaves a durable, patient-important harm through the microbiome is an insufficient-evidence gap, not a demonstrated null, and the fabric holds no source that settles it.
Two things are already settled without needing that answer. First, “don’t take antibiotics you don’t need” is sound advice on antibiotic-resistance grounds alone, independent of any microbiome cost. Second, where antibiotics are clinically indicated, the stakes of an untreated infection and of resistance dominate any speculative, unmeasured harm to the resident flora — so the microbiome argument changes no decision at the point of prescribing.
§8 — Leaky gut and candida: demarcating real physiology from the syndrome
A functional test sorts these claims: is there a falsifiable, quantified claim on a real human outcome?
Intestinal permeability is real and measurable, and it is genuinely altered in coeliac disease, inflammatory bowel disease, and some infections. “Leaky gut syndrome” is a different proposition — a general, diet-treatable increase in permeability said to drive systemic disease — and that is not established. Worse for the causal story, altered permeability may be a consequence of gut disease rather than its cause, the same direction-of-causation confound that runs through the whole field. Keep the real conditions distinct from the syndrome; the measurable pathology must not launder the unfalsifiable version.
Candida runs the same way. Medical candidiasis — oral thrush, invasive infection — is real and diagnosable. “Chronic candida overgrowth” corrected by an anti-candida diet makes no falsifiable, quantified claim on a measured human outcome, and sits on the Esoterik side of the demarcation line . The Camilleri intestinal-permeability review that would let the fabric adjudicate the real permeability findings against the syndrome is not held — a named gap.
§9 — Attention is an anti-signal — rank the topic low
The gut microbiome is discussed far out of proportion to its established, outcome-level effect sizes. By the attention-is-an-anti-signal rule (Layer 1 - Ranking Interventions for a Stratum), that volume is evidence about the field’s novelty and marketability, not about the size of the lever.
The genuine levers that run through the microbiome — fibre and overall diet quality — are already ranked, and ranked without needing the microbiome frame at all. Someone acting on fibre and a plant-diverse diet has captured the actionable part; the microbiome vocabulary adds discussion, not a new decision.
§10 — What the evidence cannot yet answer (the gaps)
- Direction of causation is the pervasive confound. For almost every dysbiosis-disease association, no one has resolved whether the disrupted community causes the illness or the illness (and its diet) causes the disruption. Flag it; the held evidence does not resolve it.
- Several backbone syntheses are unheld. A Cochrane FMT-for-CDI systematic review would pool the effect that the single, early-stopped van Nood trial over-estimates, with a proper confidence interval. A strain-level probiotics SR, the Camilleri intestinal-permeability review, and a microbiome-and-health umbrella review would each anchor a claim the fabric now carries on thinner ground.
- No source bears on antibiotics’ long-term microbiome harm, and none on birth mode and early-life colonization — both are named gaps, not answered questions.
§11 — How this compares to the popular framing (the guidance null)
The popular framing sells the microbiome as a general-purpose health dial: probiotic tonics for immunity and mood, “leaky gut” diets, microbiome supplements. On those marketed interventions this appraisal diverges — the outcome evidence for a general tonic is not there.
This appraisal converges with the disciplined clinical message. The World Gastroenterology Organisation’s central point is that «the effects of probiotics are strain-specific and dose-specific» (World Gastroenterology Organisation et al., 2023) — benefit attaches to a named strain, a named indication, and a dose, never to “probiotics” as a class. That is the Is the Food Category Doing Any Work problem at the level of the organism.
The largest real lever the microbiome offers is also the least marketed one: fibre and dietary diversity, the boring recommendation that is already made on its own evidence.
Caveats
- Open loop. This appraisal grades the coherence and source-fidelity of the evidence, never validity. Nothing here has been tested against a realized outcome; a clean appraisal is not a validated recommendation.
- Appraise, not prescribe. This is not a probiotic-brand guide, not a strain-and-dose prescription, and not clinical management of C. difficile, invasive candidiasis, or IBD — those are medical, and out of scope.
- General, applied per person. This is a population-level appraisal; the weighting of outcomes and the fit to any individual’s situation happen case by case.
- One axis. The wiki holds a health axis only. It says nothing about the cost, marketing, or ecology of the gut-health industry beyond whether its claims move a health outcome.
Evidence box
Question ’Which modifiable exposures (diet/fibre, prebiotics, probiotics by strain, fermented foods, antibiotics, FMT) change the gut microbiome in a way that reaches a patient-important outcome, how good is that evidence versus a composition-shift surrogate or a mechanism-only claim, and which microbiome claims are effectively unfalsifiable?‘ Evidence included 7 sources — 1 gold, 4 high, 2 moderate Overall certainty Medium (see Rating Certainty of Evidence) Source-selection note 2 source(s) below the gold evidence bar feed this page: Valdes (narrative review, moderate); Suez (RCT, moderate). Each labelled by tier; none load-bearing for the core claims. Last updated 2026-08-11 · Independently reviewed: No · Full edit history