This page has been folded. Fermented foods are not one exposure — the lay question splits along the food that was fermented, and each strand now lives with the food it belongs to.
- The gut / live-culture question — is the effect the live cultures, and must they be alive? — now lives in Gut Microbiome. It carries the live-versus-pasteurized discriminator (no held source runs the dead-culture control), the strain- and dose-specific bar that a mixed live ferment does not obviously clear (WGO), the Wastyk diversity finding read as indirect remodeling rather than colonization, and the trendy ferments (kimchi, kombucha, sauerkraut, kefir) as a named insufficient-evidence gap — unstudied, not shown inert.
- Fermented dairy — the yogurt and fermented-milk cardiovascular and type-2-diabetes associations (Zhang, Guo, Gijsbers) — is a dairy question, worked in Dairy alongside the rest of the dairy subtypes, where the non-independence of the overlapping cohorts is handled in one place.
- Fermented soy (miso, tempeh, natto) is a distinct exposure appraised in Soy Products, not covered by the dairy evidence.
The short answer is unchanged: the fermented-food health halo runs well ahead of what has been measured. Fermented dairy carries a modest observational signal that is really a dairy signal; the fashionable ferments have almost no hard-outcome human evidence (insufficient evidence, not a null); and whether the live cultures are doing the work is untested. Treat the halo as a claim to check, not a credential — and note this is a small lever, well below the big rocks.
Evidence box
Question ’What does the evidence show about live-fermented foods” effect on each patient-important outcome — direction, magnitude, for whom, how certain — and where does any effect come from: the live cultures, the food matrix, or the biochemical modification fermentation performs (and must the cultures be alive)?’ Evidence included 0 sources Overall certainty Not stated (see Rating Certainty of Evidence) Source-selection note All sources are gold or high tier. Last updated 2026-09-09 · Independently reviewed: No · Full edit history