A Menopause Probiotic Moved Symptom Scores by One Point in 245 Women

It is late, and a woman in her early fifties is awake again - the second time tonight, sheets damp, the thermostat blameless. On the phone in her hand is a shopping page for a capsule that promises menopause support, and beneath it forty more that promise the same thing in slightly different type. Somewhere in the middle of that scroll is the week's headline, which said a daily probiotic might ease menopause symptoms, and which is the reason the page is open at all.

The word doing the work in that headline is might. Underneath it sits a real trial - randomised, placebo-controlled, published in the journal Menopause on 4 August 2026 - and the trial is both more specific and more mixed than the coverage built from it. It tested three named bacterial strains at a stated dose, not the category on the label. And it reported a result on one of its two registered primary measures while the other stayed flat, which is the single most useful fact about it and the one least likely to survive a headline.

The study was run by AB-Biotics, the Spanish company that makes the strains, and registered as NCT06604234 before it began. It enrolled 245 women aged 45 to 60 who were in perimenopause or postmenopause - within five years of their last period - through a site in Madrid, running from September 2024 to a primary completion date of 14 December 2024. Entry required a Cervantes quality-of-life score of at least 38, so every participant arrived already carrying a measurable symptom burden. Women who had used hormone therapy in the previous three months were excluded, as were those with recent antibiotic or probiotic use, prior hysterectomy or ovariectomy, active cancer, type 1 diabetes, untreated thyroid disease, severe renal, cardiac or hepatic disease, and chronic gastrointestinal conditions.

Allocation was two-to-one and quadruple-masked: 160 women to the probiotic, 85 to a maltodextrin placebo, with participants, care providers, investigators and outcome assessors all blinded. The unbalanced split is a design choice with a cost. It puts more participants on the active product, which is often easier to recruit for, but for a fixed total of 245 it buys less statistical power than an even split would have - the smaller arm sets the ceiling on what the comparison can detect.

The intervention was 1x10^9 colony-forming units a day of a formula combining Levilactobacillus brevis KABP052, Lactiplantibacillus plantarum KABP051 and Pediococcus acidilactici KABP021, for three months. The authors describe the trial as decentralised, using what they call a real-world approach. Two instruments were registered as primary outcomes - change in the Cervantes Quality of Life Scale and change in the Menopause Rating Scale, both measured from baseline through 90 days. The Utian Quality of Life questionnaire, the subscales of the other two, gastrointestinal symptoms and vaginal infection symptoms were all registered as secondary.

In the full analysis set, the Menopause Rating Scale II total score fell further in the probiotic group than in placebo across the study, with a global mean difference between groups of -1.00 [0.47] and a P value of 0.033. The bracketed figure is the standard error, a reading the P value is consistent with. The somato-vegetative domain - the physical-symptom cluster - accounted for a clear share of that, at -0.54 [0.20], P=0.007. In the per-protocol analysis, which counts only the participants who completed the protocol as written and is therefore the more forgiving population, the psychological domain also separated, at -0.51 [0.25], P=0.040.

The Cervantes scale showed no significant difference between the groups. That is the sentence the coverage left out, and it carries more weight than a null usually would, because the Cervantes was not an afterthought: it was registered as a co-primary outcome alongside the Menopause Rating Scale, and it was the instrument used to decide who was symptomatic enough to enrol. A trial that nominates two primary yardsticks and clears one of them has produced a split decision, not a clean win.

The third instrument landed on the positive side. The change in Utian Quality of Life score was higher in the probiotic group, at 2.21 [1.07], P=0.040 - a secondary outcome, and one whose P value sits close enough to the conventional threshold that it would not survive much correction for the number of comparisons the trial ran. The authors' own conclusion is appropriately narrow: the formula was associated with improvements in menopause-associated symptoms and quality of life, the global Cervantes score appeared similar between groups, and future investigations are required to confirm the clinical effect and its mechanism.

The formula was not assembled at random, and the reasoning behind it is the most interesting science in the file. Estrogens leave the body conjugated to glucuronic acid; certain gut bacteria carry an enzyme, beta-glucuronidase, that strips that tag in the gut and frees the hormone to re-enter the bloodstream. In 2024, a group including several of the same authors screened 84 strains across 16 species of lactic acid bacteria and bifidobacteria for exactly that activity, and reported that L. brevis KABP052 had the highest.

That earlier, exploratory trial measured hormones rather than symptoms. After twelve weeks, women on the probiotic formula held higher serum estradiol than placebo (31.62 plus or minus 7.97 pg/mL versus 25.12 plus or minus 8.17) and higher estrone (21.38 plus or minus 8.57 pg/mL versus 13.18 plus or minus 8.77), with the placebo group's levels declining over the same period. That is a coherent mechanistic story, and it is why the symptom trial exists.

It is also a chain with an untested link. The published abstract of the symptom trial reports no hormone measurements at all, so nothing in it demonstrates that the small movement on the Menopause Rating Scale came through the estrogen pathway rather than through any of the other routes - gastrointestinal comfort, sleep, or the placebo response that symptom questionnaires are famously sensitive to - that a three-month capsule could plausibly travel.

Set against the shelf it will be sold from, this trial is unusually good. Most of what ranks for menopause supplement queries is written by the people selling the capsules, cites nothing that tested the product in question, and names no strains at all. This one registered its outcomes in advance, blinded four ways, used a placebo, and published a result that was partly negative. The finding is small and it is honest about being small.

What the trial cannot do is transfer. Probiotic effects are strain-specific, and the evidence here attaches to three named organisms at 1x10^9 colony-forming units a day - not to a bottle marked women's blend, not to yoghurt, and not to a different formulation of the same species. A shopper holding the study up against a label is comparing two things that share a category and very little else. That is the practical distance between what was measured and what is on sale, and it is the reason a headline about probiotics and menopause is nearly always broader than the study underneath it.

This is a better piece of evidence than almost anything else supporting the menopause supplement aisle, and it is still thin. The tell is not the size of the effect - a one-point shift on a symptom scale is a plausible thing for a modest intervention to produce - but the split verdict across the trial's own two primary outcomes, which the coverage resolved in the favourable direction without mentioning that there was a split to resolve. Add the funder's authorship of the paper, the single country, the three-month horizon and the absence of posted results, and what is on the table is a small, single-trial signal attached to three specific organisms, not a demonstrated treatment for menopause. It moves the honest answer from nothing to something small and unconfirmed, which is a real move. What clinicians weigh alongside a result like this - symptom burden, the options with larger evidence bases behind them, cost, and what a person has already tried - is nowhere in this paper, and no symptom scale is going to settle it.