Most Probiotics Aren't Chosen for What They Actually Do, a Sweeping Pharmacy Survey Finds

Walk the supplement aisle at any large US pharmacy and the promises stack up fast. Gut health. Vaginal health. Immune support. Mood. Pick almost any bottle, turn it over, and you will find a list of bacteria with long Latin names, presented as if each one had earned its place through careful matching of microbe to malady. It turns out that, for the most part, it hasn’t.

Emma Glass, Glynis Kolling and Jason Papin at the University of Virginia went looking for the logic behind those labels. They pulled every over-the-counter probiotic on sale at CVS, Walgreens and Walmart, the three biggest pharmacy chains in the country, and catalogued what was actually inside. The haul came to 352 distinct products across 70 brands. And between all of them? Just 36 unique species of bacteria. More than half the products contained only one. The most crowded formulation managed seventeen.

That alone is a bit startling, given the sprawl of the shelf. But the more telling result came when the team checked whether the species inside a bottle had anything to do with what the bottle claimed to treat. They mapped the lot using a statistical technique that clusters products by their bacterial makeup, then coloured each one by its marketed use. If gut products leaned on one set of microbes and vaginal products on another, you would expect to see the colours separate into neat territories. They didn’t. The marketed uses smeared across the whole map with no real pattern, which is to say there is no agreed recipe linking a given combination of bacteria to a given health claim.

“It is truly fascinating to discover that these probiotic bacteria hold a unique, specialized niche among the trillions of microbes in and on the human body,” says Kolling, a research faculty member in UVA’s biomedical engineering department. The trillions matter here. We carry at least as many microbial cells as human ones, and the supplements on sale are sampling a vanishingly thin slice of that diversity, dominated by a few familiar names, mostly Lactobacillus, the genus that also sours your yogurt.

A map of what bacteria can do

Knowing which species sit in a bottle tells you remarkably little, though, because what actually matters is metabolism: the chemistry each microbe runs, the compounds it eats and the compounds it leaves behind. So the team built something to get at that. They call it HaPaPro, a collection of more than a thousand computer models, each a genome-scale reconstruction of a single bacterium’s metabolic wiring, spanning probiotics, disease-causing pathogens and the ordinary host-associated bugs that live on and in us without much fanfare.

Run those models and a gap opens up. The probiotic species, it turns out, cover only a narrow band of the metabolic range that host-associated bacteria collectively possess. They cluster tightly, biochemically speaking, doing similar things to one another while a vast functional territory sits unoccupied. The supplement industry, in other words, has been fishing from a very small pond. Worth saying that the US Food and Drug Administration has approved only two microbial products as actual therapeutics, both for stubborn Clostridioides difficile gut infections. Everything else on the shelf is a supplement, regulated far more loosely than a drug.

To show the models could do more than diagnose the problem, the researchers turned them on a concrete target: bacterial vaginosis, the common and miserably recurrent condition that sets in when the vaginal microbiome tips out of balance and a pathobiont called Gardnerella vaginalis takes hold. The consequences run well beyond discomfort, raising the risk of pregnancy complications, pelvic inflammatory disease and sexually transmitted infection. Antibiotics knock it back, but it comes roaring back in a great many women, and resistance is creeping in.

The acid that does the work

Here the modelling earned its keep. It pointed to vaginal microbes whose chemistry overlaps with Gardnerella‘s in just the right way to crowd it out, and the team took eleven of them into the lab to test directly. They grew each isolate, stripped out the cells, and let Gardnerella try to grow in the leftover broth. Some broths barely slowed it. Others shut it down almost completely. The difference came down largely to one molecule: D-lactic acid, a particular mirror-image form of the acid that keeps a healthy vagina slightly acidic and hostile to invaders.

The textbook story credits Lactobacillus with that protective acidity, and the textbook is not wrong. But the surprise was that Lactobacillus held no monopoly. Several non-Lactobacillus species, with names like Anaerococcus tetradius, churned out plenty of inhibitory D-lactic acid too. The protective trick, it seems, is a chemical function rather than a club membership, and that reframing is the whole point: design probiotics around what a microbe does, not what genus it happens to belong to. There is a catch, mind. Some of those potent acid-producers are themselves linked to dysbiosis, so they make lousy candidates for a supplement no matter how well they suppress the pathogen in a dish. When the team weighed inhibition against clinical safety, one species, Lactobacillus jensenii, a hallmark of the healthy vaginal community, came out as the standout candidate.

“It is remarkable how much microbes play a role in human health and well-being,” says Papin. “I love seeing how computational models of these complex biological systems are leading to new ideas for therapies and helping us understand such fundamental biological processes.” (Papin discloses a financial stake in Cerillo, which makes some of the instruments used in the experiments.)

What this points toward is a different way of building the bottle on the shelf, one where the bacteria are picked for a job they can demonstrably do rather than for tradition or manufacturing convenience. The same framework could be tuned to other niches and other patients; the team notes, for instance, that more than a fifth of vaginitis cases turn up in postmenopausal women, whose biochemistry differs enough that the best microbial target might differ too. “By combining our advanced methods, we have the potential to vastly expand the pool of beneficial bacteria and pave the way for targeted solutions to support human health,” says Kolling. Whether that pool reaches the pharmacy aisle, and how soon, is the open question. For now the labels keep promising. The science of making them mean something is only just catching up.

Frequently Asked Questions

Source: Glass, E.M., Kolling, G.L. & Papin, J.A. “Genome-scale metabolic modelling identifies vaginal microbiome members as potential probiotics.” Nature Microbiology (2026). https://doi.org/10.1038/s41564-026-02380-w