How we evaluate scientific evidence

Winnow operates in a space where the science is moving quickly, public concern is real, and the distance between what has been shown in a laboratory and what has been demonstrated in a human body is still considerable. We think it matters to be explicit about how we read the research, what weight we assign to different kinds of evidence, and where the honest answer is still, simply, we do not know yet.

Not all evidence does the same kind of work. Some studies help clarify mechanisms. Some test whether a signal holds in a living system. Some show association in humans. Some begin to answer whether an intervention changes anything meaningful in practice.

The probiotic and microplastic binding story sits across all of those layers. That does not make the evidence weak. It makes it mixed, and worth reading with care. Below is how we think about the main categories of research in this area, what each contributes, and where their limits still are.

Mechanistic studies: how something could work

Mechanistic research asks a basic question: is there a plausible biological pathway? Does the chemistry make sense? Can we describe, at a molecular or cellular level, how an interaction might occur?

In the context of probiotic binding of microplastics, this includes work on bacterial surface properties, exopolysaccharide layers, surface hydrophobicity, and the interfacial chemistry that may allow a microbial cell to adhere to a plastic particle. It also includes work on how microplastics interact with the gut barrier itself, including mucus layers, tight junction proteins, and epithelial surfaces.

This kind of evidence is foundational. It tells you whether an idea is biologically coherent. A claim with no mechanistic basis is speculation. A claim with mechanistic support is a hypothesis worth taking seriously.

But mechanistic plausibility is not the same as proof in a living system. Chemistry that behaves one way in a controlled setting may behave differently in the complexity of a human gut. Mechanistic evidence tells you that something could happen. It may even tell you that something is likely to happen. But it is not a guarantee.

We treat mechanistic evidence as scientific groundwork. It helps shape formulation decisions and supports the idea that the binding hypothesis is grounded rather than invented.

In vitro studies: what happens in the dish

In vitro studies take the next step. They create controlled conditions, expose bacterial cells to plastic particles, and measure what happens. Does binding occur? How much? How quickly? Does it vary by strain, particle size, polymer type, or environment?

This is where much of the strongest current evidence for probiotic-microplastic interaction lives. In vitro studies have shown measurable binding across multiple polymer types. They have shown that binding capacity can vary dramatically between strains, sometimes by more than tenfold even within the same species. They have also helped identify the surface features that appear to matter most, including exopolysaccharides, S-layer proteins, and cell-surface hydrophobicity.

In vitro evidence is valuable because it is controlled, reproducible, and precise.

Its limitation matters just as much. A dish is not a gut. In vitro systems remove much of the biological complexity that defines real life: competing microbes, mucus, bile salts, transit, immune signaling, food matrices, and the protein corona that forms around particles in biological fluids. A strain that binds strongly in a clean buffer may behave quite differently in the crowded, dynamic environment of an intestine.

Nevertheless, in vitro data is a key part of strain selection. It is also the level at which we describe binding most directly. When we say Winnow’s strains bind microplastics in laboratory testing, this is the category of evidence we are referring to.

Animal models: what happens in a living system

Animal studies add biological complexity. They can show whether a probiotic survives transit, whether binding occurs inside a functioning gut, whether barrier function changes, whether microbiome composition shifts, and whether downstream markers move with it.

In the microplastic-probiotic literature, animal models have shown that certain strains can reduce intestinal permeability, restore tight junction protein expression, shift microbiome composition toward more favorable profiles, and in some cases increase fecal excretion of plastic particles. These are meaningful findings. They suggest that binding is not only a surface-chemistry phenomenon, but rather translates into measurable biological effects.

While similar in this context, animal data is not identical to human data. Rodent gut transit times, microbiome structure, immune signaling, diet, and exposure patterns differ from ours. Animal studies can enormously strengthen a hypothesis. But once again, the transfer to humans is not 100% guaranteed. Biomedical history is full of interventions that looked strong in animal models and then failed, or changed character, in human trials.

Observational human studies: what patterns appear in people

Observational research looks at what is happening in actual human populations. In the microplastics field, that includes detection studies finding plastic particles in stool, blood, placenta, arterial plaque, brain tissue, and breast milk. It also includes epidemiology examining associations between microplastic exposure and health outcomes.

This category of evidence has changed the conversation. Detection studies have made it increasingly difficult to argue that microplastics are only an environmental issue and not a human one. Observational findings have also made the health conversation harder to ignore.

But observational evidence has real limits. Detection is not causation. Finding microplastics in arterial plaque does not yet prove they caused a cardiovascular event. Correlations in population studies can reflect confounding, reverse causation, or variables that were never measured. Observational data can show that two things occur together. It cannot, on its own, tell you why.

Randomized controlled trials

The strongest way to show that an intervention works in humans is a well-designed randomized controlled trial. You take a group of people, randomly assign some to the intervention and others to a control, and measure outcomes in a way that reduces bias as much as possible.

We do not have that for probiotic binding of microplastics in humans. No one does.

That fact matters, and we think it should be stated plainly rather than buried or worked around.

The absence of human trial data does not mean the hypothesis is wrong. It means it has not yet been tested at the level that would justify the strongest claims, such as guaranteed removal and detox. The mechanistic evidence is coherent. The in vitro evidence is encouraging. The animal data is supportive. But the final tier, the one that would allow anyone to say with confidence that this works in people, does not yet exist.

That is not unusual for a young field. Human trials are expensive, slow, and methodologically difficult, especially when the exposure is ubiquitous, there is no negative control as everyone is exposed, and the mechanism sits at the intersection of microbiology, materials science, and gastroenterology. But the youth of a field does not change the evidentiary standard. It only explains why the gap remains.

We acknowledge the gap directly. We do not describe Winnow as clinically-proven or clinically-tested for microplastic binding in humans. We do not imply that trial evidence exists when it does not. We describe what our strains do in preclinical laboratory testing, because that is what the evidence currently supports.

An honest look

In a field like this, we think credibility comes from naming uncertainty rather than hiding it.

Here is what we do not yet know.

Whether binding in the gut meaningfully reduces the total systemic burden of microplastics in a human body over the long term. The mechanism is plausible and the animal data is meaningful. Human confirmation does not yet exist.

Whether the amount of binding that occurs during ordinary gut transit is large enough to matter at a population level. In vitro studies control dose and exposure time. A human gut is a moving target, shaped by transit, food, bile, mucus, and microbial competition.

How long microplastics persist in the body once they move beyond the gut. Detection studies show presence in distant tissues. Lab studies show that cellular degradation happens slowly. They do not yet tell us what happens over a lifetime of constant exposure.

Whether some polymer types, particle sizes, or exposure routes matter more than others for human health. The literature increasingly suggests that smaller particles may behave differently from larger ones, and that polymer chemistry matters. But the dose-response relationships for each and every variant in humans remains a challenge.

Whether the correlations seen in human epidemiology reflect causation. Some findings are striking. Others are scary. They are not yet definitive.

These are not rhetorical hedges. They are the actual boundaries of the evidence as it stands. We believe the direction of the research is concerning enough to justify careful action. We also believe that pretending the science is further along than it is would eventually weaken trust in the entire field.

Why this matters for how we communicate

This framework shapes how we write, how we explain, and how we make claims.

When we say Winnow’s strains bind microplastics in laboratory testing, that language reflects our laboratory evidence. When we describe what animal studies have shown about gut barrier protection, we say animal studies have shown it. When we discuss the broader concern around microplastic exposure, we cite human observational work and try to be explicit about what it does and does not establish.

We are not doing this only because compliance matters, though it does. We are doing it because we think it is the right way to talk about science in a field where the stakes are high and the evidence is still developing.

In an uncertain field, trust is usually lost when certainty gets ahead of the data.

Our aim is simpler than that: read carefully, speak precisely, be transparent, and leave room for what is still unknown.

Good in. Bad out. Every day.