The evidence ladder
Research on an ingredient generally climbs a ladder. Each rung is more expensive, slower and more convincing than the last — and most ingredients never reach the top.
- In vitro (cell culture). The compound is applied to cells in a dish. Cheap, fast, useful for working out mechanism. Tells you almost nothing about what happens in a living animal, because it skips digestion, absorption, metabolism and dose entirely.
- Animal models. Usually rodents. The compound is given to a living system, so digestion and metabolism are in play. The model species is chosen for convenience, not because it resembles your dog.
- Target-species trials. Actual dogs, actual cats. This is the rung that matters, and it is the one most often missing.
- Replication. More than one group, in more than one population, finding the same thing. This is where confidence properly comes from, and in companion animal nutrition it is rare.
A cell-culture or rodent result described in language that implies a benefit in your dog. It is technically true that the ingredient "has been shown to" do the thing — in a dish. The rung of the ladder is the whole story, and it is almost never stated on a label.
Why species extrapolation is genuinely risky
Moving a result from one species to another is not a formality. Real, well-known differences make it a hazard.
- Cats are not small dogs. As obligate carnivores they have a distinctly different metabolism, including limits on how much taurine they can make and how they handle certain compounds. Ingredients that are routine for dogs can be inappropriate for cats.
- Dogs are not small humans. Body size, gut transit time, and enzyme activity all differ. Dose does not scale linearly with weight, and a per-kilogram conversion from a human trial is an estimate, not a translation.
- Breed differences exist within a species. Some breeds carry drug-sensitivity variants that affect what crosses into the brain. Population averages hide individual animals.
None of this means human research is useless. It means it is a starting hypothesis about your animal, not a finding about your animal.
The vocabulary that separates strong studies from weak ones
- Controlled
- There is a comparison group that did not get the ingredient. Without one, you cannot separate the supplement from time, season, or a change in the weather.
- Randomised
- Animals were assigned to groups by chance, not by choice. Prevents the healthiest animals landing in the treatment group.
- Blinded
- Owners, and ideally the assessing veterinarian, do not know which animal got what. This matters enormously — more on why below.
- Placebo-controlled
- The comparison group gets an identical-looking product with no active ingredient.
- Crossover
- Each animal receives both treatment and placebo in sequence, acting as its own control. Efficient with small groups.
- Powered
- Enough animals were enrolled to detect a real effect if one exists. Small studies routinely miss real effects and also produce false ones.
The placebo problem belongs to the owner
Here is the wrinkle unique to veterinary research: your dog has no opinion about whether the supplement is working. You do.
Owner-reported outcomes — "he seems more comfortable", "she's playing more" — are the most common measure in pet supplement studies because they are cheap to collect. They are also the most susceptible to expectation. An owner who spent money and hope on a product tends to see improvement, and this is not a character flaw; it is how human perception works.
This is exactly why blinding matters more here than almost anywhere else, and why objective measures carry so much more weight. Force-plate gait analysis measures how much weight a dog puts on a limb. Accelerometers count actual activity. Blood work measures markers. None of these care what the owner hoped for.
When you read that a supplement "improved comfort in 80% of dogs", the question worth asking is: measured how, and compared to what?
Why companion animal research is thin
It is worth understanding the economics, because they explain the state of the literature better than any conspiracy would.
Human drug trials are funded by the prospect of a patented, reimbursed medicine. Livestock research is funded by production economics at industrial scale. Companion animal supplement research has neither. Supplements generally cannot be patented, insurance rarely reimburses them, and the market — while large — is fragmented.
The result is predictable: studies tend to be small, short, single-site, and frequently funded by a company with a commercial interest in the outcome. Industry funding does not automatically invalidate a study, but it is a fact you should weigh, and a well-run company will disclose it rather than hope you do not ask.
Five red flags in a claim
- "Clinically proven" with no citation. Proven is a strong word. A company with a study is usually delighted to name it.
- A study on the ingredient, sold as a study on the product. These are different things. A finished formula is not the sum of its citations.
- No dose stated. If you cannot compare the amount in the product to the amount in the study, the citation is decorative.
- Disease language on a supplement. Treats, cures, prevents, reverses. A supplement is not evaluated by the FDA to do any of those, and using that language is a regulatory problem as well as an honesty one.
- Testimonials standing in for data. Reviews tell you about satisfaction and palatability, which are real and worth knowing. They do not tell you about efficacy.
Published research on an ingredient is not a claim about any particular product — including ours. What a company can legitimately show you is what is in the box and at what amount, which brings us to certificates of analysis.
Frequently asked
No, but it is a real consideration. Much of the nutrition research that exists is industry funded, because few other bodies pay for it. Look at the methods rather than the funder: was it randomised, blinded, controlled, and did it use objective measures? A well-designed industry-funded study beats a sloppy independent one.
It depends on the effect being measured — a large, obvious effect needs fewer animals than a subtle one. In practice, companion animal trials often enrol 20 to 60 animals, which is enough to detect a large effect and easily small enough to miss a modest one. Treat a single small study as a signal worth following, not a settled answer.
PubMed indexes most peer-reviewed veterinary and nutrition research and is free to search. Abstracts are usually public even when full papers are paywalled, and the abstract will normally tell you the species, the design and the sample size — the three things that matter most.
That you cannot evaluate the claim. That is not proof of anything bad, but the burden sits with the seller. A company confident in its formulation will tell you the ingredient amounts and show you the batch testing without a fight.
Not sure it is right for your pet?
Our eligibility screener checks your animal's medications and conditions against every ingredient on the formulation sheet. One minute, eight questions, and it will tell you if we are not a fit.