Explainer
Why animal results are not human results
When a vendor page says “studies show”, it is almost always describing work done in cells or in rodents. That work is real and it is not worthless — it is how research starts. But it does not establish that anything happens in a person, and the gap between those two statements is where most of the marketing in this category lives. This page sets out what each tier of evidence can and cannot support, so you can read a citation and know what it is actually claiming.
7 min read · updated 2026-07-23
The short version
Ask three questions of any citation: what population did it observe, was it the same molecule that is being sold, and is the outcome quoted the outcome the paper measured. In this category all three regularly fail, and the first one fails most often.
What each tier can support
- In vitroIn vitro — cells in a dish
Can establish: Show that a compound does something to cells under controlled conditions, and suggest a mechanism worth investigating.
Cannot establish: Establish anything about a living organism. A dish has no circulation, no liver, no immune system and no dose that has to survive getting there.
- AnimalAnimal — a living system, but not yours
Can establish: Show an effect in a whole organism, including absorption and distribution, and justify a human trial.
Cannot establish: Establish a human effect. Most drugs that work in animal models fail in human trials; the attrition between the two is the single largest filter in drug development.
- Case reportCase report — one person, described
Can establish: Record that something happened to someone, and flag a signal worth studying.
Cannot establish: Establish incidence, causation or typicality. With no control and a sample of one, a case report cannot separate the compound from coincidence.
- Human — observationalHuman observational — people, watched
Can establish: Show associations across real populations at a scale trials cannot reach.
Cannot establish: Establish causation. People who take a thing differ from people who do not in ways that are hard to fully adjust for.
- Human — controlled trialHuman controlled trial — the standard
Can establish: Establish an effect in the population studied, using the product studied, at the exposure studied.
Cannot establish: Transfer automatically to a different population, a different product, or material of unverified identity and purity.
- Meta-analysisMeta-analysis — trials, pooled
Can establish: Increase precision and detect effects too small for a single trial.
Cannot establish: Improve on the trials it pools. A meta-analysis inherits the products, populations and biases of its inputs.
Three failures to watch for in a citation
The species swap. A rodent result described in the language of human benefit. The tell is a sentence about outcomes with no subject — “supports tissue repair” rather than “reduced lesion area in rats”. If the sentence would be false with “in rats” appended, it is being used to imply something the paper does not say.
The molecule swap. Research about a related but different compound. The clearest example in this index is TB-500: nearly all cited work concerns thymosin beta-4, the full parent protein, while the product sold is a fragment. Evidence about one is not evidence about the other.
The product swap. Trial evidence for an approved medicine cited on a page selling research-grade powder. The trials used a manufactured product of verified identity and purity. A research vial is established only by its certificate of analysis — which is why reading that certificate properly matters more than the citation does.
How this index applies it
Every study record carries its evidence type, and every compound digest has a required section stating what is not established. Of the 28 digests published, 1 rest entirely on preclinical work — meaning no study indexed for those compounds observed a human being.
This page describes how to read published research. It is not medical advice, contains no dosing information or protocols, and does not suggest that any compound is suitable for any purpose.