On this page
Every clinical page on this site carries an evidence label. Those labels are not decoration and they are not scores we invented — each one describes what kind of human research exists, and each is meant to be checkable against the studies cited on that page.
Why a label at all
“Studies show” is the least useful phrase in medical writing. It flattens a laboratory experiment on cultured cells and a randomised trial in nine hundred patients into the same claim.
Those are not the same claim. One is a reason to do further research; the other is a reason to change what you do to patients.
So every condition and treatment page here states which it has, at the top, before anything else.
The five labels
Used consistently across the site.
- Established clinical use — accepted medical practice, supported by regulatory approval or equivalent professional consensus.
- Higher-level human evidence — multiple controlled human trials, usually with systematic review or meta-analysis, though limitations remain.
- Early human evidence — human studies exist but are preliminary, small, heterogeneous or inconclusive.
- Preclinical evidence — mostly laboratory or animal work, with little or no human data.
- Insufficient evidence — reliable evidence is inadequate to support any meaningful conclusion.
A treatment being offered commercially says nothing about which label it earns.
The hierarchy behind the labels
Not every study carries equal weight, and the ordering is not arbitrary.
At the top sit systematic reviews and meta-analyses, which pool multiple studies and assess their quality. Below those, randomised controlled trials, where allocation is random and ideally blinded. Below those, controlled studies without randomisation, then prospective cohorts, then case series, then individual case reports.
Underneath all of it sit animal studies and laboratory work, which establish that something is worth investigating rather than that it works.
Where this matters practically is that the same treatment can sit at two different levels for two different conditions. MSCs in knee osteoarthritis have randomised trials and meta-analyses behind them; the same cells for a neurological condition may have almost nothing in humans. One label cannot cover both, which is why the grade is set per page rather than per product.
Why a case series is not a trial
A case series describes what happened to a group of patients who had a treatment. It has no comparison group, so it cannot tell you what would have happened without it.
That matters enormously in conditions that fluctuate. Osteoarthritis pain varies week to week; people present when it is bad, and much of the improvement that follows would have happened anyway.
Placebo response in injection studies is substantial and well described. Any result not measured against a sham comparison has to be read with that in mind.
None of which makes case series worthless. They are often the first signal that something deserves a proper trial, and in rare conditions they may be all that will ever exist. The error is treating them as though a trial had been done.
What a trial measured, and for how long
Two questions worth asking of every study cited anywhere, including here.
What was the endpoint? A pain score is not cartilage thickness. A biomarker is not a symptom. A change in an imaging measure is not a change in how a knee feels.
How long was follow-up? Six months tells you little about a condition that runs for thirty years. Most of the evidence discussed on this site has follow-up measured in months.
A third question is worth adding: how many people dropped out, and were they counted? Studies that report only the patients who completed treatment can look considerably better than studies that count everyone who started.
Mechanism is not evidence
A plausible biological explanation for why something ought to work is a reason to test it, not a result.
Medicine is full of interventions with excellent mechanistic rationale that failed when tested properly. The rationale was not wrong about the biology; it was incomplete about the body. Cells behave differently in a dish than in a joint, and differently in a mouse joint than in a sixty-year-old human one.
Where this site describes a proposed mechanism — as on MSC immunomodulation — it is labelled as a proposal, and the clinical evidence is reported separately.
How these labels are assigned here
Each label is set in the site’s own data file, and the page cannot contradict it. If a page claims a higher grade than its data allows, the build fails and the page does not publish. That is a mechanical check rather than an editorial intention, which is the point — intentions drift over a few dozen pages and mechanical checks do not.
Citations are checked the same way. Every study referenced on this site is resolved against PubMed during the build, and a reference that cannot be resolved is removed rather than paraphrased.
That mechanism exists because it caught real errors during construction — identifiers that looked correct and pointed at unrelated papers.
What the labels deliberately do not do
They do not tell you whether a treatment is right for you. Evidence describes populations; you are one person with a specific history.
They do not rank clinics, and they do not express an opinion about price. A clinic offering a treatment graded here as early evidence is not necessarily doing anything wrong — a great deal of legitimate medicine sits at that level. What matters is whether the clinic describes it that way to you, or describes it as settled.
And they do not stay fixed. Evidence moves, and a label here reflects the literature at the date shown on the page — see the limits of current evidence.
They also carry no opinion about legality. A treatment can be lawfully sold in one country and unavailable in another while the underlying evidence is identical in both — see regulation in Thailand.
Frequently asked questions
What does “early human evidence” actually mean?
Human studies exist, but they are preliminary — often small, varied in method, or without a sham comparison. It is real evidence, and it is not enough to call something established.
Why is a meta-analysis ranked above a trial?
Because it pools several trials and assesses their quality, which reduces the influence of any single small or unusual study.
Does a treatment being legal mean it is proven?
No. Legality and evidence are separate questions decided by different bodies for different reasons.
Why do you keep mentioning follow-up length?
Because most evidence here runs for months against conditions that run for decades. Durability is genuinely unknown for much of it.
What is wrong with “studies show”?
It hides the difference between a dish of cultured cells and a randomised trial. Both are studies; only one should change a treatment decision.
Can I check your citations?
Yes, and you should. Every reference links to its PubMed record, and each was resolved against PubMed during the build rather than typed from memory.
Requesting a medical evaluation
Nothing on this page establishes whether any treatment is appropriate for you. That needs your history, your imaging and your current medications read by a clinician. If you would like that review, send your records and we will tell you honestly whether there is anything worth discussing — including when the answer is no.
This page is general information, not medical advice, and does not create a doctor–patient relationship. Regenerative treatments discussed here are in most cases investigational. Discuss any treatment with a clinician who knows your history.