Method · Evidence grades
How we grade evidence.
One letter per dossier. It answers one question: how much good human evidence exists for the effects people actually care about? Here is the rubric, how we apply it, and what a grade does not mean.
The atlas today
- A5
- B3
- C9
- D6
23 dossiers · reviewed 2026-09-26. Dashed bars have no glow on purpose.
The rubric
Four grades, published word for word
This is the exact text every grade in the atlas is held to. If the wording ever changes, the change will be noted here.
- A
Strong human evidence5 in atlas
Multiple adequate human RCTs and/or FDA approval for a defined indication.
- B
Human RCTs, narrow3 in atlas
Human RCTs exist but narrow, early, or not approved for the popular use.
- C
Thin human data9 in atlas
Human data thin (pilots, uncontrolled, or none completed); strong animal + large anecdotal volume.
- D
Little useful signal6 in atlas
Mechanism speculation, marketing, or almost no useful human signal.
Explore
Every peptide, filed by grade
Pick a grade. The list shows each dossier carrying it and the one-line evidence note that explains why.
Grade A · Strong human evidence
Multiple adequate human RCTs and/or FDA approval for a defined indication.
- SemaglutideMetabolic · FDA-approvedGraded A on the strength of multiple large RCTs and FDA approval for weight management, type 2 diabetes and cardiovascular risk reduction. The evidence belongs to the approved product, not to compounded or gray-market copies.
- TirzepatideMetabolic · FDA-approvedGraded A on the strength of large RCTs and FDA approval for three indications. The evidence belongs to the approved product, not to compounded or gray-market copies.
- RetatrutideMetabolic · Clinical-stageGraded A for weight-loss efficacy on the strength of large randomized trials. Not approved, and long-term safety is still being characterized.
- TesamorelinGH-axis · FDA-approvedGrade A for its approved indication. The popular off-label use (general fat loss, physique, anti-ageing) has no comparable trials.
- PT-141Sexual · FDA-approvedGrade A for the approved indication, on two Phase 3 RCTs. Effect sizes are modest, satisfying sexual events did not beat placebo, and there is no comparable evidence in men.
Evidence ladder
From a dish to a drug label
Study designs differ in how well they can separate a real effect from chance, bias and hope. Most compounds in the gray market stall on the lower rungs.
- 01Cell / labD
A molecule does something in a dish. Says little about a living body.
- 02AnimalC or D
Effects in a whole organism, usually rodents. Doses, metabolism and outcomes often fail to translate.
- 03Case reportsC
A clinician describes one or a few patients. Good for spotting signals and side effects. No comparison group.
- 04Uncontrolled pilotC
A small group is measured before and after. Cannot separate the drug from placebo, time or expectation.
- 05Randomized trialB
People are randomly assigned to drug or control, ideally blinded. The first rung that can show cause and effect.
- 06Multiple RCTs / approvalA
Several adequate trials agree, or a regulator has reviewed the full data for a defined use.
Method
How grades are assigned
- 01
Humans first
Randomized human trials and FDA labels carry the most weight. Animal and cell work explains mechanism, but on its own it cannot lift a compound above C.
- 02
The use matters
A grade reflects the best human evidence for the uses a dossier covers. When the popular use runs ahead of the data, the evidence note says so on the first screen. Tesamorelin is Grade A for HIV-associated lipodystrophy, not for general fat loss.
- 03
Anecdotes are counted, not scored
Large anecdotal volume is part of the C definition because it shows people are using something. It cannot tell an effect apart from expectation, regression to the mean or a mislabeled vial.
- 04
Design over headlines
We look at size, controls, blinding, endpoints and whether results were published in full. Topline results in a press release weigh less than a peer-reviewed paper.
- 05
Ties go down
When a compound sits between two grades, it gets the lower one until better data arrive.
- 06
Re-checked on every review
Grades are revisited on each review cycle and whenever a trial reads out, a label changes or FDA acts.
Limits
What a grade is not
Not a safety rating
An A-grade drug can carry serious risks. GLP-1 labels include a boxed warning for thyroid C-cell tumors. A D-grade compound is not proven dangerous. It is unproven, and that includes unproven safe.
Not a recommendation
Grade A means the evidence for a defined use is strong. It does not mean anyone should take it, and it says nothing about whether it suits a particular person.
Not permanent
Grades move when evidence moves. A trial readout can lift a compound, and a failed replication can sink one. Every page shows when it was last reviewed.
Dose labels
Every number says where it came from
Each dose figure in the atlas carries one of three labels. The label tells you the source and how much weight the number can bear. None of them is a dose for you.
- FDA labelMost weight
The dose printed on the FDA-approved prescribing information for the approved indication.
Reviewed by FDA for a specific indication and population. Printed on the prescribing information.
Example in the atlasSemaglutide · Wegovy injection label (weight management): 0.25 → 0.5 → 1 → 1.7 → 2.4 mg, once weekly; 4-week steps - Clinical trial rangeReal human data
Doses used in published or company-reported human trials. Trial doses are not approved doses.
Measured under controlled conditions in specific trial populations. Not an approved dose.
Example in the atlasTirzepatide · SURMOUNT-1 trial doses: 5, 10 or 15 mg, once weekly for 72 weeks - Commonly reported rangeLeast weight
Ranges repeated across research-community discussion. Not validated by trials. Not a recommendation.
Repeated across forums, clinic handouts and logs. Repetition is not validation.
Example in the atlasPinealon · Commonly reported injectable range: 1–2 mg, once daily for 10–20 days
Examples are pulled from the dossiers and shown with their labels to illustrate the system. They describe the published or discussed landscape and are not instructions.
Disagree with a grade?
Good. Bring a trial, a label or a registry entry, and the grade gets re-checked.