Each indication-specific claim receives both a development stage and an evidence-confidence grade. The grade applies to a precise claim, not to the molecule as a whole.
- A: established for a labeled use. Typical support: Current approval plus adequate controlled trials and post-market context
- B: moderate human evidence. Typical support: Multiple controlled studies or a strong pivotal study, but no current approval for the claim
- C: preliminary human evidence. Typical support: Small, uncontrolled, surrogate-endpoint, or early-phase studies
- D: preclinical only. Typical support: In vitro or animal evidence with no adequate human efficacy evidence
- E: anecdotal or marketing. Typical support: Testimonials, extrapolation, or vendor claims without adequate scientific support
X is a separate branch: adequate negative evidence, a failed program, or evidence inconsistent with the studied claim or material.
Grading ladder
| Grade | Meaning | Typical support |
|---|---|---|
| A | Established for a specific labeled use | Current approval plus adequate controlled trials and post-market context |
| B | Moderate human evidence | Multiple controlled studies or a strong pivotal study, but no current approval for the claim |
| C | Preliminary human evidence | Small, uncontrolled, surrogate-endpoint, or early-phase studies |
| D | Evidence from in vitro or animal studies with no adequate human efficacy evidence — the atlas Grade D lane. Definition source: Evidence grading methodology · Glossary only | In vitro or animal evidence with no adequate human efficacy evidence |
| E | Anecdotal/marketing claim | Testimonials, extrapolation, or vendor claims without adequate scientific support |
| X | Evidence contradicts or does not support the claim | Adequate negative evidence, failed program, or claim inconsistent with the studied material |
- AGrade A: Established for a specific labeled use
- BGrade B: Moderate human evidence
- CGrade C: Preliminary human evidence
- DGrade D: Preclinical only
- EGrade E: Anecdotal/marketing claim
- XGrade X: Evidence contradicts or does not support the claim
Approval status and evidence grade answer different questions. An approved drug may have grade A evidence for one indication and grade D or E evidence for an off-label claim. Grade X: Evidence contradicts or does not support the claim — adequate negative evidence, failed program, or claim inconsistent with the studied material. Definition source: Evidence grading methodology · Glossary is a separate contradictory or non-supportive outcome, not a weaker position below E.
For applied examples, compare the claim rows for semaglutide, retatrutide, and BPC-157. The companion explainers Evidence Grades A to E Explained and How to Read a Peptide Clinical Trial show how the same vocabulary is used across the atlas.
Study-level fields
Evidence tables record the exact claim, population, comparator, sample size, route, duration, endpoint, result, important limitations, and identifier. The six assessment inputs below are kept visible before studies are synthesized at the claim level.
| Field | Question it answers | Common overreach |
|---|---|---|
| Exact claim | What effect, indication, and material was actually assessed? | Assigning one grade to the molecule as a whole |
| Population/comparator | In whom, and against what control or alternative, was the claim tested? | Generalizing beyond the enrolled population or treating an uncontrolled result as comparative evidence |
| Sample size | How much human information contributes to the result? | Treating a small study as conclusive or ignoring small-study bias |
| Route/duration | How was the studied material used, and for how long was follow-up observed? | Transferring findings across routes or using short follow-up for chronic benefit or safety claims |
| Endpoint/result | What was measured, what happened, and was the endpoint clinically relevant? | Treating statistical significance as clinical importance or silently converting a biomarker change into patient benefit |
| Limitations | Which design, reporting, product, or program constraints qualify the inference? | Omitting bias, non-equivalent products, termination, retraction, rejection, or withdrawal |
Statistical significance is not treated as clinical importance, and biomarker changes are not silently converted into patient benefit.
Exact claim, population and comparator, sample size, route and duration, endpoint and result, and limitations feed the study-level assessment, then the claim-level synthesis. Current product authorization supports A only for the exact labeled claim. A biomarker result is not patient benefit unless the endpoint supports that inference.
Common bias checks
randomization, concealment, blinding, attrition, selective reporting;
multiplicity and post-hoc analyses;
small-study and sponsor bias;
short follow-up for chronic-benefit or safety claims;
non-equivalence of the tested pharmaceutical product and a marketed "research" vial;
retractions, expressions of concern, trial termination, and regulatory rejection or withdrawal.
Three-question summary
What exact claim? Keep the indication, material, population, and authorization scope specific.
What human design and endpoint? Identify the comparator and whether the endpoint supports the claimed benefit.
What limitations and status? Preserve uncertainty, bias checks, product equivalence, and current authorization as separate considerations.