How we find, verify and grade research

How records are found

The corpus comes from one PubMed query, run through the E-utilities API: ipamorelin OR "NNC 26-0161", which PubMed translates to "ipamorelin"[Supplementary Concept] OR "ipamorelin"[All Fields] OR ("nnc"[All Fields] AND "26 0161"[All Fields]). At retrieval it returned 55 records [PubMed search].

Every record the search returns is indexed. Nothing is dropped for being inconvenient, off-topic or unflattering. Where a record turns out to be marginal — an analytical paper that does not name ipamorelin in its title, or a registry entry that is not really about the compound — it is kept and labelled as such, so the search remains reproducible. One record has no abstract indexed in PubMed at all, and its entry says so rather than inventing content [Cox 2015].

Trials come from the ClinicalTrials.gov v2 API [ClinicalTrials.gov], chemistry from PubChem PUG REST [PubChem], and regulatory records from FDA pages and documents read directly [FDA Category 2].

How records are verified

Bibliographic data is never typed by hand. Title, journal, year, authors, DOI and publication types are pulled from the PubMed record itself and merged with editorial annotations at build time. The build fails if any indexed record lacks an annotation, or if any annotation names a record the search did not return — so the database cannot contain an invented reference, and cannot silently drop a real one.

A separate verification script checks that every cited PMID resolves on PubMed with a matching title and year, that every DOI resolves, and that every other cited URL returns successfully. Any page citing a PMID that is not in the database fails the build rather than rendering a broken citation.

How records are classified

Three classifications are applied to every record, and all three are filters in the explorer.

Research model

• Human study
• Animal study (in vivo)
• Ex vivo (excised tissue; no animal received the substance)
• In vitro (cells or receptor preparations)
• Analytical chemistry
• Review or commentary

Substance actually tested

This distinction is specific to ipamorelin and matters more here than on most compounds, because much of the literature a search returns is about analogues built from the ipamorelin scaffold rather than ipamorelin itself.

• Ipamorelin itself
• An analogue built from the ipamorelin scaffold
• Ipamorelin among several compounds tested
• A modified form found in seized or market product

Of 42 primary records, 12 have ipamorelin itself as their subject. A finding for NN703 or for Gly-ipamorelin is never presented as a finding for ipamorelin.

Evidence grade

• A: randomised controlled human trial
• B: other human study (no control group, or not randomised)
• C: animal study (in vivo)
• D: cell, tissue, receptor or analytical study
• R: review or commentary (context, not evidence)

Grades are assigned against these criteria, considered together:

• study design, and whether there was randomisation and a control group;
• relevance of the model to the claim being evaluated;
• population or species, and sample size where reported;
• quality of controls, including whether comparisons were against untreated or drug-treated animals;
• relevance of the outcome measured to any clinical question;
• whether results have been replicated independently;
• whether safety findings were reported at all.

What is kept separate

• Reviews from primary research. 13 of the 55 records are reviews or commentary. They are graded R, never counted in primary-study totals, and never cited as evidence for a claim.
• Species. A finding in rats, mice, pigs, ferrets, fish or cells is labelled with its species everywhere it appears.
• Route. Intravenous, subcutaneous, intranasal, oral and culture exposures are never pooled.
• Endocrine findings from clinical outcomes. A change in a hormone level is not a patient benefit, and the two are separated on every page that touches both.
• The compound from the product. Analytical findings about what is in market material are classified separately from research on pharmaceutical-grade ipamorelin.

How claims are written

Each record carries a line describing what was measured, written under fixed rules:

• the study is the subject of the sentence, not the compound;
• the model is named;
• results are reported as the authors reported them, including null and negative results;
• no line says what the compound “does” in a body, or what it is “for”.

Editorial workflow

Every record moves through five states: Draft → Source Verified → Scientific Review → Approved → Published. Records reach Source Verified when the automated checks pass against the primary record. Scientific Review is a human judgement on interpretation and grading, and is not something this site's automated pipeline can grant itself. Approval states are stored in a git-tracked file, and a production build can be configured to refuse to render any record that is not approved. The rules each page follows are on the editorial policy page.

What this site will not publish

• Dosing, administration, reconstitution or self-use guidance of any kind.
• Combination, stacking or cycling protocols.
• Pooled effect estimates across heterogeneous studies, or any combined efficacy score.
• A claim that a mechanism demonstrates a clinical benefit.
• A fabricated citation, study count, or regulatory fact. Where a source could not be read, the site says so.

The corpus was last verified on 2026-11-11 and the regulatory records on 2026-11-11. Errors can be reported as described on the about page.