lifescore Edge compounded GLP-1 evidence audit

An audit of 6 editorially selected publications bearing on one bounded question: do compounded GLP-1 products have the same human evidence as the approved ones? This is not a systematic review and not a literature census. Each of the 6 audited sources is coded against five explicit gates, so a reader can recompute the finding rather than accept it. Of those sources, 2 administered a compounded product to a person, 0 tested equivalence against the approved medicine, and 1 passed the independence gate outright. On that gate the six sources fall out as 3 authored, funded or staffed by a party selling one side of the comparison, 2 disclosing ties that place no product on either side, and 1 industry-free. The administered studies cover 95 participants in total. Selection. The query compounded semaglutide OR compounded tirzepatide OR compounded GLP-1 was run against PubMed via the NCBI E-utilities API and Europe PMC on 2026-08-18 and returned 106 records. The inclusion and exclusion rules, the reading depth and the full list of limits are in PROTOCOL.md inside this package. Gate dictionary compoundedArticle — Was the article actually studied a compounded preparation rather than the approved medicine? pass = a compounded product was the test article; partial = compounded products are involved but their identity rests on self-report or is mixed with other classes; stop = no compounded product was examined. humanAdministration — Did this study give the product to a person and observe what happened? pass = administered and observed; partial = human exposure is reported but the study did not administer; stop = no administration. comparator — Was there a control group or an originator comparator? pass = present; partial = a reference standard or a non-compounded reporting group, not a control arm; stop = none. equivalenceTested — Does the study test whether a compounded product performs the same as the approved one? pass = equivalence or bioequivalence tested; partial = composition or signal compared without a clinical equivalence claim; stop = not tested. independentOfStake — Is the work free of a party with a commercial interest in the answer? pass = no such party; partial = disclosed ties that do not place a product on either side; stop = authored, funded or staffed by a party selling one side of the comparison. Boundary. Editorial evidence audit of published sources. Not medical, dosing, sourcing or purchase advice, not a safety verdict on any product, and not proof that an untested claim is false. What this dataset cannot show. It does not reproduce the 106-record screen that would establish how few studies administered a compounded product. The screened record list was not captured, so the audit cannot demonstrate its own denominator. Do not present it as the screen, and do not read the six rows as the complete literature. Independence of the auditor. A set that codes other people's interests has to declare its own. Edge does not currently carry advertising, sponsorships, affiliate links or paid placements. The people and organizations covered do not approve articles before publication. None of the audited parties commissioned, funded, reviewed or saw this audit before publication. lifescore sells no GLP-1 product, compounded or approved, and holds no position on either side of the comparison. Limits of that declaration. This audit was not externally peer reviewed. The auditor is the publication itself; there is no institutional affiliation and no third-party editorial board. Authorship is organizational rather than individual, so the coding cannot be attributed to a named expert with a public track record. Independence from a commercial interest is not a claim of expertise, and it does not make a coding decision correct. Licence scope. The selection of sources, the gate scheme, the coding decisions, and the original summary text written by lifescore Edge. It does not cover the cited publications themselves, which remain with their own rights holders and are referenced by DOI or PMID rather than reproduced; and the underlying facts, which are not subject to copyright. Each record was read from the indexed PubMed record or the publisher's own article page, including conflict-of-interest and funding statements, rather than from a search snippet. Published by lifescore Edge. Editorial standards, source hierarchy, correction policy and the independence boundary: https://lifesco.re/edge/methodology/. Machine-readable copies of this dataset are also served at JSON and CSV.