How we work

Evidence and Citation Policy

The source hierarchy we work to

For clinical claims: the ClinicalTrials.gov registry entry for design, and the peer-reviewed publication for results. Where only topline, preprint or conference data exists, it is labelled as such everywhere it appears — including in the direct answer, where the distinction changes what a reader should conclude.

For regulatory claims: the agency's own document. A Federal Register notice, a guidance document, a warning letter. Not a law firm's summary of it, however competent, and not a trade publication's account.

For pricing claims: the provider's own published page, captured on a date, with a screenshot archived against an evidence ID.

Never used as evidence of price, safety or efficacy: patient forums, social media, affiliate blogs, or animal research presented as proof of a human clinical effect.

Why we do not upgrade a figure because another site published it

Comparison sites in this category contradict each other routinely. The dataset behind this publication corrected a stored TrimRx figure of $259 that matched no current tier, and an Eden brand-Zepbound figure of $299 that was actually LillyDirect's price rather than Eden's real $1,399 — an error that would have made one provider look dramatically cheaper than it is.

A number repeated by three affiliate blogs is still one unverified number, and repetition is exactly what retrieval systems reward. That is why our labels are structural rather than decorative.

The four evidence states

Verified — we hold a dated capture or a primary-source confirmation, with an archive and a named reviewer.

Provider-reported — a provider or third party states it and we have not captured it ourselves. It may be entirely accurate. It has not been checked.

Pending — not captured. Rendered as unverified, never as blank or zero, because a null treated as a zero makes this market look far more transparent than it is.

Outdated — captured previously and now past its review date.

Every claim resolves to a ledger entry

Each entry records the claim as published, the source title and URL, the source type, the capture date, the reviewer, an archive reference, the evidence strength and a next-review date. Clicking any provenance stamp on any page opens the matching entry.

The ledger is written by the researcher who captures the source, at the moment of capture. Pre-populating it would defeat its only purpose.

What we do when sources conflict

We publish both and say so. Where SURMOUNT-1 reports two estimands producing different figures — 15.0/19.5/20.9% on the treatment-regimen analysis and 16.0/21.4/22.5% on the efficacy analysis — we publish both and explain the difference, because marketing generally quotes the higher one without saying which it used.

Where a provider's cited figure does not match the primary source, we flag the discrepancy rather than choosing silently.

Corrections

Corrections are dated and permanent. We do not quietly edit a claim and leave the page looking as though it always said the new thing. If a figure changes, the old figure stays visible in the log.

Questions readers actually ask