Scientific source policy and evidence hierarchy

Our source hierarchy

Medical statements on this site are supported by primary or authoritative sources, in this order of preference:

1. FDA drug labels, safety communications, databases and official guidance
2. NIH and PubMed-indexed original research
3. ClinicalTrials.gov registrations
4. Peer-reviewed systematic reviews and meta-analyses
5. Major professional medical guidelines
6. Government sources: CDC, CMS, the NPI registry, state pharmacy and medical boards
7. A provider's own pages — for that provider's prices and service claims only

What we will not treat as evidence

• Reddit and patient forums — never as evidence of price, safety, efficacy or legitimacy. They may be described as anecdotal consumer sentiment, and labelled as such.
• Affiliate comparison sites — never as proof of a medical claim, and never sufficient to mark a price Verified. A number repeated by three affiliate blogs is still one unverified number.
• Animal research — never as proof of a human clinical effect.
• Press releases — never as the sole evidence for a scientific conclusion.
• A single study — never as support for a universal claim.
• Provider marketing — never as independent proof of its own quality.

How far a result actually stretches

This is where most misuse happens, and it is not misquotation — it is extrapolation. A real result gets stretched past the population, the dose, the duration or the dosage form that was actually tested.

Every efficacy figure in the GLP-1 field comes from an FDA-approved subcutaneous injection. SURMOUNT-1 tested 5-15mg injections. STEP 1 tested a 2.4mg injection. Neither tested a compounded preparation, a 1mg microdose, or an orally disintegrating tablet. When those trial numbers appear on a page selling an ODT, evidence has been transferred across a dosage form without justification. We do not do that, and every study page we publish carries a 'what this does not prove' section for exactly this reason.

We do not invent citations

We do not invent a study, an author, a DOI, a PMID, a trial registration number or a result. Where we do not have a source, we say we do not have a source. Where a price is unverified, we label it unverified rather than publishing a plausible-looking number.

How this works in practice

A policy that is not operationalised is decoration. Here is what ours actually changes about the pages you read.

Every price carries a status. Verified means we hold a dated capture of the provider's own page. Reported — pending verification means a provider or third party reports it and we have not captured it ourselves. Evaluation in progress means we are not asserting it. We do not upgrade a price to Verified because a comparison site published it — sites in this category contradict each other routinely, and a number repeated by three affiliate blogs is still one unverified number.

Every medical claim traces to a primary source. FDA labels and guidance for regulatory status; PubMed-indexed randomised trials for efficacy; ClinicalTrials.gov for trial design. Reddit and patient forums are never used as evidence of price, safety, efficacy or legitimacy — they may be described as anecdotal sentiment, labelled as such. Animal research is never presented as proof of a human clinical effect.

Every ranking shows its arithmetic. Where a provider we have a commercial relationship with ranks well, the calculation that produced that result is printed on the page. If the arithmetic is wrong, you can see that it is wrong, and tell us.

Commercial relationships and what they do not buy

The publisher and certain principals have financial relationships with some of the telehealth providers listed on this site. That is how this publication is funded, and we state it in the footer of every page rather than burying it.

What compensation does not do: it does not change a score, a rank, an inclusion decision, or a negative finding. Providers cannot pay for placement, cannot suppress an accurate criticism, and cannot review their own page before publication. Where a commercially-related provider loses a category, we say so — a comparison in which one provider wins everything is an advertisement, and the fastest way to tell the difference is to look for the losses.

Corrections

We publish prices in a market that changes them frequently, and we will get things wrong. When we do, we correct the page, date the correction, and say what changed — we do not quietly edit a number and pretend it was always right. Both readers and providers can submit corrections with evidence, through the same process and to the same standard.

Our own record so far includes removing a set of provider prices we had sourced from a third-party comparison site and could not substantiate, and correcting brand-pricing figures that had gone stale after a manufacturer price cut. Both corrections made the site less flattering to conclusions we had already published. That is the point.