Skip to content
Filia
Menu
Back

Treatment science

What the evidence actually says about the medicines we offer.

Each medicine we offer, graded on the trials behind it rather than on how it is marketed — including the ones that grade badly. Where a blood test decides whether a treatment is worth anything to you, it says so and links to it.

How we weigh evidence

Two things are recorded for every claim, because one without the other misleads. What kind of study it was, weakest to strongest — and separately what it found, because “a trial exists” is not the same as “a trial found something”.

  1. 01 In silico. A computer model suggests a pathway. Nothing has been tested.
  2. 02 In vitro. Cells in a dish. Shows activity is possible, not that it happens in a person.
  3. 03 In vivo. A whole living organism, usually an animal, before any human data.
  4. 04 Ex vivo. Living human tissue outside the body. Human material, but not a human outcome.
  5. 05 Observational. People followed or compared without being assigned. Shows association, not cause.
  6. 06 Product analysis. What is actually in the bottle, measured. Says nothing about whether the molecule works — only whether you are getting it.
  7. 07 Open-label. Human results where everyone knew what they were taking. Real, but flattered by expectation.
  8. 08 Randomised trial. Assigned by chance, usually blinded, against a placebo or a comparator.
  9. 09 Systematic review. Every trial that meets the criteria, pooled. Settles more than any single trial.

Positive. Found a benefit.

Mixed. Benefit on some measures and not others, or trials that disagree.

No effect. Looked properly and found nothing. A real result, not a missing one.

Negative. Did worse than the comparator, or caused harm.

How each treatment is scored

Two numbers, not a letter. A single grade cannot tell we do not know apart from we know it does not work, and those are opposite situations. Saw palmetto has been through a 27-trial review that found nothing — excellent evidence, and the answer is no. An untested peptide has no evidence at all. A letter puts them in the same bucket.

So certainty runs 1 to 10 for how settled the question is, and the finding says which way it settled. Both are computed from the claims underneath; neither is typed in by hand.

Established. Good human evidence that it works for this.

Promising. Human evidence leans positive, and it is not settled.

Unsettled. Real evidence on both sides, or it depends which outcome you mean.

Untested. No convincing human evidence for this use, in either direction.

Disproven. Looked at properly and it does not work. That is knowledge, not absence of it.

The index

3 treatments, every one written up

Research evidence score10/10EstablishedSemaglutideFDA-approved (as studied) · Weight management in adults with obesity, or overweight with a weight-related condition

Large, well-replicated weight loss in placebo-controlled trials — less than tirzepatide head to head — and the evidence is about weight rather than about health outcomes in women specifically.

Compounded medicines are not FDA approved. This score reflects the cited research on the studied medicine. It does not rate the compounded preparation shown on this site.

See the evidence →
Research evidence score10/10EstablishedTirzepatideFDA-approved (as studied) · Weight management in adults with obesity, or overweight with a weight-related condition

The larger effect of the two — shown head to head in a randomised trial — with more serious adverse events in the pooled comparisons and the same catch about stopping.

Compounded medicines are not FDA approved. This score reflects the cited research on the studied medicine. It does not rate the compounded preparation shown on this site.

See the evidence →

A treatment with no grade has not been written up yet. We would rather show you an empty row than a confident-looking summary nobody has checked — the pages that do this badly are the ones that fill every row.