Levels of evidence A/B/C
How sources are selected, what levels A, B and C mean, and why a dose is not invented when no studies exist for a goal. This is the app's Methodology section retold as a plain page.
What the levels mean
A Proven
Systematic reviews, meta-analyses and positions of professional societies. The effect reproduces in independent studies — the strongest support the base has.
B Likely
Individual good randomised controlled trials or reviews. The effect exists but is modest, inconsistent or depends on how the product is used.
C Limited data
Preliminary and mechanistic data, single observations and case reports. These are not a basis for firm conclusions, and the review page says so.
A level is assigned by the type of research, not by how popular an item is: the reviews are collected in supplements reference, and the terms themselves are explained in glossary.
Priority of sources
- Systematic reviews and meta-analyses
- Randomized controlled trials
- Cohort and observational studies
- Official positions and consensus statements: ISSN, ACSM, IOC, ESPEN and the Endocrine Society
Blogs, forums, shops and commercial sites are not used as sources: they can only point to the original publication. Publication metadata is never typed in by hand — title, journal, year and authors come from PubMed/MEDLINE, and a verification pass re-checks every record and stops the build if an article is not found. The dataset itself can be downloaded on the open data.
Why a dose is not invented
Every item has its own entry for every claimed goal. Where no good studies exist for a "supplement × goal" combination, an honest note stands in place of a number rather than an average from a study about another task. A confirmed effect is not carried over to a neighbouring goal: an item working for one task says nothing about another.
All such cases are collected on the «Where data is missing» — that is not a gap in the reference but its honest part. An empty cell means exactly one thing: no good studies exist for that combination.
A level is the strength of evidence, not benefit
Levels A, B and C describe how well an effect is confirmed, not how well an item suits a particular person. A higher level means the effect is better supported by research and says nothing about your goal, health, medications or limits. That is why the reference offers no "what to choose" advice: it compares characteristics and data, and the decision stays with the reader and their doctor.
Update
A new publication goes to review first and only then into the base: automatic publication of unchecked data is excluded. What changed and when is visible in the «Change log».