How our evidence grades work
We don't use an opaque “research score.” Every compound's grade is computed from published fields by the rubric below, and the derivation is shown on each profile. You can always see why a compound got the grade it did.
Strongest tier, requires at least one indexed study that included female subjects.
Solid human evidence, but female-specific outcomes not established. Approved compounds with no female studies are capped here.
Human evidence exists but is preliminary.
No human trial data. Any human dosing figure is extrapolated.
Absence of evidence, not evidence of absence.
How we count studies
The three numbers on each reviewed profile, PubMed records, female-tagged, and sex-analysis tagged, come from live queries against the NCBI PubMed database, so they are real and reproducible. Specifically:
- PubMed records, title/abstract matches for the compound name and its aliases.
- Female-tagged, of those, the count carrying PubMed's Female MeSH term (a standard index tag for studies involving female subjects).
- Sex-analysis tagged, the count carrying the Sex Factors MeSH term, a proxy for studies that analyzed results by sex.
These MeSH filters are automated proxies, not a manual read of every paper. They undercount in real cases, for example, a compound developed and approved specifically for women can still show zero “sex-analysis” tags if indexers didn't apply the term. That is exactly why the qualitative evidence-in-women block on each profile is authored by hand and does not simply mirror these counts. Each profile shows the date its counts were pulled.
The female-evidence modifier
Alongside the letter grade, every compound carries a female-evidence type: Direct (sex-stratified outcomes reported), Extrapolated(women were included but results weren't analyzed by sex), or None(no female subjects). A compound can hold a strong letter grade on male-dominated evidence while still showing “Female evidence: None”, and we show both, because for our readers the second number is often the one that matters.
Grades describe the state of the published literature, not our opinion of a compound. They are not medical advice and not a safety rating. “Unstudied” states mean no qualifying study was found.