Five weighted dimensions feed a single 0-100 score. The score gets rounded, mapped to a tier, and shown next to claims that need a confidence tag. This page is the long-form spec. The narrative summary lives at /methodology.
| Dimension | Weight | High | Mid | Low |
|---|---|---|---|---|
| Trial strength | 35% | Phase-3 randomized controlled trial with active comparator and pre-registered endpoints, or a meta-analysis of multiple Phase-3 trials. | Phase-2 trial, or Phase-3 with limitations like open-label design or single-center recruiting. | Preclinical / animal data, case series, or anecdote. |
| Sample size | 20% | n ≥ 10,000 (for cardiovascular outcomes) or n ≥ 1,500 for symptom endpoints. | n = 100 to 1,000. | n < 100, or pilot work. |
| Mechanism | 15% | Receptor-level pharmacology established and reproducibly demonstrated in independent labs. | Plausible mechanism, with one or two supporting studies. | Hypothetical or borrowed from a different drug class. |
| Reproducibility | 15% | Two or more independent trials, run by different sponsors, report consistent results. | One large trial plus supportive observational work. | Single trial, no replication yet. |
| Real-world fit | 15% | Trial population reflects our reader: women, mixed ages, real-world comorbidities and adherence. | Trial enrolled people similar to our reader on most dimensions. | Lab-only data, healthy young men only, or non-generalizable population. |
We treat this as settled enough to act on. Independent replication exists, mechanism is clear, the trial population fits.
Solid evidence, but one of the five dimensions is short of perfect, usually replication or real-world fit.
Worth reporting, worth acting on with caveats. We name the limitation in the body of the post.
Early signal. We report it because readers ask, and we flag exactly what is missing.
Mechanism is plausible or anecdote is loud, but the trial literature is too thin to be confident. We say so in plain language.
Posts flagged with the medical-disclaimer requirement get an extra evidence check before publish: every dose, indication, and rate is verified against the current FDA prescribing information and the published trials, and the check is dated on the post. We do not claim review by a named outside clinician. If a credentialed reviewer joins the team, their name and credentials will appear on the post.