This is the module that makes you "better than a researcher who only knows their niche." A compound's star rating is only as trustworthy as your ability to check it yourself.

The hierarchy of evidence (weakest → strongest)

1. In vitro ("in glass") — cells in a dish. Shows a mechanism is possible, not that it works in a person. Most "this molecule kills cancer cells" headlines are here.
2. Animal (in vivo) — mice/rats. Establishes it works in a living body — but most animal findings don't translate to humans. This wiki caps animal-only compounds at ⭐⭐ for exactly this reason. BPC-157's problem is that it's stuck here.
3. Observational human studies — cohort/epidemiology. Finds associations (spermidine-eaters live longer) but cannot prove causation (confounding — maybe spermidine-eaters also exercise more).
4. Randomized controlled trials (RCTs) — randomly assign drug vs placebo. The gold standard, because randomization cancels out confounders. Double-blind = neither patient nor researcher knows who got what.
5. Meta-analysis — pools many RCTs for the most reliable estimate. The top of the pyramid.

The numbers you must be able to read

• p-value — the chance the result was a fluke. p<0.05 is the convention, but "statistically significant" ≠ "meaningful."
• Effect size — how big the effect is. A supplement can be statistically significant and practically useless (2% change). Always ask "significant and large?"
• Confidence interval — the plausible range of the true effect. Wide interval = uncertain (small study).
• Number needed to treat (NNT) — how many people must take it for one to benefit. A powerful reframing of "does this matter."
• Relative vs absolute risk — "cuts risk 50%" (relative) can mean 2%→1% (absolute). Marketers quote relative; think in absolute.

The traps

• Publication bias — positive results get published, null results get buried, so the literature looks rosier than reality.
• Funding bias — industry-funded supplement trials skew positive.
• Surrogate endpoints — a drug that improves a marker (LDL, testosterone) hasn't proven it improves an outcome (heart attacks, lifespan).
• N and duration — a 12-person, 4-week study proves little.

The regulatory maps (so "approved" means something)

• FDA drug approval: Preclinical → Phase 1 (safety, small) → Phase 2 (does it work?) → Phase 3 (large confirmatory) → approval. "Phase 3" in an entry means late-stage, strong-but-unfinished (e.g., retatrutide).
• Supplement vs drug: supplements are regulated as food — not tested for efficacy before sale. "OTC" ≠ "proven."
• Off-label: an approved drug used for an unapproved purpose (rapamycin for longevity) — legal for doctors, but the evidence for that use may be thin.
• WADA/USADA: the sport-doping banned lists — most SARMs, peptides, EPO, stimulants. Relevant if you compete.
• The 2026 peptide situation: removal from FDA Category 2 (April 2026) ≠ approval; it only lifted a prohibition. Watch the July 2026 PCAC compounding decision. A literate reader holds "reclassified" and "approved" as different states.