How we flag evidence.

A study that pools data from multiple independent RCTs Randomized controlled trial: people are assigned to the treatment or a comparison by chance, so the groups differ only in what they received. It is the design that can show cause and effect. to calculate an aggregate effect. The strongest form of evidence because it reduces the influence of any single trial's quirks. When a meta-analysis of 7+ RCTs agrees, the signal is robust.

Grade impact: supports Grade A if the endpoint The outcome a trial measures to decide whether the treatment worked, such as heart attacks, blood pressure or a symptom score. is a hard clinical outcome, an objectively measured function, or a validated scale scored by a clinician blinded to treatment. Required for our highest confidence ratings.

Examples: Omega-3 cardiovascular meta-analysis, Magnesium 34-RCT BP meta-analysis, Melatonin 19-RCT meta-analysis

Participants randomly assigned to treatment or placebo An inactive substitute made to look like the real treatment, so participants cannot tell which one they received. with controlled conditions. The gold standard for establishing causation. EBL requires at least one RCT for any grade above D. The quality, size, and blinding of the RCT determine whether it supports a B or an A.

Grade impact: minimum requirement for Grade B or above. Single RCT with biomarker A measurable sign in the body, such as a blood test value, used to track a condition or the effect of a treatment. endpoints = Grade B. Multiple RCTs on a hard clinical outcome, an objectively measured function, or a validated scale scored by a clinician blinded to treatment = Grade A.

Examples: REDUCE-IT (Omega-3, n=8,179), Q-SYMBIO (CoQ10, n=420), Yoshino 2021 (NMN, n=25)

Researchers observe existing behavior without intervention. Can show correlation but not causation. Large observational studies (thousands of participants over decades) generate hypotheses that RCTs then test. Observational data alone cannot support a Grade A on EBL, regardless of effect size How large a difference the treatment made, as distinct from whether the difference could be detected at all..

Grade impact: can support Grade B as secondary evidence alongside RCTs. Cannot be the sole basis for any grade above C. The Finnish sauna cohort (↓ CVD risk (observational)) is observational. This is why sauna is flagged, not graded at A.

Examples: Laukkanen 2015 (sauna, n=2,315; PMID PubMed ID: the permanent number the US National Library of Medicine gives a published paper. Each one links to that paper's record on PubMed. 25705824), Kiechl 2018 (spermidine dietary intake)

A preliminary study without blinding or placebo control, often with small sample sizes (n<30). These generate early signal but cannot establish efficacy. Participants and researchers both know who's getting treatment, which introduces bias. Most "promising" supplements are stuck at this level.

Grade impact: maximum Grade C. Cannot support B or A regardless of positive results. Upgrade requires a proper blinded, placebo-controlled RCT.

Examples: Arterosil (n=19 uncontrolled), Vascanox (n=12 open-label), Efrati HBOT telomere study (n=35)

Multiple independent research groups have confirmed the finding. This is the difference between "one lab found something" and "the scientific community agrees." Replication is what separates a promising lead from reliable evidence.

Grade impact: required for Grade A. A single unreplicated RCT, no matter how large, stays at Grade B until an independent group confirms the finding.

Only one research group has published on this. The result may be real, but until an independent team confirms it, there's a meaningful chance the finding is specific to that lab's conditions, population, or methodology. Over 50% of unreplicated findings fail to hold up in subsequent trials.

Grade impact: under Rule 03 a single positive trial without independent replication is a Grade C, regardless of trial quality. Most Grade C supplements carry this flag.

The trial measured outcomes that directly matter: did people live longer, have fewer heart attacks, break fewer bones, or develop less cancer? Hard endpoints are unambiguous, you either died or you didn't. These are the same endpoints used in drug approval trials.

Grade impact: a hard endpoint plus replication reaches Grade A. So does replication on an objectively measured function, or on a validated scale scored by a clinician blinded to treatment: the other two qualifying routes. This is the highest bar in medicine and only a handful of supplements clear it.

Examples: REDUCE-IT (Omega-3, −25% MACE), VITAL (Vitamin D3, −25% cancer mortality in follow-up), GISSI-P (Omega-3 post-MI mortality)

The trial measured validated biomarkers that correlate with health outcomes but aren't the outcomes themselves. Lowering LDL-C is associated with fewer heart attacks, but "LDL went down" is not the same as "fewer heart attacks happened in the trial." Biomarker trials are valuable but inherently one step removed from proof.

Grade impact: biomarker endpoints support Grade B. Upgrade to A requires hard endpoint confirmation in a separate trial.

Examples: Berberine HbA1c reduction, CoQ10 FMD improvement, Ashwagandha cortisol reduction, NMN insulin sensitivity