The EBL Evidence Pyramid
Every supplement and intervention on this site is assigned a tier. The tier reflects the quality of human evidence — not the enthusiasm of researchers, not the plausibility of the mechanism, not the quality of the marketing. Human outcomes first. Everything else below.
Randomized controlled trials in humans measuring real clinical outcomes — all-cause mortality, cardiovascular events, fracture incidence, hospitalisation. These are the studies that matter. A supplement earns Grade A only when multiple independent RCTs show consistent benefit at these endpoints. Examples on EBL: EPA/DHA Omega-3 (REDUCE-IT), Creatine (500+ RCTs), Magnesium (34-RCT meta-analysis on BP), Vitamin D3 (VITAL).
Pooled analyses of multiple RCTs. Higher statistical power than any single trial, but effect sizes can be diluted by study heterogeneity or inflated by publication bias. A Cochrane review of high-quality RCTs sits near the top of Tier 2; a meta-analysis of three small poorly-designed trials sits near the bottom. We note the quality of the underlying studies.
Human RCTs that measure biomarkers (HbA1c, LDL-C, bone density, VO₂max) rather than clinical outcomes. This is where most supplement research lives. The fundamental problem: biomarker improvement does not reliably predict clinical benefit. The CAST trial gave antiarrhythmic drugs that normalized heart rhythm — and increased mortality. A supplement that lowers LDL-C or raises NAD+ is promising. It is not proven.
Mouse lifespan studies, rodent cognition, cell culture. Mechanistically interesting. Historically unreliable as predictors of human outcomes — approximately 90% of findings at this tier fail to replicate in human trials. We mention Tier 4 data when it's the best available and the mechanism is plausible, but we say clearly that it hasn't been tested in humans.
Pathway plausibility, conference presentations, case reports, theoretical extrapolation. Not evidence. We rarely cite Tier 5 as support for a recommendation. When we do, it's clearly labelled as speculative. The longevity field has a significant problem with Tier 5 claims presented with Tier 1 confidence — this pyramid exists precisely to counter that.