Our Methodology

EvidenceCheck uses a systematic, transparent approach to evaluate health claims. We believe in showing our work so you can verify our conclusions.

Evidence Hierarchy

We prioritize evidence sources in this order (highest to lowest quality):

  1. Meta-analyses of randomized controlled trials (RCTs) — The gold standard, combining results from multiple rigorous studies
  2. Systematic reviews — Comprehensive, methodical reviews of all available evidence
  3. Individual RCTs — Controlled experiments with random group assignment
  4. Observational studies — Cohort, case-control, and cross-sectional studies
  5. Animal/in-vitro studies — Preliminary evidence that may not translate to humans

Confidence Scoring

Every claim receives a confidence score (0-100%) based on three factors:

1. Base Score (Evidence Quality)

  • Meta-analysis of RCTs exists → starts at 80%
  • Systematic review exists → starts at 70%
  • Multiple RCTs (3+) → starts at 60%
  • Single RCT → starts at 40%
  • Only observational studies → starts at 30%
  • Only animal/in-vitro → starts at 10%

2. Consistency Adjustment

  • All studies agree → +15%
  • Most agree (>70%) → +5%
  • Mixed results → -10%
  • Mostly contradictory → -20%

3. Recency Adjustment

  • Latest major study <2 years old → +5%
  • Latest major study 2-5 years old → 0%
  • Latest major study >5 years old → -5%

Confidence Levels

Score Level Meaning
75-100% ✅ Strong Evidence Consistent, high-quality evidence supports this claim
55-74% 🟡 Moderate Evidence Good evidence exists, but with some limitations or inconsistencies
35-54% 🟠 Weak Evidence Limited evidence; more research needed
15-34% ⚖️ Conflicting Evidence Studies significantly disagree with each other
0-14% ❌ No Evidence No reliable human studies found

Review Process

  1. Claim identification — We select claims based on search volume and community requests
  2. Literature search — We search PubMed, Cochrane Library, and Google Scholar for relevant meta-analyses and RCTs
  3. Evidence evaluation — We assess study quality, sample sizes, and methodology
  4. Score calculation — We apply our scoring methodology transparently
  5. Plain-language summary — We write accessible summaries at an 8th-grade reading level
  6. Review and publish — Human review before publication

Update Schedule

Claims are reviewed and updated when significant new evidence is published. At minimum, each claim is re-evaluated annually. The "Last Reviewed" date on each page shows when it was last checked.

Conflict of Interest

EvidenceCheck has no financial relationships with supplement companies, food manufacturers, or pharmaceutical companies. Our confidence scores are never influenced by commercial interests. We are funded through premium subscriptions and API access, not advertising or sponsorship.

Report an Error

If you believe a claim contains an error, outdated information, or a misrepresented study, please contact us. We take corrections seriously and will update claims promptly when errors are identified.