How We Research Supplements
Updated 3 November 2026 · Root & Evidence editorial content · No independent clinical review claimed
Curiosity starts the question. Transparent methods help you judge the answer.
On this page Jump to a section ↓
Start with a specific question
We define the population, supplement and formulation, comparator, outcome, and follow-up. A claim about sleep does not stand in for a claim about stress. Traditional use provides context, not proof of benefit.
Search and source selection
For the current selected-study guides, we checked PubMed-indexed publications and NIH ODS or NCCIH background and safety resources. We followed relevant source links and included negative findings. This was a targeted search, not a comprehensive systematic search; we do not claim that all eligible studies were found. Sources were checked on 3 November 2026. Future full reviews should publish databases, dates, exact search strings, and a study-selection log.
Study inclusion criteria
We prioritize randomized human trials relevant to the stated outcome and systematic reviews of such trials. We record the tested form, dose, population, comparator, duration, and measured outcomes. Observational evidence may inform longer-term safety but cannot establish causality on its own. Cell and animal findings are background only. Abstract-only details are treated cautiously; missing information is not invented.
Handle conflicting results
We keep null findings alongside favorable ones. We compare population, formulation, dose, baseline risk, outcome definitions, comparator, and follow-up before interpreting disagreement. Secondary and subgroup findings are identified as such. We avoid double-counting a trial and the meta-analysis containing it.
Assess quality and certainty
We consider randomization, blinding, attrition, selective reporting, precision, consistency, directness, and potential publication bias. Funding and conflicts are relevant context, not automatic proof that a result is wrong. Our current guides do not include completed formal risk-of-bias assessments and remain Not yet rated. The five editorial labels are explained on the Evidence ratings page; they are not a validated score or formal GRADE assessment.
Explain practical meaning
We distinguish statistical significance from the size and importance of an effect, primary from secondary endpoints, and surrogate markers from outcomes people experience. Study doses describe research, not personal treatment recommendations. Quality testing and clinical efficacy are separate questions.
Update and correct transparently
Readers can submit a source and concern through Contact. We check the cited material, correct substantive errors when confirmed, and add a dated note to the affected page. Safety-sensitive reports receive priority. New evidence may change a conclusion. We do not promise continuous literature monitoring or a fixed response deadline.
Editorial independence
Shopping opportunities do not determine evidence conclusions or ratings. Current Amazon links are ordinary searches with no affiliate tracking or commission. If monetization is introduced, relationships must be clearly disclosed beside relevant links. Sponsors must not buy favorable conclusions or suppress limitations. No external clinician review or professional credentials are claimed for these guides.
Evidence ratings ↗ · Corrections policy ↗ · Affiliate disclosure ↗
Method references: GRADE Working Group ↗ · Cochrane Handbook, certainty of evidence ↗