Glossary

Plain-language definitions for the research terms used across the site. If you've ever wondered what "p < 0.05" means, this is for you.

RCTRandomized controlled trial

The strongest single-study design. Participants are randomly assigned to receive the intervention (e.g. a supplement) or a control (placebo or standard care). Randomization balances confounders so any difference in outcome is attributable to the intervention.

DBPCDouble-blind, placebo-controlled

An RCT where neither the participant nor the researchers measuring outcomes know who received the active compound. Eliminates placebo and observer bias. The gold standard for supplement evidence.

CrossoverCrossover trial

Each participant receives both the intervention and the control, separated by a washout period. Smaller sample size needed but only works for short-term, fully-reversible effects.

CohortCohort study

Observational. Follows a group of people over time and observes which ones develop an outcome. Can show association, not causation. Confounding (e.g. healthy-user bias) is the main weakness.

Case-controlCase-control study

Observational. Compares people with a condition to people without it, looking backward for differences. Cheap and fast but prone to recall bias.

In vitroIn vitro

Latin for "in glass". Done in cells or test tubes, not in living humans or animals. Hypothesis-generating; rarely translates directly to human effects.

SRSystematic review

A structured search across all published evidence on a question, using pre-specified inclusion criteria. Reduces cherry-picking. Often paired with a meta-analysis.

Meta-analysisMeta-analysis

Pools the numeric results of multiple RCTs to compute a more precise summary effect. Cochrane reviews are considered the highest-quality form. The site's Tier 1 calls usually rest on a meta-analysis with positive results.

NMANetwork meta-analysis

A meta-analysis that compares 3+ interventions even when not all of them have been directly compared head-to-head. Used to rank treatments (e.g. the 2025 Cheng NMA ranking nutraceuticals against antidepressants).

GRADEGRADE evidence rating

A standardized framework that rates the certainty of evidence as High, Moderate, Low, or Very Low based on study design, consistency, directness, precision, and publication bias.

PRISMAPRISMA reporting standard

The PRISMA checklist standardizes how systematic reviews report their methods, so you can reproduce the search and judge what was excluded.

SMDStandardized mean difference

The intervention's effect expressed in standard deviations. SMD 0.2 ≈ small, 0.5 ≈ moderate, 0.8 ≈ large. Lets you compare effect sizes across trials that used different scales.

RRRelative risk

The ratio of an event happening in the treated group vs control. RR 0.7 means 30% lower risk; RR 1.5 means 50% higher risk.

OROdds ratio

Like RR but uses odds (event:no-event) instead of probability. Approximates RR when the event is rare.

HRHazard ratio

Like RR but accounts for time-to-event. Used in survival analysis.

CIConfidence interval (95% CI)

The range within which the "true" effect probably lies. If the 95% CI for an RR crosses 1.0, the result is not statistically significant.

p-valuep-value

The probability that a result this large would happen by chance if the intervention had no real effect. p < 0.05 is the conventional cutoff for "statistically significant" — but it doesn't mean clinically meaningful.

NNTNumber needed to treat

How many people you'd need to treat to prevent one bad outcome. NNT 10 = treat 10 people for one to benefit. The lower the better.

PMIDPubMed identifier

A unique number assigned by the U.S. National Library of Medicine to every paper indexed in PubMed. Lets you look up the exact source — every PMID on this site is a clickable link.

DOIDigital Object Identifier

A persistent ID for any published document. Resolves to the publisher's page (where the full PDF often lives behind a paywall, sometimes not).

PMCPubMed Central

The free full-text repository for biomedical research. ~6 million papers; a subset of PubMed.

COIConflict of interest

Any financial or personal relationship that could bias the research. We tag every cited paper with its disclosed COI status.

Funder typeFunder type

Who paid for the research. Categorised on this site as public (NIH, NHS, etc.), nonprofit (American Heart Association, etc.), industry (a supplement company), mixed, or none_disclosed. Tier 1 calls require at least one public/nonprofit citation.

ULTolerable Upper Intake Level

The highest daily intake of a nutrient unlikely to cause harm in nearly all people. Set by NIH/NAS/EFSA. Above the UL, risk of toxicity rises.

RDARecommended Dietary Allowance

The intake sufficient to meet the requirement of 97–98% of healthy people. Below the RDA, deficiency risk rises.

AIAdequate Intake

Used when an RDA can't be set due to insufficient data. The AI is the average intake assumed to maintain adequacy.

EAREstimated Average Requirement

The intake that meets the requirement of 50% of healthy people in a group. Below the EAR, deficiency is statistically common.

CYP450Cytochrome P450

The liver's main drug-metabolising enzyme system. CYP3A4, CYP2D6, CYP1A2, CYP2C9, CYP2C19 are the most clinically important. Many supplements either induce (speed up) or inhibit (slow down) these — which changes prescription drug levels.

P-gpP-glycoprotein

A pump that ferries drugs out of cells. Some supplements modulate P-gp activity, which changes how much of an oral drug actually gets absorbed and how much gets cleared.

Half-lifeHalf-life

The time it takes for half of a dose to clear from the bloodstream. After 5 half-lives, ~97% is gone. Determines dosing frequency.

AUCArea Under the Curve

Total drug exposure over time. A higher AUC means more drug reached the bloodstream — relevant when evaluating bioavailability or absorption-enhancing pairings (e.g. piperine + curcumin).

CmaxCmax

Peak blood concentration after a dose.

SNPSingle nucleotide polymorphism

A common one-letter difference in DNA. Some SNPs affect how you metabolise nutrients or drugs.

MTHFRMTHFR variant

The methylenetetrahydrofolate reductase gene. Common variants (C677T, A1298C) reduce conversion of folic acid to its active form. Carriers benefit from supplementing methylfolate (5-MTHF) directly.

APOE4APOE4 carrier

A variant of the apolipoprotein E gene linked to higher Alzheimer's risk. Some supplement effects (e.g. omega-3, B vitamins) differ by APOE status.

Tier 1Tier 1 (Strong Evidence)

Composite score ≥ 72 AND at least one cited PMID with public or nonprofit funding. The "I'm comfortable taking this" tier.

Tier 2Tier 2 (Promising)

Composite score 60–71. Real evidence but limited to specific populations, doses, or outcomes.

Tier 3Tier 3 (Trending)

Popular in wellness culture but weak or inconsistent clinical evidence.

Tier 4Tier 4 (Risky / Avoid)

Documented safety risks (organ damage, drug interactions, deaths). Regulatory warnings issued.