Supplements — What The Evidence Actually Supports

The supplement industry is built on an asymmetry: the cost of making a claim is near zero, and the cost of disproving one is a multi-year clinical trial. That leaves buyers sorting a shelf where a substance with hundreds of randomized controlled trials sits next to one with a press release, at similar prices, in similar packaging.

This hub sorts them. Every guide here is evidence-ranked, dose-specific, and explicit about which supplements only work if you are actually deficient — because that last category is where most supplement spending is wasted.

The evidence tiers

Ranked by strength of the underlying human evidence, not by popularity or marketing spend. "Strong" means multiple randomized controlled trials with consistent, replicated results at the stated dose. "Moderate" means real trial support with smaller effect sizes or fewer replications. "Weak" means the mechanism is plausible but human outcomes have not delivered.

Test before you supplement — the six that require it

For most supplements, taking them without deficiency produces no benefit. For a few, it produces active harm. These six are where a blood test before purchase changes the decision.

How to read a supplement label

Four things separate a legitimate product from an expensive placebo, and all four are visible on the label before you buy.

1. Elemental content, not compound weight. A capsule listing "400mg magnesium glycinate" contains roughly 50–60mg of actual elemental magnesium. Reputable brands print both. Brands that print only the salt weight are relying on the bigger number looking better.
2. The form matters more than the brand. Magnesium oxide is poorly absorbed and functions mostly as a laxative; glycinate and threonate are absorbed well. Folic acid is inferior to methylfolate for the large share of people carrying MTHFR variants. The form is a bigger quality signal than the label design.
3. Third-party testing. NSF Certified for Sport, Informed Sport, and USP verification each mean an outside lab confirmed the contents. In an industry regulated as food rather than medicine, this is the only real assurance that the capsule contains what the label claims.
4. No proprietary blends. A "proprietary blend" discloses total weight but hides the per-ingredient dose. This exists almost exclusively so the manufacturer can include a study-backed ingredient at a fraction of the study dose. If the per-ingredient amounts are not printed, assume they are too low to matter.

The five mistakes that waste the most money

• Supplementing without a baseline. Without a starting measurement there is no way to know whether a supplement corrected anything. This is the single largest source of wasted supplement spend.
• Stacking 15+ items. Beyond about five supplements, interactions and redundancy accumulate and attribution becomes impossible. If everything changes at once, nothing is measurable.
• Under-dosing. Most consumer multivitamins and blends contain clinically meaningless amounts of their headline ingredients. Ashwagandha below 300mg, omega-3 below 1g combined EPA/DHA, and creatine below 3g are all effectively placebos.
• Chasing novelty. The supplement with the newest podcast coverage is, by definition, the one with the least human evidence. Creatine has been studied since the 1990s — that is a feature.
• Treating supplements as a substitute for the fundamentals. No stack compensates for chronic short sleep, no aerobic training, or excess alcohol. Supplements operate at the margin of a foundation that has to already exist.

Buying guides

• How to read a supplement label — what 21 CFR 101.36 actually requires in the Supplement Facts panel, and which reassuring parts of a label carry no information
• Berberine dosage and timing — the doses used in trials, HCl vs phytosome, and the drug interactions that matter
• Berberine vs metformin — mechanism, evidence strength, and an honest answer to the "natural Ozempic" claim
• Best creatine supplement — why the form question is already settled, the active-creatine arithmetic per serving, and which third-party marks mean anything
• Best collagen supplement — the two choices that are not interchangeable: gram-scale hydrolysed peptides, or 40 mg undenatured type II
• Does collagen help joints? The answer depends on which pain scale a trial used, and neither arthritis guideline recommends it or advises against it
• Best cortisol supplement — ashwagandha, phosphatidylserine, rhodiola and what the evidence supports
• Best deep sleep supplement — nothing on the shelf has good evidence for raising slow-wave sleep, and what moves it instead
• Best electrolyte powder — the consensus statement says the dangerous hydration error is drinking too much, and no powder’s sodium prevents it
• Best magnesium supplement — glycinate vs threonate vs citrate vs oxide
• Best protein powder — DIAAS amino-acid scores, scoop-purity arithmetic, and the 2024 independent heavy-metal results
• Best vitamin D supplement — D3 over D2, and the three things that decide a blood level which are not printed on the bottle
• Best time to take vitamin D — why fat matters more than time of day
• Best B12 supplement — the NIH states no form absorbs better than another, and sublingual is no different from a swallowed tablet
• B12 injections vs oral: high-dose tablets reach the same blood levels in head-to-head trials, while guidelines still specify injections for autoimmune gastritis and terminal ileal resection
• Best vitamins for energy — what works when you are deficient, and what does not
• Creatine and kidney markers — why it raises creatinine and lowers eGFR without harming the kidney
• Creatine for longevity — the best-evidenced option here, and the lab artifact it creates
• Creatine for women — dosing, the menopause and bone evidence, and the water-weight question
• Creatine monohydrate vs HCl — the solubility claim, and the head-to-head trial nobody has published
• NMN vs NR — the NAD+ precursors, their trials, and the 2025 regulatory reversal
• Magnesium side effects: the dose-dependent gut effects, the 350 mg limit, and the kidney exception
• Magnesium glycinate and insomnia: what the trial evidence actually shows about the paradoxical response, and how much glycine a capsule really delivers
• NMN dosage: what each human trial administered, against what the labels recommend
• NMN side effects: what the trials reported, and how far that reassurance extends
• Senolytic supplements — fisetin, quercetin and what the human trials actually showed
• What is NAD+: the two jobs the coenzyme does, the five precursors that feed it, and what the human trials measured
• NAD+ and exercise — training raised muscle NAD+ by 127%; three biopsy trials of nicotinamide riboside raised it by nothing
• Tongkat ali vs Fadogia agrestis — five pooled trials and 267 people against zero human studies
• Whey vs plant protein — whey wins on both axes: a DIAAS of 100 to 107 against 62 to 84, and 28% of whey products over the Proposition 65 lead limit against nearly 80% of plant ones

Research Peptides

Clinics offer these under one heading, but their regulatory positions are not the same: one has an approved product, one sits on the US Food and Drug Administration's current restricted-compounding list, and the rest were nominated for compounding and then withdrawn. Each page below reports what was tested, in whom, and what the federal tables say. None of them gives a dose.

• Peptide therapy FDA status: seven molecules, their approved products and their position on the federal compounding tables
• PT-141 evidence: the one with an approved product, and the Limitations of Use that exclude most people searching for it
• KPV peptide evidence: two mouse colitis studies, a transporter-dependent mechanism, and no human exposure data
• Dihexa evidence: the blocking experiments that tie the effect to one receptor system, all of them in rodents
• DSIP peptide evidence: named for a deep-sleep effect its own trial authors called clinically insignificant
• Peptide therapy cost: why the membership, not the peptide, is most of the monthly bill

Prescription Longevity Drugs

• Rapamycin for longevity: the mouse lifespan data, the Dog Aging Project trial, and how thin the human evidence still is
• Rapamycin vs metformin: one has the better animal data, the other has the better human safety record

Related

• Biomarker guides — the tests that tell you what to supplement
• Normal vs optimal ranges
• Nutrient density — food-first alternatives to supplementation
• Nutrition guides
• Best fish oil supplement: what separates one bottle from another
• DHEA and testosterone: the trials found no change in men and a measurable rise in women
• Do fish oil supplements work: the trials, outcome by outcome
• What is creatine — the definitional guide to the most-studied supplement on this site
• Urolithin A — what the trials showed, and what the price buys
• Red yeast rice — why a product that works is a statin by another name
• Collagen side effects — the oxalate question answered properly, and which reported effects have no mechanism
• Collagen dosage and timing — the trial doses, and the fact that time of day has never been tested
• Collagen types explained — what Types I, II and III are, and why "all five types" is a marketing count
• Collagen vs collagen peptides — hydrolysis is the one format difference that changes what reaches your blood
• Best probiotic supplement — the evidence attaches to named strains at named doses, which is why most rankings answer the wrong question
• Ashwagandha benefits — claim by claim, what the trials support, how long it takes, and the liver cases