Human-Trial Evidence Index — v1
Pure City Research · purecityresearch.org
Version: v1 (first 25 ingredients — expanding to 50)
Last reviewed: 2026-09-04
License: CC BY 4.0 (dataset + article)
Authorship: Pure City Research (organization-only; no personal bylines)
---
Editorial Introduction
Pure City Research is editorially independent. We are building a reader-supported funding model — no advertising, no manufacturer sponsorship. This index is the first public artifact of that promise: a single, openly licensed table ranking 25 widely used dietary-supplement ingredients by the strength of their human clinical-trial evidence — not by marketing, popularity, or mechanistic plausibility. Every row is tied to a public source (PubMed, Cochrane, or NIH Office of Dietary Supplements). Where the human evidence is thin, the table says so plainly. We grade conservatively: a supplement does not earn "Strong" by tradition or testimonials, only by replicated, human-trial proof. This is v1 — 25 ingredients drawn from topics the site already covers deeply, with new rows added as the evidence is reviewed. The CSV mirror is intended for Zenodo DOI assignment and a GitHub mirror so journalists, clinicians, and AI engines can cite and verify every cell.
---
Methodology
Scope. v1 covers 25 ingredients selected from topics purecityresearch.org already covers (magnesium, CoQ10, nattokinase, melatonin, omega-3, berberine, ashwagandha, L-theanine, creatine, vitamin D, collagen, probiotics, garlic, psyllium, zinc-L-carnosine, L-glutamine, rhodiola, St John's wort, saffron, L-tyrosine, tongkat ali, myo-inositol, glutathione, astaxanthin, shilajit). v1 will expand toward 50.
Evidence standard. Human evidence only: randomized controlled trials (RCTs), systematic reviews, and meta-analyses, plus Cochrane reviews, NIH Office of Dietary Supplements (ODS) fact sheets, NCCIH, FDA/DailyMed, and NCBI Bookshelf/StatPearls for safety and regulatory context. Mechanistic or in-vitro data is never presented as clinical proof — consistent with the site's existing nattokinase and magnesium articles.
Grading rubric (matches the site's published Strong/Moderate/Weak/Insufficient vocabulary).
- Strong — multiple high-quality meta-analyses of RCTs with consistent, clinically meaningful effects; large, replicated participant pools; independent (non-industry) replication; recognized by guideline bodies or carrying an authorized health claim.
- Moderate — meta-analyses or several RCTs showing a real but modest effect, or a strong effect confined to a narrow population; some heterogeneity or funding concerns; not yet sufficient for "Strong."
- Weak — small, short, few, or single-arm trials; high risk of bias; inconsistent or industry-confounded results; positive signals that disappear under independent scrutiny.
- Insufficient — no robust human outcome trials; only biomarker or pilot data; poor bioavailability undermines the claimed route; or claims rest entirely on in-vitro/mechanistic evidence.
How counts were derived. Human RCT counts and pooled participant figures are taken from the largest accessible named meta-analysis or systematic review per ingredient, supplemented by PubMed/academic search verification (pplx_sdk). Where a precise pooled N could not be independently verified, the cell is marked "needs verification" rather than estimated.
Funding flag. Industry-funded evidence is treated as a downgrade signal, exactly as the site's magnesium article downgrades the industry-funded magnesium-L-threonate study. Several rows (collagen, rhodiola, tongkat ali, shilajit, ashwagandha) note industry affiliation explicitly.
Limitations. (1) Counts reflect trials indexed in PubMed/Cochrane under common search terms and may omit regional or non-English-language trials. (2) "Pooled participants" is taken from a single named meta-analysis per row and is not a total across all trials ever published. (3) Grades are a structured editorial judgment, not a formal GRADE or Cochrane risk-of-bias assessment — they are reproducible but not equivalent to one. (4) This is v1; rows will be revised as new trials publish. (5) The index is educational, not treatment guidance; users should consult clinicians.
---
The Evidence Index (v1, 25 ingredients)
| # | Ingredient | Primary claimed use | Human RCTs (found) | Pooled participants (key meta-analysis) | Effect direction | Evidence grade | Grade justification (one line) | Key source citation | Typical effective dose (trials) | Headline safety note |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Magnesium | Blood pressure / migraine prevention adjunct | 34 | 2,028 (34-RCT BP meta-analysis) | Positive (modest) | Moderate | Consistent modest BP reduction + moderate migraine support; weak for sleep/cramps | Kass et al., Eur J Clin Nutr 2012; NIH ODS Magnesium | 300–600 mg elemental/day (migraine); >400 mg (BP) | Supplemental UL 350 mg/day; hypermagnesemia risk in renal impairment |
| 2 | CoQ10 (ubiquinone/ubiquinol) | Statin muscle pain / heart-failure adjunct | 12 | 575 (JAHA 12-RCT statin-myalgia meta) | Positive | Moderate | Consistent statin-myalgia effect + Q-SYMBIO HF adjunct; null for general energy/anti-aging | JAHA meta-analysis (12 RCTs); Mortensen Q-SYMBIO, JACC 2014 | 100–200 mg/day (myalgia/migraine); 300 mg/day (HF) | May lower warfarin INR; fat-soluble, take with food |
| 3 | Nattokinase | Blood pressure / cardiovascular markers | 6 | 546 (2023 6-RCT meta-analysis) | Positive (modest BP) | Weak | Modest BP effect in a small meta-analysis; clot-dissolution/Long-COVID claims have zero human RCT support (in-vitro only) | 2023 systematic review & meta-analysis, PMID 39076715 | 2,000 FU/day (~100 mg) | Bleeding risk; avoid with anticoagulants/antiplatelets; stop pre-surgery |
| 4 | Melatonin | Circadian timing / jet lag / DSWPD | Many (multiple) | needs verification (large across Cochrane + sleep-latency meta-analyses) | Mixed (positive circadian; modest insomnia) | Moderate | Robust for jet lag/circadian phase (Cochrane); only ~7–10 min sleep-onset latency gain in chronic insomnia | Cochrane review (jet lag); AASM clinical guideline | 0.3–1 mg (physiological); 30–60 min before bed | Next-day grogginess at high doses; commercial potency variability (JAMA gummy study 74–347% of label) |
| 5 | Omega-3 (fish oil, EPA/DHA) | Cardiovascular risk / triglyceride lowering | 86+ | 135,291 (14-RCT CHD meta-analysis, Gong 2022) | Mixed | Moderate | Large trial base; modest MACE reduction in some meta-analyses, null in others; benefit concentrated in high-risk/high-dose EPA (REDUCE-IT) | Cochrane (Abdelhamid 2020); AHA Advisory 2017; Gong et al. 2022 | 1 g/day EPA+DHA (general); 4 g/day icosapent ethyl (high-risk Rx) | Bleeding risk at high doses; atrial-fibrillation signal at high doses; fish allergy |
| 6 | Berberine | Blood glucose / HbA1c support (T2D adjunct) | 37+ | 3,048 (Xie 2022, 37-RCT meta-analysis) | Positive | Moderate | Consistent HbA1c (~−0.6–0.7%) and lipid improvements; "Nature's Ozempic" claim false | Xie et al. 2022, Front Pharmacol; NIH ODS | 900–1,500 mg/day divided, with meals | Potent CYP3A4 inhibitor (drug interactions); EFSA genotoxicity concerns; avoid in pregnancy |
| 7 | Ashwagandha | Stress / cortisol reduction | 15 | needs verification (2025 15-trial meta-analysis) | Positive (modest) | Moderate | Modest cortisol/stress/sleep gains in meta-analyses; trials small/short; testosterone/athletic claims weak | 2025 meta-analysis of 15 RCTs; NIH LiverTox | 300–600 mg/day standardized root extract (KSM-66/Sensoril) | LiverTox Likelihood B (drug-induced liver injury); avoid in pregnancy, autoimmune/thyroid disease |
| 8 | L-Theanine | Calm focus / caffeine synergy | Multiple (small) | needs verification (small acute trials) | Mixed | Weak | Reliable caffeine-synergy cognitive benefit; null for clinical GAD (8-wk RCT, p=0.73) | Randomized crossover EEG trials; GAD RCT | 100–200 mg (2:1 ratio with caffeine) | GRAS; theoretical additive effects with sedatives/antihypertensives |
| 9 | Creatine monohydrate | Muscle strength / lean mass (ergogenic) | 23 | large across 2 meta-analyses (2024: 23 RCTs; 2025: 20 studies, n=1,093) | Positive | Strong | Most-researched ergogenic; large meta-analyses + ISSN position stand; excellent safety profile | 2024 meta-analysis of 23 RCTs (+4.43/+11.35 kg strength); ISSN position stand | 3–5 g/day (or 20 g/day load ×5–7d) | Benign water retention; no kidney damage in healthy; DHT/hair-loss myth unreplicated |
| 10 | Vitamin D | Bone health / deficiency correction | Many (large) | >25,000 (VITAL trial) | Mixed | Moderate | Definitive for deficiency/osteomalacia; VITAL/D2d null for broad cancer/CVD/diabetes prevention in replete adults | VITAL (NEJM 2019); D2d; Cochrane | 600–800 IU/day (RDA); 2,000 IU (VITAL) | UL 4,000 IU/day; hypervitaminosis/hypercalcemia at very high chronic doses |
| 11 | Collagen peptides | Skin aging / joint support | 23 | needs verification (2025 AJM 23-RCT meta-analysis) | Mixed | Weak | Positive skin effects disappear when industry-funded studies excluded; tendon biomarker signal only | 2025 meta-analysis of 23 RCTs, Am J Med | 2.5–10 g/day (cosmetic); 15 g gelatin+vit C pre-exercise (tendon) | Generally safe; heavy-metal contamination risk (animal-derived) — third-party testing advised |
| 12 | Probiotics (strain-specific) | Antibiotic-associated diarrhea prevention | Many (Cochrane/meta) | large across strain-specific meta-analyses (AAD RR ~0.63) | Positive (strain-specific) | Moderate | Robust for specific strains (S. boulardii, L. rhamnosus GG) for AAD; broad immune/weight claims unproven | Cochrane/meta-analysis AAD (RR ~0.63); ESPGHAN guidelines | 5–20 billion CFU/day of a validated strain | Generally safe; serious infection risk in immunocompromised/critically ill/premature infants |
| 13 | Garlic (Aged Garlic Extract) | Blood pressure reduction | Many (meta-analyses) | needs verification (RCT meta-analyses) | Positive (modest, hypertensives) | Moderate | Consistent modest BP reduction (−3.7 to −4.4 mmHg systolic) in hypertensives; null in normotensives | Meta-analyses of RCTs; Cochrane | 600–1,200 mg/day Aged Garlic Extract | Antiplatelet/bleeding risk; stop 7–10 days pre-surgery; GI upset, odor |
| 14 | Psyllium husk | LDL cholesterol / constipation | >40 | large (>40-RCT meta-analysis) | Positive | Strong | Robust meta-analyses + FDA-authorized health claim for LDL reduction; effective bulk-forming laxative | Meta-analysis of >40 RCTs; FDA 21 CFR 101.81 health claim; Cochrane | 7–10 g/day soluble fiber (LDL); 5–15 g/day (constipation) | Esophageal/bowel obstruction if taken without water; separate from medications 2–4 h |
| 15 | Zinc-L-carnosine (polaprezinc) | Gastric mucosal healing | Moderate (Japanese RCTs) | needs verification | Positive (mucosal) | Moderate | Approved anti-ulcer drug in Japan with RCT support for mucosal healing & H. pylori-eradication adjunct; "leaky gut" claims weak/industry-affiliated | Japanese clinical approval data; RCTs; Cochrane | 150 mg/day complex (34 mg elemental Zn), divided | Copper depletion with long-term use near 40 mg Zn UL |
| 16 | L-Glutamine | Post-infectious IBS-D / intestinal permeability | Moderate | 106 (Zhou et al. RCT) | Positive (narrow) | Moderate | Strong RCT for post-infectious IBS-D (15 g/day); FDA-approved (Endari) for sickle cell; athletic & general "leaky gut" claims unsupported | Zhou et al. RCT (106 adults); FDA Endari approval | 15 g/day (5 g ×3) for IBS-D | Contraindicated in severe hepatic impairment/cirrhosis (hyperammonemia) & advanced CKD |
| 17 | Rhodiola rosea | Fatigue / stress ("adaptogen") | 11 | needs verification (11-trial systematic review) | Mixed | Weak | Independent reviews cite high bias, small samples, inconsistent outcomes; inferior to sertraline; industry-confounded | University of Alberta systematic review of 11 RCTs; EMA herbal monograph | 200–600 mg/day standardized extract (3% rosavins/1% salidroside) | Generally well-tolerated; theoretical serotonergic interactions with SSRIs/SNRIs |
| 18 | St John's wort | Mild-to-moderate depression | 29 | 5,489 (Cochrane 29-trial review) | Positive | Moderate | Cochrane: superior to placebo, comparable to SSRIs for mild-mod depression; but potent CYP3A4/P-gp inducer with dangerous interactions | Cochrane systematic review of 29 trials (n=5,489); Apaydin et al. meta | 900 mg/day standardized extract (0.3% hypericin) | Dangerous interactions — oral contraceptives, warfarin, immunosuppressants, antiretrovirals; serotonin syndrome with SSRIs |
| 19 | Saffron (Crocus sativus) | Depressive/anxious mood (self-report) | 34 | 1,769 (2026 GRADE 34-RCT meta-analysis) | Mixed | Moderate | 34-RCT meta shows self-report improvement but null on clinician-rated scales; high heterogeneity; mostly Iran | 2026 GRADE meta-analysis of 34 RCTs (n=1,769) | 30 mg/day standardized stigma extract (15 mg ×2) | Contraindicated in pregnancy (uterine stimulation); theoretical antiplatelet/hypotensive additive effects |
| 20 | L-Tyrosine | Acute-stress cognitive preservation | Multiple (small military/lab) | needs verification (small crossover trials) | Mixed | Weak | Preserves cognition under acute stress (cold/sleep deprivation/multitasking); null for baseline cognition/depression/ADHD | USARIEM/Dutch Navy RCTs; neurochemical reviews | 100–150 mg/kg (~7–10 g) single dose pre-stress | Contraindicated with MAOIs (hypertensive crisis); competes with levodopa; caution in hyperthyroidism |
| 21 | Tongkat ali (Eurycoma longifolia) | Testosterone support (hypogonadal men) | Small number | needs verification (small trials) | Mixed | Weak | Modest testosterone rise in hypogonadal/older men via SHBG displacement; small trials; eugonadal/athletic claims unproven | Placebo-controlled RCTs; FDA/EFSA reviews | 100–200 mg/day standardized hot-water extract | Hepatotoxicity case reports; heavy-metal (lead/mercury) & sildenafil adulteration risk |
| 22 | Myo-inositol | PCOS metabolic/hormonal support | Many (2023 PCOS Guideline meta) | large across meta-analyses (4,000 mg/day trials) | Positive (metabolic) | Moderate | 2023 International PCOS Guideline: improves insulin sensitivity/menstrual cyclicity; live-birth evidence low-certainty; 40:1 ratio contested | 2023 International PCOS Guideline meta-analyses; Cochrane | 4,000 mg/day (2,000 mg ×2) + folic acid | Excellent tolerability; mild GI effects at >12 g/day |
| 23 | Glutathione (oral) | Antioxidant support / "detox" / skin | Small number | small (no robust outcome meta-analysis) | Mixed | Insufficient | Oral bioavailability <1%; small RCTs show biomarker elevation only with novel delivery (liposomal/orobuccal); no large outcome trials | Allen & Bradley 2011 RCT; Sinha et al. 2018 liposomal RCT; NIH ODS | 500–1,000 mg/day (liposomal/orobuccal better absorbed) | Generally well-tolerated; "skin-whitening" high-dose use unproven; avoid inhaled forms |
| 24 | Astaxanthin | Skin moisture / elasticity | 11 | needs verification (11-study meta; 7-RCT lipid meta n=280) | Mixed | Weak | Moderate skin moisture/elasticity signal (SMD 0.53/0.77) but no wrinkle reduction; "internal sunscreen" & lipid/BP claims null | 2021 systematic review/meta of 11 studies; 7-RCT lipids meta (n=280) | 4–12 mg/day | Excellent safety profile; benign orange-tinted stools at high doses; take with fat |
| 25 | Shilajit | Testosterone / vitality (older men) | 1 (small) | ~60–75 (single 90-day RCT) | Mixed | Insufficient | Single small manufacturer-affiliated RCT; energy/performance claims rest on uncontrolled pilot; raw resin heavy-metal risk | 90-day RCT (healthy men 45–55); regulatory advisories | 250–500 mg/day purified extract (PrimaVie) | Raw resin heavy-metal (lead/arsenic/mercury) risk; avoid in hemochromatosis/gout/hypotension |
Reading the table. "Human RCTs (found)" and "Pooled participants" are drawn from the largest accessible named meta-analysis per ingredient and are not exhaustive totals. "Effect direction" summarizes the overall human-trial signal for the primary claimed use: Positive = consistent benefit; Mixed = benefit in some populations/endpoints but not others, or positive self-report with null objective measures; the corresponding grade reflects this nuance. Cells marked "needs verification" could not be independently confirmed at publication and are flagged for the next review cycle rather than estimated.
---
Draft Show HN post
Title: Show HN: Human-Trial Evidence Index — 25 supplements ranked by actual human-trial strength (open CSV, every cell cited)
Body:
I help run purecityresearch.org, an editorially independent, reader-funded supplement-evidence site (no ads, no manufacturer sponsorship). The biggest gap we saw: when people ask "which supplements actually have evidence?", the answer usually comes from a single ad-supported database. So we built an openly licensed, machine-readable index that ranks 25 widely used supplement ingredients purely by the strength of their human clinical-trial evidence — not popularity or mechanistic plausibility.
Each row records: primary claimed use, number of human RCTs, pooled participants from a named meta-analysis, direction of effect (positive/mixed), an evidence grade (Strong/Moderate/Weak/Insufficient), a one-line justification, a verifiable source citation (PubMed/Cochrane/NIH ODS), the typical effective dose from trials, and a headline safety note.
Every value traces to a public source. Where we couldn't independently verify a number, we wrote "needs verification" instead of guessing. Industry-funded evidence is treated as a downgrade signal — for example, collagen's positive skin effects disappear once industry-funded trials are excluded. The full table plus a CSV mirror (intended for a Zenodo DOI + GitHub mirror) is in the repo.
This is v1 (25 of a planned 50 ingredients). We grade conservatively: creatine and psyllium reach Strong; melatonin, magnesium, berberine, St John's wort land Moderate; rhodiola, tongkat ali, shilajit, glutathione are Weak/Insufficient. We'd love feedback on the grading rubric and any rows where the evidence picture is wrong.
---
Dataset mirror
- CSV: `evidence-index-v1.csv` (same 25 rows, machine-readable, CC BY 4.0)
- Intended for Zenodo DOI assignment + GitHub mirror at `github.com/pure-city/evidence-index`
- JSON-LD `Dataset` + `MedicalWebPage` schema to be added at publication (per Proposal 1, §2)
---
*Pure City Research is editorially independent and reader-supported. This index is educational, not medical advice; consult a clinician before starting, stopping, or changing any supplement, especially if you take prescription medications. Several ingredients listed (St John's wort, berberine, nattokinase, garlic, omega-3) carry clinically significant drug-interaction or bleeding risks.*
Have a question — or want us to cover something?
Ask about anything on this page, or request the next deep dive: an ingredient, a supplement, or a health concern. We use published research, evidence syntheses, and regulatory guidance, with clear source links.
One daily research roundup
Get the topics, key findings and links from our new articles in one email. At most one digest a day, only when there is something new.