Creatine

Separating scientific consensus from scientific uncertainty.

This is not a case for creatine. It is the evidence base, laid out so it can be attacked. Every claim carries its design, sample size, heterogeneity, publication-bias test, funding source and GRADE certainty. Where the evidence is weak, the site says so and shows you why.

01 · Mechanism

What creatine actually is

A guanidino compound, synthesised endogenously at roughly 1 g/day, stored at about 95% in skeletal muscle, and used as a phosphoryl shuttle rather than a simple reservoir. The mechanism is not in dispute. What follows from it is.

~1 gSynthesised per day by liver, kidney and pancreas
95%Of the body pool held in skeletal muscle
~20%Typical increase in muscle total creatine on supplementation
1.5–2%Of the pool lost per day as creatinine — non-enzymatically
The pathway
Click any step for the enzyme, the reaction and why it matters.
The molecule
Drag to rotate. Hover an atom for its assignment. Switch to phosphocreatine to see the phosphate land on the guanidino nitrogen.
02 · Evidence hierarchy

Not all of this evidence weighs the same

Rotate the pyramid. Note where the widely-quoted professional position stand actually sits — at the base, with expert opinion, not at the apex. That single placement decides how most of the public conversation about creatine should be weighted.

03 · The syntheses

What the meta-analyses actually found

Six syntheses, each opened up in full: the pooled estimates with confidence intervals, the heterogeneity, the publication-bias tests, the funding, the declared conflicts, and the weaknesses. Nothing is summarised away.

04 · Critical appraisal

How this literature answers the standard questions

The nine questions a reviewer asks of any body of evidence — and, honestly, how creatine research answers each one. Three of the nine do not go well.

05 · Why confidence differs

Sample size did not decide these verdicts

The older-adult synthesis and the depression synthesis pooled the same number of participants — 1,093. One is rated HIGH certainty, the other VERY LOW. Heterogeneity, publication bias and effect magnitude did that, not n.

Heterogeneity (I²) across every reported outcome
The proportion of variation between trials that is real disagreement rather than chance.
What publication bias looks like
Compare a symmetric funnel with an asymmetric one from the same meta-analysis.
Effect magnitude, in the units you live in
Statistical significance and clinical importance are different questions.
Every synthesis, side by side
Click a column heading to sort.
06 · Where the evidence fails

The case against creatine for depression

This is the section that should decide whether you trust the rest of the site. A statistically significant result, 1,093 participants, published in a respected journal — and it still does not support the claim. Watch the scale.

07 · Safety atlas

Organ by organ, evidence and unknowns

Select an organ. Each entry separates what has been demonstrated, the mechanism that explains it, what remains genuinely unknown, and who should not take it.

08 · The chemistry

Why phosphocreatine, and not simply more ATP

The thermodynamic answer, the bond that carries it, and the isoenzyme shuttle that most popular accounts leave out.

Phosphocreatine + ADP + H⁺  ⇌  Creatine + ATP
09 · Both columns

The argument, with receipts on each side

Click any claim to see the evidence behind it — including the claims against. Nothing here is hidden, and the conflicts of interest are flagged on the claims they affect, including the ones favourable to creatine.

Arguments for

Arguments against / limits

10 · Your standards

Set your own evidentiary bar

You do not have to accept this site's weighting. Impose your own. Discard industry-funded work, demand large samples, insist on low heterogeneity and no detected publication bias — the conclusions recompute live, and the site tells you honestly what survives.

11 · Professional bodies

What the organisations say — and where they diverge

Consensus statements are useful and are also the weakest tier of evidence. Each entry below carries the caveat that belongs with it.

12 · In practice

What the evidence supports doing

Stated conservatively, at the level the evidence actually licenses.

13 · Sources

Every source, with its DOI

Search and filter. Anything the site could not verify against the primary source is flagged rather than quietly presented as fact.