“Creatine works” is not one claim — it is one established claim wearing the clothes of a dozen
unestablished ones, and the whole decision turns on keeping them apart. The single held source is an
ISSN position stand (Kreider 2017), a persuasion-shaped, industry-funded consensus (see The source
caveat below); it is read here against the wiki’s own four-evidence-states discipline, not taken at its
enthusiasm. Sorted by evidence state. confidence: low (demoted from medium 2026-08-09): the sole source
is moderate-tier consensus with a sponsor COI — the ergogenic benefit is real but rests on a single,
non-systematic, industry-funded document, so the page cannot carry medium confidence.
| Claim | State | Basis |
|---|---|---|
| Ergogenic — high-intensity performance, and lean mass/strength with resistance training | benefit (surrogates) | consensus of ISSN + ADA + Dietitians of Canada + ACSM; effect from primary trials |
| Bone — BMD added to resistance training, older adults | no meaningful effect on the BMD surrogate (pooled); insufficient on fracture/bone-geometry | gold MA, 5 RCTs n=193, pooled MD ~0 at all four sites (Forbes 2018) |
| Safety — no renal/hepatic harm at 3-30 g/day; only consistent effect is weight gain | no meaningful harm | >1000 studies, doses to 0.8 g/kg/day for 5 yr |
| Cognition (working memory; under sleep-deprivation/stress) | insufficient / emerging | small human studies, brain creatine +5-15% |
| Neuroprotection (TBI, spinal cord, ischemia) | insufficient — mechanism | almost entirely rodent models |
| Neurodegenerative disease (Parkinson’s, ALS) clinical outcomes | no clinical benefit | the stand’s own cited large trials (Bender: 1687 pts) |
(inferred from Kreider et al., 2017)
The established lever — ergogenic benefit, on surrogates
The benefit that survives appraisal is ergogenic: creatine loading raises intramuscular phosphocreatine, and «After creatine loading, performance of high intensity and/or repetitive exercise is generally increased by 10–20% depending on the magnitude of increase in muscle PCr» (Kreider et al., 2017). The society’s position, and the reason to weight this above the other claims, is that it is a cross-body consensus, not one group’s view: «creatine monohydrate is the most effective ergogenic nutritional supplement currently available to athletes in terms of increasing high-intensity exercise capacity and lean body mass during training … Recent position stands by the American Dietetic Association, Dietitians of Canada, and the American College of Sports Medicine on nutrition for athletic performance all drew similar conclusions» (Kreider et al., 2017).
- Mechanism (human-corroborated, directional). Creatine buffers ATP resynthesis via the creatine kinase / phosphocreatine energy shuttle, which is why the benefit concentrates in maximal-effort, short-duration, repeated efforts (sprints, sets to near-failure) and not endurance (Kreider et al., 2017). Supplementation raises muscle creatine/PCr «by 20–40%» from a normally 60-80%-saturated store (Kreider et al., 2017).
- The benefit is on SURROGATES, like its sibling lever. Performance, lean mass, and strength are surrogates (Surrogate Outcomes), not patient-important endpoints — no mortality or function trial is held. This is the identical boundary as Protein and Resistance Training for Muscle and Strength: creatine is a second ergogenic adjunct to resistance training, working through a different pathway (PCr energy buffering) than protein (muscle-protein-synthesis substrate), so the two are additive, not redundant (inferred from Kreider et al., 2017).
Dose — a loading region and one evidenced form
(inferred from Kreider et al., 2017)
- Loading vs slow-fill. «The quickest method of increasing muscle creatine stores may be to consume ~0.3 g/kg/day of creatine monohydrate for 5–7-days followed by 3–5 g/day thereafter to maintain elevated stores. Initially, ingesting smaller amounts of creatine monohydrate (e.g., 3–5 g/day) will increase muscle creatine stores over a 3–4 week period, however, the initial performance effects of this method … are less supported» (Kreider et al., 2017). So loading buys speed to saturation, not a higher ceiling — the decision is time-to-effect, not magnitude.
- Monohydrate is the only form the evidence backs. «Claims that different forms of creatine are degraded to a lesser degree than creatine monohydrate in vivo or result in a greater uptake to muscle are currently unfounded» — citrate, serum, ethyl ester, buffered, and nitrate forms show no greater retention (Kreider et al., 2017). This is a label-is-not-the-exposure instance in the buyer’s favour: the cheapest, best-studied form is the one to buy -> Is the Food Category Doing Any Work.
- Baseline status modifies the response. Vegetarians, with lower muscle creatine stores (90-110 vs ~120 mmol/kg dry muscle), «may observe greater gains in muscle creatine content from creatine supplementation» (Kreider et al., 2017) — a repletion-vs-enhancement gradient (route (a)/(b)) analogous to the deficiency structure on Deficiency Repletion vs Enhancement, though here on a surrogate, not a hard outcome.
Safety — the decision-relevant finding, and the creatinine trap
The safety profile is the part of this document with the widest evidence base, and it directly answers a common belief (creatine harms the kidneys): «The only consistently reported side effect from creatine supplementation that has been described in the literature has been weight gain» (Kreider et al., 2017), and across dosing to 0.8 g/kg/day for up to 5 years «there is no compelling evidence that creatine supplementation negatively affects renal function in healthy or clinical populations» (Kreider et al., 2017).
- The creatinine confound (decision-change). Creatine is metabolised to creatinine — the same marker used to estimate kidney function — so supplementation can raise serum creatinine (and thus lower an estimated GFR) without any change in actual renal function; controlled studies of GFR and creatinine clearance found no true impairment (Kreider et al., 2017). The practical consequence: a raised creatinine in a creatine user is not, by itself, evidence of kidney damage — but it will be read as such by a clinician who does not know the person supplements. (inferred from Kreider et al., 2017)
- The weight gain is water + lean mass, not fat — partly acute osmotic fluid retention (~0.5-1.0 L on loading) (Kreider et al., 2017); a consideration for weight-class athletes, not a health harm.
- Symmetric-standards caveat: the safety conclusion is stated by an industry-funded body with an interest in it; it rests on a genuinely large literature, but a favourable safety claim from a conflicted source earns the same scrutiny as an unfavourable one (see The source caveat).
The unestablished claims — kept out of the benefit column
The position stand narrates a long list of potential medical uses; under the four-states discipline these are insufficient / not-yet / null, and must not be laundered into the ergogenic benefit:
- Neurodegenerative disease — a NULL from the stand’s OWN cited trials. The sharpest instance: the large human trials show no clinical benefit — «A total of 1,687 patients took an average of 9.5 g/day of creatine for a total of 5,480 patient years. Results revealed no clinical benefit on patient outcomes in patients with PD or ALS. However, there was some evidence that creatine supplementation slowed down progression of brain atrophy in patients with HD (although clinical markers were unaffected)» (Kreider et al., 2017). So on patient-important outcomes in Parkinson’s/ALS, creatine is no meaningful effect, not benefit — the document’s own evidence, against its framing.
- Neuroprotection (TBI, spinal cord injury, cerebral ischemia) — mechanism, mostly ANIMAL. The supporting studies are rats and mice (cortical damage, infarct size, gray-matter loss) (Kreider et al., 2017). This is the insufficient-evidence / not-yet state under the transportability caveat (rodent != human), not a human-outcome finding (inferred from Kreider et al., 2017).
- Cognition — emerging, small human studies. Brain creatine rises 5-15% and small trials report gains in working memory and in resisting the cognitive decrement of sleep deprivation / stress (Kreider et al., 2017) — a genuine candidate, held as insufficient (small samples, mixed tasks), not asserted as benefit.
- Pregnancy, ischemic heart — rationale + animal data only; not-yet (Kreider et al., 2017).
- Inborn creatine-synthesis deficiency (AGAT/GAMT) — a genuine but rare therapeutic repletion case (high-dose creatine improves outcomes), distinct from supplementing a replete person (Kreider et al., 2017).
Older adults — the one clinical stratum where the surrogate benefit is decision-relevant
The ergogenic benefit becomes more than athletic in the elderly, where muscle/strength/function are themselves on the outcome menu (sarcopenia). Two meta-analyses of elderly resistance-trainers found creatine added to training: participants «experienced greater gains in muscle mass, strength, and functional capacity» (357 individuals, ~64 y) and a second (405 individuals) corroborated greater lean mass and upper-body strength «compared to training alone» (Kreider et al., 2017). A third MA from a different group sharpens the point by contrast: Choi 2021 pooled 22 RT-vs-RT+nutrition RCTs in healthy older adults and found creatine was the only nutritional intervention to beat training alone — «Among the nutritional interventions, only those with creatine showed significant effects on lean body mass (n = 4, MD 2.61, 95% CI 0.51 to 4.72)» — every other nutrient subgroup was null (Choi et al., 2021). Its distinctive contribution is the within-MA head-to-head — creatine against protein/multinutrient in one pool — which the earlier MAs do not provide; not scored as an independent-E leg (Choi’s 4 creatine trials may overlap those MAs’ constituents, unverified, and the between-subgroup test was not significant), so it corroborates the DIRECTION on a small pool, not a precise magnitude. This is an F-refinement of Sarcopenia Definition and Diagnosis and Protein and Resistance Training for Muscle and Strength: for the stratum whose big rock is preserving strength and independence (Big Rocks (Elderly)), creatine is a small, evidenced adjunct on top of the resistance training that remains the driver — never a substitute for it (inferred from Kreider et al., 2017).
Bone — the adjunct benefit does NOT extend to BMD (a within-exposure dissociation)
The elderly adjunct story above (creatine adds to resistance training for muscle mass, strength, and function) has an intuitive extension — more muscle pull and direct osteoblast/anti-resorptive pathways should build bone — that the pooled evidence does not support. Forbes 2018, a gold meta-analysis (5 RCTs, n = 193, adults >50 or postmenopausal, creatine + resistance training vs resistance training + placebo, >=12 weeks), found no added bone-density effect at any site: «Meta-analyses revealed no greater effect of creatine and resistance training compared to resistance training alone on whole body BMD (MD: 0.00, 95% CI −0.01 to 0.01, p = 0.50), hip BMD (MD −0.01, 95% CI −0.02 to 0.01, p = 0.26), femoral neck BMD (MD 0.00, 95% CI −0.01 to 0.01, p = 0.71), and lumbar spine BMD (MD 0.01, 95% CI −0.01 to 0.03, p = 0.32).» (Forbes et al., 2018)
- The dissociation is the finding (type-F sharpening of the adjunct claim). Every included trial found creatine enhanced muscle mass or strength over placebo, yet «these adaptations did not translate to greater BMD» (Forbes et al., 2018). So creatine’s evidenced adjunct value is outcome-specific — real on lean mass / strength / function (the section above), absent on bone-mineral density. The plausible bone mechanisms (muscle-pull on bone; direct osteoblast stimulation and reduced bone-resorption markers) are a directional prior that the surrogate outcome did not confirm — net-effect-not-intended (inferred from Forbes et al., 2018).
- Four states, kept apart — a precise surrogate null, but NOT a foreclosed bone case. The pooled CIs are tight (roughly ±0.01 to 0.03 g/cm² around zero), so on the specific question — does creatine add to resistance training for BMD, at studied doses over 12 weeks–1 year — this is no meaningful effect, not merely insufficient evidence. It stays insufficient (unproven, not disproved) for three things the MA could not settle: (i) longer or higher-frequency / relative-dose protocols — the two relative-dose (0.1 g/kg/d) and higher-frequency (3x/wk) trials showed positive bone signals, a subgroup hypothesis generated post-hoc across 5 heterogeneous trials, never tested (Forbes et al., 2018); (ii) bone geometry / strength — «Creatine supplementation (10 g·d−1) during 12 months of resistance training (3 days/week) decreased femoral neck bone loss and increased femoral shaft subperiosteal width (indicator of bone strength) in postmenopausal women compared to placebo» (Forbes et al., 2018), and BMD is blind to such geometric strength gains; (iii) fracture, the patient-important endpoint, which no trial measured. The evidence base is also thin and mostly moderate-risk (only 1 of 5 trials low risk of bias), so the null is precise but lightly warranted -> Surrogate Outcomes.
- BMD is a surrogate for fracture — so the bone case is doubly limited. Creatine has neither a BMD effect nor any fracture data, so even the surrogate->outcome transmission step is untestable here. This contrasts with the exercise bone lever, where the loading effect on BMD is small-positive AND a fracture MA now closes the transmission step -> Exercise and Bone Mineral Density. a creatine-on-fracture endpoint — a genuine G-gap.
- Symmetric-standards note (a friendly-analyst null). Forbes is by the creatine research community itself (Candow and Chilibeck ran two of the positive individual trials pooled here), so a null from these analysts, who had every prior reason to find benefit, carries more weight against the bone-benefit belief than a skeptic’s would — the standards cut the same in both directions. (inferred from Forbes et al., 2018)
The source caveat — a conflicted consensus, held to the recommendation-summary bar
(inferred from Kreider et al., 2017)
- It is a position stand, not a fresh meta-analysis. An ISSN consensus is a recommendation-tier document — authoritative for what a sports-nutrition body advises, but its effect estimates are borrowed from primary studies it cites, mostly narratively rather than pooled. Where it asserts a magnitude (the 10-20% performance gain, the elderly MAs), the underlying trial/MA — not this stand — is the evidence; the wiki holds only the stand.
- Industry conflict is explicit and heavy. «Support to prepare this manuscript was provided by the Council for Responsible Nutrition» (a supplement-industry trade association) (Kreider et al., 2017), and the authors disclose co-founding the ISSN, industry creatine-research grants, company advisory roles, product royalties, and supplement patents (Kreider et al., 2017). Symmetric standards apply in both directions: the conflict does not falsify the well-corroborated ergogenic/safety claims (which cross-body consensus supports), but it does warrant discounting the document’s enthusiasm on the unestablished clinical claims, exactly as done above.
Layer-1 placement — an ergogenic adjunct, NOT a disease-prevention supplement
For a healthy adult, creatine is a small, well-tolerated adjunct to resistance training on surrogates (performance, lean mass, strength) — it earns its place only after the training itself is in place, and ranks with the other adaptation refinements, not the big rocks -> Layer 1 - Ranking Interventions for a Stratum. Scope firewall: this is a performance/adaptation lever and belongs apart from the general-population disease-prevention supplement question, which is a null-to-harmful lever on hard CV/ cancer/fracture/mortality endpoints -> Vitamin and Mineral Supplements for Disease Prevention. Creatine does not transport into that page as a “supplement that works”: it works on different outcomes (surrogate performance, not disease prevention), so pooling it with the multivitamin/omega-3 nulls would be a category error (inferred from Kreider et al., 2017; Manson et al., 2019). Its rank rises with the elderly/sarcopenia stratum, where the surrogate is itself patient-important.