Creatine for adults over 55: what 8 trials actually show

Creatine is the one sports-nutrition ingredient with a European health claim written specifically for older adults. Commission Implementing Regulation (EU) 2017/672 authorised the wording “Daily creatine consumption can enhance the effect of resistance training on muscle strength in adults over the age of 55”, on conditions: 3 g of creatine a day, combined with resistance training at least three times a week at 65–75% of one-repetition maximum, for several weeks. That claim is nine years old. The evidence behind it has kept accumulating since.

A systematic review and meta-analysis published in European Review of Aging and Physical Activity in December 2025 pooled the trials that sit closest to that claim: creatine plus supervised resistance training versus placebo plus the same training, in adults with a mean age of 50 or over. It is a small synthesis — eight trials — and it reaches a split verdict. Lower-body strength and lean tissue mass improved. Upper-body strength did not. Everything that improved, improved by a little. For anyone specifying an ingredient for this market, the split is more useful than a headline would be.

What the meta-analysis measured

Liu and colleagues searched eight databases on 27 February 2025, limited to 2015–2025, and retrieved 2,171 records. After removing 1,372 duplicates, 799 titles and abstracts were screened, 16 full texts assessed and 8 randomised controlled trials retained, totalling 482 participants. Populations were untrained healthy older men and postmenopausal women. All trials used supervised resistance training two or three times a week. Durations ran from 8 to 104 weeks.

Dosing is worth pausing on, because it is not the claim dose. Six of the eight trials dosed by body mass at 0.1 g/kg/day — roughly 7 g a day for a 70 kg adult. One used 0.14 g/kg/day. One used a flat 5 g a day. Not one trial in this pool dosed at the 3 g/day written into the authorised claim. Strength was measured as one-repetition or ten-repetition maximum on chest press and leg press; lean tissue mass by DXA. Adverse events were documented in all eight trials and were limited to transient gastrointestinal discomfort and mild cramping, with none serious.

The numbers, outcome by outcome

Pooled with random-effects models and reported as standardised mean differences, the results are modest and uneven.

Lower-body strength improved by SMD 0.29 (95% CI 0.00–0.57; p = 0.05; I² = 40%). That is a borderline result in the literal sense: the interval's lower limit is zero. Lean tissue mass improved by SMD 0.27 (95% CI 0.02–0.53; p = 0.03; I² = 18%), the cleanest of the three with the least heterogeneity. Upper-body strength did not separate from placebo: SMD 0.25 (95% CI −0.03–0.54; p = 0.08; I² = 42%). On the authors' own effect-size scale, anything from 0.20 to 0.59 is “small”.

The duration subgroups are the paper's main contribution. In trials of 32 weeks or less, upper-body strength reached SMD 0.45 (0.14–0.77) and lean tissue mass SMD 0.57 (0.19–0.96), both with no measurable heterogeneity. In trials of 52 weeks or more, the same outcomes collapsed to 0.01 (−0.43–0.41) and 0.08 (−0.17–0.34). For lower-body strength the authors tested the interaction and found no significant difference between the two duration strata, so the pattern there is suggestive rather than established. A robust variance estimation re-analysis, run to handle multiple correlated effects per trial, returned a pooled Hedges' g of 0.327 (95% CI 0.063–0.591; p = 0.017) — the same direction and the same modest size.

Where the 2025 syntheses disagree

Three syntheses published in 2025 asked overlapping questions about older adults and returned different answers on strength. A network meta-analysis in Frontiers in Nutrition compared protein, creatine and HMB against resistance training alone across 19 trials in 997 adults aged 60 and over. Protein improved strength (SMD 0.45, 95% CI 0.20–0.69). Creatine did not (SMD 0.03, 95% CI −0.35–0.42). HMB did not (SMD −0.22, 95% CI −0.57–0.12). But for muscle mass the ranking inverted: creatine gave the largest effect of the three (MD 2.18 kg, 95% CI 0.92–3.44), ahead of protein (MD 0.37 kg, 95% CI 0.04–0.70), and HMB gave nothing (MD 0.05 kg, 95% CI −0.33–0.44).

A third synthesis, covering 69 trials and 1,937 participants of all ages, reported creatine gains on bench or chest press (+1.43 kg) and squat (+5.64 kg) overall, then found in its age subgroup that these held in younger adults and not in older ones. Its abstract gives those subgroup estimates without confidence intervals, so they are directional here, not quantitative.

These are not three independent datasets. The first two share primary trials, which makes the disagreement on strength more interesting, not less: the same underlying evidence supports different conclusions depending on comparator, age threshold and model. The one claim that survives all three is the one about mass. The same instability shows up outside the older-adult literature: in a 2026 network meta-analysis of 35 trials in trained athletes, creatine ranked first for strength and protein first for endurance — the opposite assignment to the one reported here, in a much younger population.

What it means for a formulator

Three consequences for anyone designing a product for this population.

The evidence dose and the claim dose are different numbers, and the claim dose wins. The trials pooled here mostly delivered 7–10 g a day by body mass. The authorised claim is fixed at 3 g a day. A product built to carry that claim has to deliver 3 g at the end of its shelf life, not at packing — and creatine converts to creatinine in solution, in low-pH matrices and under heat, so the overage needed to land on 3 g at expiry is a specification decision taken before the artwork is signed off. Specify assay at release and at expiry, separately, and state which one the certificate of analysis reports. CREAPRIME™ is supplied at ≥99.9% creatine monohydrate with creatinine ≤30 ppm and a batch-by-batch HPLC assay to USP <621>, so the starting point for that calculation is a measured number rather than a nominal one. That degradation product matters a second time in the clinic rather than the plant: creatine raises serum creatinine by about 0.14 mg/dL and lowers creatinine-based eGFR while directly measured filtration does not change, which is worth knowing for a product aimed at a population screened routinely with that marker.

Every trial in this pool used creatine monohydrate. No other salt or form appears. That means the evidence does not transfer to a different form by analogy, and it means the assay method has to separate creatine from creatinine — a total-creatine assay that does not resolve the degradation product will pass a batch that has already partly converted. Write the method and the per-batch acceptance limit into the purchase specification, not just a purity percentage.

The duration finding is a compliance problem before it is a biology problem. The benefit concentrated in trials of 32 weeks or less, under supervision. In the real world an adult over 55 has to take the product daily for months without a trial coordinator. That pushes serving format, dosing accuracy and palatability from marketing concerns to efficacy concerns. It also pushes batch continuity: a thirty-week regimen sold as a programme needs enough material from a single qualified source to cover the run, which is where pallet-scale supply and a per-batch certificate matter more than a sample does. If the formulation pairs creatine with protein — reasonable, given that protein carried the strength signal in the network analysis — the LACTOPRIME™ systems are specified on the same per-batch basis. Our minimum order is one pallet, 1,000 kg, and every licence carries a publicly verifiable code; technical documentation is available on request.

Limits of this evidence

The meta-analysis is small. Eight trials and 482 participants is not enough to assess publication bias — the authors deliberately did not attempt a funnel plot or Egger's test, since both lose power below ten studies, and said so. Risk of bias was rated unclear overall in roughly 75% of the included trials, mostly because allocation concealment was reported inadequately. Six trials had attrition above 15% and four did not report intention-to-treat analysis; the authors note both may inflate the efficacy estimates. GRADE certainty was moderate for upper- and lower-body strength and low for lean tissue mass, downgraded for risk of bias and, in the case of mass, for imprecision from a sample under 400.

Two further gaps matter. No sex-stratified analysis was performed, despite six trials enrolling postmenopausal women and five enrolling men. And the inclusion threshold was a mean age of 50, below the over-55 population the EU claim is written for — the pooled estimate describes a slightly younger group than the label does. The review was registered on PROSPERO (CRD420251018405). The postmenopausal subgroup has pooled evidence of its own, reviewed separately rather than here: seven trials in 608 postmenopausal women. Women at the other end of the age range have theirs too: 21 trials and 424 training women.

Taken together: creatine added to resistance training in older adults produces a small, real effect on lean tissue mass, a borderline effect on lower-body strength, and nothing demonstrable on upper-body strength. The honest version of the commercial case is not that creatine transforms ageing muscle. It is that a measured, well-characterised monohydrate delivering the claim dose at expiry is the difference between a product that can legally say something and one that cannot.