Effectiveness of Creatine in Metabolic Performance: A Systematic Review and Meta-Analysis.

The researchers tested the effect of oral creatine monohydrate supplementation on metabolic performance outcomes. The intervention was creatine monohydrate at various doses (most commonly 20 g/day for 5–7 days loading phase, followed by 3–5 g/day maintenance). Comparators were placebo (usually maltodextrin or dextrose) or no supplementation. The primary outcome was anaerobic power output measured via Wingate anaerobic test (WAnT). Secondary outcomes included blood lactate concentration after exercise, time to exhaustion during high-intensity cycling or running, and ratings of perceived exertion (RPE). The meta-analysis synthesised data from 10 randomised controlled trials (RCTs) that met inclusion criteria.

The meta-analysis included data from 10 RCTs with a total of 237 participants (range per study: 14–40 participants). Participants were predominantly healthy, recreationally active to moderately trained adults aged 18–35 years. The majority were male (~85%), with two studies including female participants. Exclusion criteria across studies included: pre-existing kidney or liver disease, diabetes, use of medications affecting muscle metabolism, and current use of other sports supplements. Baseline fitness levels varied from sedentary to moderately trained (VO₂max range: 38–52 mL/kg/min). No studies included elite athletes or clinical populations. All studies were conducted in university or laboratory settings.

• Anaerobic power output: Measured using the Wingate anaerobic test (WAnT) — a 30-second maximal cycling sprint against a fixed resistance. Outcomes included peak power (watts), mean power (watts), and fatigue index (percentage decline). Units: watts per kilogram of body weight (W/kg) for normalisation.
• Blood lactate concentration: Measured via finger-prick blood samples taken 3–5 minutes post-exercise, analysed using enzymatic-amperometric lactate analysers (e.g., Lactate Pro, YSI 2300). Units: mmol/L.
• Time to exhaustion: Measured during constant-load cycling or running at 70–85% of VO₂max until volitional failure. Units: seconds or minutes.
• Ratings of perceived exertion (RPE): Borg CR-10 scale (0–10, where 0 = nothing at all, 10 = maximal exertion) or Borg 6–20 scale (6 = no exertion, 20 = maximal exertion).
• Body composition (secondary): Some studies measured lean body mass via dual-energy X-ray absorptiometry (DXA) or bioelectrical impedance analysis (BIA).

Study design: This is a systematic review and meta-analysis of randomised controlled trials. The authors searched PubMed, Scopus, Web of Science, and Cochrane Library up to June 2023. Inclusion criteria were: (1) RCTs with a placebo or control group, (2) creatine monohydrate as the sole intervention, (3) metabolic performance outcomes measured, (4) adult human participants, (5) published in English. Exclusion criteria included: non-randomised designs, animal studies, studies combining creatine with other supplements (e.g., beta-alanine, caffeine), and studies with <7 days of supplementation.

Risk of bias assessment: The authors used the Cochrane Risk of Bias tool (RoB 2) for RCTs. Of the 10 included studies, 4 were rated as low risk of bias, 5 as some concerns (primarily due to unclear allocation concealment or lack of blinding of outcome assessors), and 1 as high risk (due to incomplete outcome data). Publication bias was assessed via funnel plot asymmetry and Egger's test — no significant asymmetry was detected (p = 0.34), though the small number of studies limits power.

Statistical approach: A random-effects meta-analysis was used (DerSimonian-Laird method) because heterogeneity was expected across studies due to differences in dosing protocols, participant fitness, and outcome measures. Effect sizes were reported as standardised mean differences (SMD) with 95% confidence intervals (CI). Heterogeneity was quantified using I² statistics (0% = no heterogeneity, 100% = maximal). Subgroup analyses were planned for: (1) loading vs. no-loading protocols, (2) trained vs. untrained participants, (3) duration of supplementation (<4 weeks vs. ≥4 weeks). Sensitivity analyses were performed by removing one study at a time.

What this design can and cannot prove:

• Can prove: That creatine supplementation, on average across the included studies, produces a statistically significant and clinically meaningful improvement in anaerobic power and lactate clearance compared to placebo. The meta-analytic approach increases statistical power and generalisability beyond any single study.
• Cannot prove: Causality at the individual level — meta-analyses give average effects, not individual responses. Cannot determine optimal dose or duration because studies varied widely. Cannot rule out that effects are driven by a subset of "responders" (genetic or dietary factors). Cannot assess long-term safety or effects beyond 8 weeks because most studies were short (2–8 weeks). Publication bias remains possible despite non-significant Egger's test, as small negative studies may be unpublished.

Major methodological weaknesses:

• Small number of studies (k=10) and participants (n=237) — limits precision and power for subgroup analyses.
• Heterogeneity was moderate-to-high for some outcomes (I² = 45–68% for anaerobic power), suggesting true differences between studies that are not fully explained.
• Short duration — most studies lasted 2–4 weeks; no data on effects beyond 8 weeks.
• Industry funding — 3 of 10 studies disclosed funding from supplement manufacturers (e.g., Creapure®), which is associated with larger effect sizes in supplement research.
• Lack of standardised dosing — loading protocols varied (15–25 g/day for 5–7 days), and maintenance doses ranged from 2–10 g/day.
• No assessment of dietary creatine intake — participants' baseline meat consumption (which provides ~1–2 g creatine/day) was not controlled, potentially diluting effects in vegetarians who have lower baseline muscle creatine.

Primary outcome — Anaerobic power output (Wingate test):

• Creatine supplementation significantly increased peak power compared to placebo: SMD = 0.48 (95% CI: 0.28 to 0.68, p < 0.001), I² = 52%.
• Mean power also increased: SMD = 0.41 (95% CI: 0.21 to 0.61, p = 0.002), I² = 45%.
• In absolute terms, peak power increased by approximately 6–8% (range: 4–12% across studies), equivalent to ~30–50 watts in a 70 kg individual.
• Fatigue index (percentage decline in power over 30 seconds) was not significantly changed: SMD = -0.12 (95% CI: -0.38 to 0.14, p = 0.36), I² = 18%.

Secondary outcome — Blood lactate concentration post-exercise:

• Creatine significantly reduced blood lactate levels after high-intensity exercise: SMD = -0.39 (95% CI: -0.59 to -0.19, p < 0.001), I² = 38%.
• Absolute reduction: approximately 1.5–2.5 mmol/L lower than placebo (baseline lactate typically 10–14 mmol/L post-Wingate).
• This effect was more pronounced in studies using a loading protocol (SMD = -0.52) compared to no-loading (SMD = -0.21), though the subgroup difference was not statistically significant (p = 0.12).

Secondary outcome — Time to exhaustion:

• Creatine increased time to exhaustion during constant-load exercise: SMD = 0.35 (95% CI: 0.12 to 0.58, p = 0.008), I² = 41%.
• Absolute increase: approximately 45–90 seconds longer (range: 30–120 seconds across studies), representing a ~8–15% improvement.

Secondary outcome — Ratings of perceived exertion (RPE):

• No significant difference between creatine and placebo: SMD = -0.08 (95% CI: -0.30 to 0.14, p = 0.48), I² = 12%.
• This suggests that despite improved performance, participants did not perceive the effort as easier.

Subgroup analyses:

• Loading vs. no-loading: Studies using a loading phase (20 g/day for 5–7 days) showed larger effects on peak power (SMD = 0.56 vs. 0.32) and lactate reduction (SMD = -0.52 vs. -0.21) compared to studies using only maintenance dosing (3–5 g/day from day one). However, the difference was not statistically significant (p = 0.09 for power, p = 0.12 for lactate).
• Trained vs. untrained: Effects on anaerobic power were larger in untrained participants (SMD = 0.62) compared to trained participants (SMD = 0.33), p = 0.04 for subgroup difference. This suggests a ceiling effect — those with higher baseline fitness have less room for improvement.
• Duration: Studies ≥4 weeks showed slightly larger effects (SMD = 0.52) than studies <4 weeks (SMD = 0.38), but this was not significant (p = 0.21).

Sensitivity analyses: Removing the one high-risk-of-bias study did not change the direction or significance of any outcome. Removing industry-funded studies reduced the effect size for peak power from SMD = 0.48 to SMD = 0.39, but it remained significant (p = 0.02).

• Anaerobic power: A ~6–8% increase in peak power output during a 30-second all-out sprint. For context, this is roughly equivalent to the improvement seen after 4–6 weeks of sprint interval training in untrained individuals. In practical terms, if your baseline peak power is 700 watts, creatine might push it to ~750 watts — enough to notice in a cycling sprint or a set of heavy squats.
• Blood lactate reduction: A drop of ~1.5–2.5 mmol/L post-exercise. This is meaningful because lactate accumulation is associated with muscle fatigue and the "burn" sensation. A reduction of this magnitude might allow you to complete one or two additional high-intensity repetitions before failure.
• Time to exhaustion: An extra 45–90 seconds of exercise at 80% of VO₂max. This is a moderate effect — noticeable in a timed run or cycling test, but not transformative. For a 10-minute run, this represents a ~10–15% improvement.
• Overall: The effects are small-to-moderate by standardised effect size conventions (SMD 0.35–0.48 = "small to moderate"). This means that while the effects are statistically robust and likely real, they are not massive — creatine is not a "magic pill" that will double your performance. The effects are most pronounced in untrained individuals and during short-duration, high-intensity efforts (30 seconds to 3 minutes).

What the authors acknowledge:

• Small number of studies and participants, limiting generalisability.
• Moderate-to-high heterogeneity for some outcomes, suggesting true differences between studies.
• Short duration of most studies (2–8 weeks), with no long-term safety or efficacy data.
• Lack of standardised dosing protocols across studies.
• Potential publication bias despite non-significant Egger's test.
• Inability to assess individual responder vs. non-responder effects.

What a critical reader would note:

• Industry funding: 3 of 10 studies were funded by creatine manufacturers. In supplement research, industry-funded studies are 2–4 times more likely to report positive results. The sensitivity analysis showing reduced effect size when excluding these studies is concerning.
• No dietary control: Baseline creatine intake from meat/fish was not measured or controlled. Vegetarians (who have ~20–30% lower baseline muscle creatine) typically show larger responses to supplementation. If studies included mixed diets, the average effect may be diluted.
• No assessment of muscle creatine levels: None of the included studies directly measured muscle creatine content via biopsy or magnetic resonance spectroscopy (MRS). This means we cannot confirm that the observed effects are actually due to increased muscle creatine stores — the proposed mechanism.
• Lack of blinding verification: While studies were described as double-blind, none reported testing whether participants could guess their group assignment. Creatine has a distinct taste and can cause water retention (1–2 kg weight gain in the first week), which may unblind participants.
• Outcome selection bias: The meta-analysis focused on metabolic performance (power, lactate, time to exhaustion). It did not include other commonly reported creatine effects (e.g., muscle hypertrophy, strength gains, cognitive effects), which may have different effect sizes.
• No assessment of adverse effects: Creatine is generally safe, but some individuals experience gastrointestinal distress (bloating, diarrhoea) at loading doses. The meta-analysis did not systematically report adverse events, so the risk-benefit ratio is unclear.
• Population limits: Nearly all participants were young, healthy, and predominantly male. Effects in women, older adults, or clinical populations (e.g., metabolic syndrome, sarcopenia) cannot be inferred.

For someone running their own n=1 experiment:

What to test

• Intervention: Creatine monohydrate (not other forms like creatine ethyl ester or buffered creatine, which have less evidence). Use a standard protocol:
  • Loading phase: 20 g/day (split into 4 doses of 5 g each) for 5–7 days.
  • Maintenance phase: 3–5 g/day thereafter.
• Comparator: Placebo (maltodextrin or dextrose, matched for taste and appearance). Ideally, use a blinded protocol where you and a helper prepare identical-looking capsules or powders.
• Outcome: Choose ONE primary outcome to avoid cherry-picking. Good options:
  • Anaerobic power: 30-second Wingate test on a stationary bike (or a 400-metre sprint if no bike available). Measure peak power (watts) and mean power.
  • Blood lactate: If you have access to a lactate meter (e.g., Lactate Pro), measure finger-prick lactate 3 minutes post-exercise.
  • Time to exhaustion: Constant-load cycling or running at 80% of your estimated max heart rate until failure.

Minimum meaningful duration

• Loading phase: 5–7 days to saturate muscle stores.
• Total experiment: At least 3–4 weeks (1 week loading + 2–3 weeks maintenance). Effects on anaerobic power appear within 5–7 days but may increase slightly over 2–4 weeks.
• Washout period (if crossover design): At least 4 weeks between conditions, as muscle creatine levels take ~4 weeks to return to baseline after stopping supplementation.

What to measure (specific metrics)

• Primary metric: Peak power (watts) or mean power (watts) during a 30-second all-out effort. Normalise to body weight (W/kg) to account for water retention weight gain.
• Secondary metrics:
  • Blood lactate (mmol/L) at 3 minutes post-exercise.
  • Time to exhaustion (seconds) during constant-load exercise.
  • Body weight (kg) — track daily to account for water retention (expect 1–2 kg gain in first week).
  • Subjective energy/mood (1–10 scale) — to detect any placebo or nocebo effects