Does oral creatine supplementation improve strength? A meta-analysis.

The researchers asked whether taking oral creatine supplements (typically a loading dose followed by a maintenance dose) improves maximal strength and power in healthy adults, compared to taking a placebo. They specifically looked at three types of strength outcomes:

**Maximal weight lifted** (the heaviest weight you can lift once, called "one-rep max" or 1RM) for bench press, squats, and arm curls.

**Cycle ergometry sprint peak power** (the highest power output during a short, all-out cycling sprint).

**Isokinetic dynamometer peak torque** (the maximum force produced by a specific muscle group, like your quadriceps, when moving at a fixed speed on a machine).

The comparator was a placebo (usually a sugar pill or an inert powder like maltodextrin). The intervention was creatine monohydrate, taken orally, typically with a loading phase of ~20 grams per day for 5–7 days, followed by a maintenance dose of ~5 grams per day.

The meta-analysis pooled data from **16 studies** involving a total of **~300–400 healthy adults** (exact total N not reported in the abstract, but typical for this era of meta-analyses). The key demographic details:

**Age:** Most subjects were young men under 36 years old. In 7 of the 10 studies that measured maximal weight lifted, all participants were young men engaged in resistance training.

**Sex:** The vast majority were men. Women were rarely included, and the analysis found no evidence of benefit in the few women studied.

**Training status:** Participants were generally healthy, active individuals, many of whom were already engaged in resistance training programs.

**Exclusions:** Studies excluded people with kidney disease, diabetes, or those taking other supplements or medications that could affect muscle function.

The researchers extracted data from each included study using standardised forms. The specific instruments and scales were:

**Maximal weight lifted (1RM):** Measured using standard barbell bench press, squat, and arm curl exercises. The heaviest weight the participant could lift for one complete repetition was recorded in kilograms.

**Cycle ergometry sprint peak power:** Measured using a stationary cycle ergometer (like a spinning bike) during a short (typically 6–30 second) all-out sprint. Power output was recorded in watts.

**Isokinetic dynamometer peak torque:** Measured using a machine (e.g., Cybex or Biodex) that controls the speed of movement. Participants pushed or pulled against the machine's arm at a fixed speed (e.g., 60 or 180 degrees per second), and the peak torque (force) was recorded in Newton-metres.

The primary outcome was the **change from baseline to post-supplementation** in these measures, comparing the creatine group to the placebo group.

**Study design:** This is a **meta-analysis** — a statistical synthesis of results from multiple independent randomised controlled trials (RCTs). The authors systematically searched MEDLINE (1966–2000) and the Cochrane Controlled Trials Register (through June 2001), reviewed conference proceedings and bibliographies, and contacted an expert for unpublished data.

**Inclusion criteria:** Only randomised or matched placebo-controlled trials comparing creatine supplementation with placebo in healthy adults were considered. This is a strong design because randomisation helps ensure that differences between groups are due to the supplement, not pre-existing differences.

**Statistical approach:** The authors used a random-effects model to pool results across studies. This is appropriate when studies vary in their populations, doses, and protocols — it accounts for both within-study and between-study variability. Results are reported as **summary differences** (mean difference between creatine and placebo groups) with 95% confidence intervals (CIs).

**What this design can prove:**

A well-conducted meta-analysis of RCTs provides the strongest evidence for a causal effect — that creatine *causes* an increase in maximal weight lifted, because the individual studies were randomised and placebo-controlled.

The pooled estimate gives a more precise and reliable effect size than any single study.

**What this design cannot prove:**

It cannot prove that creatine works for populations not studied (e.g., women, older adults, elite athletes, or people with medical conditions).

It cannot prove safety — the authors explicitly note that safety remains unproven.

It cannot identify optimal dosing, timing, or duration, because the included studies used different protocols.

It cannot rule out publication bias (studies with null results may be less likely to be published), though the authors attempted to find unpublished data.

**Major methodological weaknesses:**

The search ended in 2001 — this is a very old meta-analysis (published 2002). Many more studies have been done since.

The total sample size across studies is modest (~300–400 people), and most participants were young men.

The analysis did not separate results by sex, age, or training status in a way that allows subgroup recommendations.

The authors did not assess the quality or risk of bias of individual studies in the abstract (though they may have done so in the full paper).

**Bench press (maximal weight lifted):** Creatine supplementation increased the maximum weight lifted by **6.85 kg** (95% CI, 5.24 to 8.47) more than placebo. This was statistically significant (p < 0.05 based on the CI not crossing zero).

**Squats (maximal weight lifted):** Creatine increased the maximum weight lifted by **9.76 kg** (95% CI, 3.37 to 16.15) more than placebo. This was also statistically significant.

**Arm curls (maximal weight lifted):** There was **no significant difference** between creatine and placebo. The effect was small and the confidence interval crossed zero.

**Cycle ergometry sprint peak power:** There was **no significant difference** between creatine and placebo.

**Isokinetic dynamometer peak torque:** There was **no significant difference** between creatine and placebo.

**Primary vs. secondary outcomes:** The primary outcome appears to be maximal weight lifted (bench press and squats), where significant effects were found. The secondary outcomes (arm curls, cycle ergometry, isokinetic torque) showed no benefit.

**Bench press:** The average improvement was about **6.9 kg (15 lbs)** more than placebo. For a young man who can bench press 80 kg, this represents roughly an **8–9% increase** in strength over the course of the training program (typically 4–12 weeks).

**Squats:** The average improvement was about **9.8 kg (21.5 lbs)** more than placebo. For a young man squatting 100 kg, this is roughly a **10% increase**.

**Arm curls:** No meaningful effect — the difference was close to zero.

**Cycle sprint and dynamometer:** No meaningful effect — the difference was close to zero.

In plain English: If you're a young man who starts taking creatine while doing resistance training, you can expect to add about 7 kg to your bench press and 10 kg to your squat over the course of a few weeks, beyond what you'd get from training alone. But you won't see improvements in arm curls, cycling sprints, or machine-based strength tests.

**What the authors acknowledge:**

The safety of creatine remains unproven — the studies were too short and too small to assess long-term risks.

The results only apply to young men under 36 who are already engaged in resistance training.

There is no evidence for improved performance in older individuals or women.

The analysis found no benefit for other types of strength and power exercises (cycle ergometry, isokinetic dynamometry).

**What a critical reader would note:**

**Outdated evidence:** This meta-analysis includes studies published up to 2000. Since then, hundreds of additional studies have been published, many showing benefits in women, older adults, and for other types of exercise. The conclusions here are likely too conservative for current practice.

**Small sample size:** With only ~300–400 total participants across 16 studies, the power to detect small effects or subgroup differences is limited.

**Publication bias:** The authors tried to find unpublished data, but it's possible that null results were underreported, inflating the apparent effect.

**Heterogeneity in protocols:** The included studies used different loading doses (15–25 g/day), maintenance doses (3–10 g/day), training programs (3–12 weeks), and outcome measures. This variability makes the pooled estimate less precise.

**No blinding assessment:** The authors did not report whether participants or assessors were successfully blinded. Creatine can cause weight gain and a distinctive taste, which may unblind participants.

**Industry funding:** Not reported in the abstract, but many creatine studies from this era were funded by supplement companies, which can bias results.

For someone running their own n=1 experiment:

### What to test

**Intervention:** Creatine monohydrate powder. Use a loading phase of **20 grams per day** (split into 4 doses of 5 grams each) for **5–7 days**, followed by a maintenance dose of **3–5 grams per day**.

**Comparator:** For a self-experiment, you can use a placebo (e.g., maltodextrin powder, which looks and tastes similar) for a baseline period, then switch to creatine. Or do a crossover design: 4 weeks placebo, 4 weeks creatine, with a 2-week washout in between.

**Exercise:** Focus on compound lifts like **bench press and squats** — these showed the largest effects. Arm curls, cycling sprints, and machine-based exercises are unlikely to show a benefit.

### Minimum meaningful duration

**Loading phase:** 5–7 days.

**Training + supplementation:** At least **4 weeks** of consistent resistance training while taking creatine. Most studies lasted 4–12 weeks.

**Washout period (if doing crossover):** At least **2 weeks** to clear creatine from muscles (creatine stores return to baseline in about 2–4 weeks).

### What to measure (specific metrics)

**Primary outcome:** One-rep max (1RM) for bench press and squats. Test at baseline, after the loading phase, and at the end of each 4-week block.

**Secondary outcomes:** Body weight (creatine causes water retention, so expect 1–2 kg gain in the first week), training volume (total weight lifted per session), and subjective energy/recovery (rated 1–10 daily).

**Confounders to track:** Sleep quality, total calorie and protein intake, training consistency, and stress levels (all affect strength gains).

### Key confounds to control for

**Training consistency:** Do the exact same workout routine during both placebo and creatine phases. Don't change volume, intensity, or frequency.

**Diet:** Keep protein and total calorie intake constant. Creatine works better with adequate carbohydrate intake (insulin helps transport creatine into muscles).

**Hydration:** Creatine pulls water into muscles, so drink extra water (at least 3–4 litres per day) to avoid dehydration and cramping.

**Timing:** Take creatine at the same time each day, ideally post-workout with a carbohydrate source (e.g., fruit juice) to enhance uptake.

**Expectation bias:** If possible, have someone else prepare your doses so you don't know whether you're taking creatine or placebo. This is hard to do perfectly because creatine causes noticeable weight gain.

### What a positive result would look like

**Bench press 1RM:** An increase of **5–8 kg** (11–18 lbs) more than what you gained during the placebo phase. For example, if you gained 2 kg on placebo, you'd expect to gain 7–10 kg on creatine.

**Squat 1RM:** An increase of **7–12 kg** (15–26 lbs) more than placebo.

**Body weight:** A rapid increase of **1–2 kg** within the first week of creatine (due to water retention), then slower gains from muscle growth.

**Training volume:** You should be able to complete more reps or sets at the same weight, or lift heavier weights for the same number of reps.

**Important caveat:** If you are a woman, over 35, or not doing heavy resistance training (e.g., you're a runner or cyclist), this meta-analysis suggests you may not see any benefit. However, more recent research (post-2002) has found benefits in women and older adults, so your results may vary. The safest approach is to test it yourself with a rigorous n=1 protocol.