Creatine, not quite the perfect molecule...

Creatine is the most studied supplement there is: hundreds of published trials, five decades of hindsight, a safety profile no other molecule in the category comes close to. That abundance has a side effect. When there is that much literature, you can always find a study to back any claim.

Here are the four claims currently attached to it, sorted by what the data actually support:

1. Muscle: benefits established and secure

Strength gains in the 5 to 15% range and one to two extra kilos of lean mass, alongside resistance training. The mechanism is known: the phosphocreatine system regenerates ATP, which buys a few more repetitions per set. Over months, those repetitions are the difference.

On safety, the International Society of Sports Nutrition position stand documents intakes up to 30 g per day for five years in healthy people with no toxicity signal. Few molecules on the shelf can say the same.

The dose, though, depends on your weight

The «5 g a day» repeated everywhere is a shortcut, and a shortcut is not always the best route to take. The maintenance dose that actually makes sense (no pun intended on my surname, which is pronounced exactly that way) is 0.03 g per kilo per day. At 60 kg, 5 g overshoots by 60%; at 95 kg, it undershoots.

2. Depression: two results, both true

This is the topic of the moment, and where sorting gets interesting. The meta-analysis published in 2025 in the British Journal of Nutrition pools 11 randomised trials and 1,093 participants. Verdict: a standardised effect of −0.34, equivalent to 2.2 points on the 17-item Hamilton scale. The commonly accepted threshold for a clinically relevant difference is 3 points.

In other words, the average effect falls short of the point where a patient feels a difference. The authors say it plainly: the benefit may be trivial, or null.

And yet the same analysis finds a remission rate multiplied by 3.6, with a confidence interval that clearly excludes no effect. Both numbers are correct. They do not contradict each other: a modest average can hide a minority of strong responders drowned in a majority of non-responders. That is exactly what the high heterogeneity between trials suggests.

The honest reading fits in one sentence: a serious lead, a possible adjunct to ongoing treatment, never a replacement. Depression is treated with a health professional.

3. Cognition: nothing below 5 g

Muscle saturates before the brain. That is why many people notice no cognitive effect: at 3 g per day, brain concentration barely moves. The threshold where cognitive benefits appear in the literature sits around 5 g per day, and studied doses can go up to 0.1 g/kg.

Second point, and the more important one: the effect shows mostly when the system is under strain. Sleep deprivation, prolonged mental load, age above 60. In a rested, healthy adult, the expected effect is small. That is not disappointing, it is logical. You do not fill a deficit that does not exist.

4. Cancer: not a claim, a caution zone

This is where I depart most from the ambient discourse, and it is better said directly: there is no demonstrated clinical benefit, and there is an unfavourable signal.

• In preclinical models the results are contradictory: suppression of tumour growth in some models, promotion of colorectal and breast cancer metastasis in others.
• In humans, published trials are negative on their endpoints: no muscle mass gain, no quality of life improvement, no added benefit over resistance training alone in treated patients.
• In January 2026, the Oncology Nursing Society urged clinicians to weigh risks and benefits carefully, explicitly mentioning a possible contribution to metastatic spread.

The trap almost nobody mentions

Creatine breaks down into creatinine. Regular supplementation therefore raises serum creatinine, with no kidney involved. That is not the problem. The problem is that creatinine is used to estimate kidney function, and that it counts among the nine markers of the biological age calculation.

Concrete consequence: you take creatine, your panel comes back with higher creatinine, your biological age worsens, and you conclude something is wrong. Nothing is wrong. It is a known, expected measurement artefact.

This is exactly the kind of conflict a single-purpose app never sees: your creatine intake is logged in one app and your blood panel in another, without those data ever being crossed. Helix tracks both and catches the consequence to adjust your biological age accordingly.

The protocol, in five points

1. 1

   Monohydrate, nothing else

   It is the form tested in nearly every trial, bioavailability is close to 100%, and it is the cheapest on the shelf. Premium forms have not proven themselves.

2. 2

   0.03 g per kilo, every day

   Rest days included. What produces the effect is saturation, not the pre-workout dose.

3. 3

   Aim for 5 g if the brain is the target

   Below that, you saturate muscle and not much else.

4. 4

   Flag it before a blood panel

   Tell the doctor reading your results that you supplement, or stop a few weeks before if you want an unbiased creatinine.

5. 5

   Give it a month before judging

   Saturation takes three to four weeks. Anything felt after three days is not the creatine.

And if you run something

Creatine's cognitive benefit shows up mostly under sleep deprivation. That cuts both ways for anyone stacking heavy weeks: yes, the molecule absorbs part of the cost of a bad night; no, it does not replace it. A shock absorber is still a shock absorber, it does not repair the road.

The useful reading is rather this one: of the four claims examined, the only one that holds without reservation requires resistance training alongside it. No molecule on this list produces an effect on its own. That is true of creatine, and true of the rest of your stack.