What the creatine trials measured instead of resting heart rate

Someone starts creatine, and three weeks later their watch shows a resting heart rate two beats above where it used to sit. Is that the creatine? The answers in circulation split in two — creatine is among the best-studied supplements there is, so no; or a study found it raises heart rate, so yes. Both answer a question the trials did not ask.

Safety and effect are different claims. A compound can be entirely safe and still move a continuous measure by a few beats a minute, and establishing the first tells you nothing about the second. So the useful question is narrower: what has anyone actually measured?

The direct evidence is one pilot study

A 2024 double-blind randomised crossover trial gave twelve sedentary older adults 20 g of creatine monohydrate a day for five days, then 5 g a day for twenty-three days, with a four-week washout between arms [1]. Its primary endpoint was flow-mediated dilation. Heart rate appears once, in one sentence:

That is the entire heart rate result — no values, no interval, no test statistic. The authors are explicit about the standing of anything that was not their primary endpoint:

A null on an unpowered secondary outcome in twelve people, reported without figures, is not evidence that creatine leaves resting heart rate alone. It is the absence of evidence either way.

What has been pooled is not resting heart rate

The one meta-analysis that pools heart rate at all is a 2024 review of creatine in swimmers — 17 studies and 361 subjects [2]. Two of those studies contributed heart rate, both measured after exercise.

The interval crosses zero comfortably (p = 0.61, I² = 2%). But read what it is: a standardised mean difference, from two trials, of heart rate after swimming, in trained swimmers. Not beats per minute, and not at rest. The review scored methodological quality with the PEDro scale, averaging 8.41 across included studies, and reported no GRADE certainty for this outcome.

The nearest thing to autonomic data is a 2026 crossover in ten resistance-trained men given 0.3 g per kg per day for three days, with heart rate variability recorded around each testing session [3]. The pre-to-post change in RMSSD was larger on creatine in the first session (p = 0.015, d = 2.99), but the main effect of condition was not significant (p = 0.154) and the difference had gone by the second session. Ten men, several indices, measured around exercise rather than at rest.

The cardiac literature answers a different question

There is a Cochrane review of creatine and creatine analogues in cardiovascular disease: 11 trials, 1,474 patients with heart failure, ischaemic heart disease or myocardial infarction [4]. Its outcomes were death, total myocardial infarction, hospitalisation for congestive heart failure, change in ejection fraction, and change in systolic and diastolic blood pressure. Heart rate is not among them, and no trial in hypertension was found at all.

Safety analyses count studies, not beats

The largest safety synthesis went through 685 human trials: 12,839 participants across 682 studies took creatine, 13,452 across 652 took placebo [5].

Gastrointestinal complaints came up in 4.9% of creatine studies against 4.3% of placebo studies, and muscle cramping or pain in 2.9% against 0.9%. Across the remaining 33 categories the difference was typically about 0.4 percentage points. That is reassuring, and it is the wrong instrument for this question: the unit of analysis is the study, not the participant — it counts how many papers mentioned a side effect. A trial that never measured resting heart rate cannot report a change in it.

What this does not establish

None of this rules out an effect of creatine on resting heart rate, of any size. There is no pooled estimate in beats per minute, no trial powered to find one, and the single crossover that recorded the number declined to print it. Nor does any of it support the opposite claim. Both positions are assertions with a literature-shaped gap where their evidence should be.

It says nothing either about people with existing cardiac disease, whose only large synthesis was inconclusive on its own chosen endpoints.

Why this is an n-of-1 problem

Resting heart rate is an unusual case. The population literature is nearly silent on it, yet it is among the most heavily measured numbers in personal health — a wearable produces one every night, for years, from the same person. Where the pooled evidence offers two studies of post-exercise heart rate in swimmers, one person can bring two hundred nights.

That abundance does not make the question easy. Resting heart rate moves with sleep, alcohol, illness, training load and room temperature. A 20 g loading dose is hard to mistake for nothing, so blinding takes real effort. And people start a supplement when they are least happy with their numbers, which is exactly when regression to the mean obliges. What would settle it for one person is a randomised on-and-off schedule run over enough periods to separate a few beats from that background. What will not settle it is starting creatine and watching the number.