Why You Should Be Skeptical About Your Wrist-Based Heart Rate

Published January 5, 2022 01:33AM

Every week or two, I’ll open an athlete’s training log and it seems like a blood-curdling scream leaps off the page.

As Edgar Allan Poe could tell you, little is more disconcerting than a screwed-up heart beat. And sometimes these athletes’ hearts seem very screwed up, a telltale sign of a major issue. Maybe they sustain 190 heart rate at a moderate effort. Or maybe it peaks at 120 on a sprint. When I first saw these issues a few years ago, worried about arrhythmia or other health crisis, I would ask them to go to the doctor immediately.

But, then, I would look at the data and notice discrepancies. Lots of studies came out that saw discrepancies too. There seemed to be a problem of faulty measurement, rather than faulty hearts. So, now, I have a standard one-question response.

Simply put, for some athletes, wrist-based heart rate is not accurate during intense activity like running, though it varies a ton. And many athletes have no idea. Let’s dig down into some of the data.

Heart-Rate-Measurement Basics

Most wrist-based heart-rate monitors rely on photoplethysmography, a word that I assume is real. Photoplethysmography involves using light to measure blood flow. That’s why your watch may have a bright LED light that shines against your skin. Light refracted off the blood flowing beneath the surface of your skin is fed into algorithms that use it to spit out a heart-rate number.

Chest straps, meanwhile, use sensors on the strap to record signals from the heart beat. This process requires some moisture, which is why you may notice readings that are off when you start a run and haven’t worked up a sweat.

Problems Wrist Monitors Face

The chest strap has the advantage of being a semi-direct measurement of heart rate. A good chest strap will be close to 100% accurate. Which makes sense, because it’s essentially reaching into your chest cavity and recording each beat with a clipboard.

As we’ll see later, many wrist monitors are not as accurate. For some, it could be as simple as not being placed correctly on the wrist, since being too loose generate a faulty reading, and being too tight can impede blood flow.

Other athletes may have physiology that is not as conducive to the technology, possibly due to locations of blood vessels or bodyfat percentage. On top of that, vigorous exercise jostles everything around, like turbulence on an airplane. You might be looking at a beautiful sunset out the window when you hit a rough patch, and now you’re looking at the sky, now the clouds, now the sun and now you’re throwing up in a bag.

Wrist-Monitor Accuracy Data

I currently have 18 windows open to different studies on wrist-based heart rate conducted since the start of 2017, and my computer fan just turned on loudly in protest. If my browser had a heart, it would currently be at VO² max.

Most of the data from the studies has overlap, even if the conclusions vary. A January 2017 research letter in the Journal of the American Medical Association Cardiology had 50 healthy adults test four different wrist monitors and a chest strap at rest, 2 mph, 3 mph, 4 mph, 5 mph and 6 mph. While the chest strap had a 0.99 correlation with electrocardiograph readings, the wrist monitors varied between 0.83 and 0.91 correlation. That study had a conclusion that we’ll see a lot: “In general, accuracy of wrist-worn monitors was best at rest and diminished with exercise.”

An August 2017 study in the Medicine & Science in Sports & Exercise Journal had even more disheartening findings. Fifty athletes spent time on the treadmill, elliptical and stationary bike, and correlation varied from 0.75 to 0.92. As that study concluded: “Electrode-containing chest monitors should be used when accurate HR measurement is imperative.”

“That sounds pretty good,” someone might say. “It’s not imperative to be exactly right, just close.” That perspective is backed up by the conclusion of some studies. For example, a 2018 study in BMC Sports Science, Medicine, and Rehabilitation came to the conclusion that wrist-based heart rate is “acceptably close” across a broad range of activities. Looking a bit deeper into the data, that study found that during running, the heart rate numbers were within 10 beats per minute of actual heart rate 95% of the time, with around three beats per minute average error. In other words, on a one-hour run, a few minutes will be spent with a heart rate monitor giving a number that is 10+ beats off.

You think that was a lot of studies? Well you ain’t seen nothing yet. A 2018 study in the journal Digital Health found absolute error of 1% to 8% depending on device and intensity level. A 2017 conference paper found that the tested watches accurately measure up to 89% of beats accurately at rest, and less during activity. A 2019 study in the Journal of Sport Sciences found error rates between 2.4% and 13.5%.

Here’s the most fascinating one to me. A 2018 study in the BMJ Open Sport and Exercise Journal found absolute error values of 3.3% to 6.2%, but using a “10% equivalence zone,” concluded that the watches were accurate. You can see why that is a problem for athletes if you think about using the heart-rate monitor to stay below aerobic threshold (AeT). Imagine a hypothetical athlete with an AeT of 150. At 150 heart rate or below, they know they can achieve the desired training stimulus, or excel in an ultra race. A 5% error rate could mean that their heart rate is actually 157 (which would lead to lack of recovery in training or a bonk in racing).

But here’s the big problem—the errors are not consistent. If it was always 5% low or 5% high, we could use that information (you’re probably having flashbacks to the accuracy v. precision distinction from high-school science class). Instead, it’s all over the place, and since 5% is an average, sometimes it might read 190 or something that makes an athlete write a panicked entry in their training log.

It’s kind of like Michael Cohen testimony. First it says one thing, then it says another, so even if it’s right one time, how can you trust it? In this analogy, I’m not sure if I’m CSPAN or Alexandria Ocasio-Cortez.

Put It All Together

When it comes to wrist-based heart rate, it’s not an obvious answer for every athlete. A 2018 commentary in the Journal of Medical Internet Research pointed out that there is no consensus, partially due to lack of uniformity in evaluation (the same data can be used to reach different conclusions based on methodology, as we were talking about with the “10% equivalence zone” above). The gold standard of product testing, D.C. Rainmaker, has found some optical sensors that work well for him (with variation), and his tests are as rigorous as any study protocol. Anecdotally, some athletes can use certain wrist monitors and return with beautiful heart-rate graphs that track effort, while others can use the same watch and do the same workout and return with a garbly mess of incomprehensible noise.

In addition, every watch is different, and some are more accurate than others. The technology is constantly improving, and it’s possible that newer, high-quality watches have overcome some of the technological hurdles uncovered by the studies. If I were a betting person, I’d guess that wrist-based heart rate will be effective relatively soon.

If you want to do some home testing, and you think your watch might be right, do 6 x 1 minute intervals hard with 2 minutes easy recovery on your next run (or similar hard effort). Does your watch return heart-rate peaks that are as stunning as the Matterhorn, lining up with how you feel? That’s good, it probably works. Calibrate it with a chest strap to be sure.

However, if you’re like some of the athletes I have seen, the heart-rate graph will look like it was created by a random-number generator. If that’s you, there’s a good chance your wrist monitor is still accurate at rest and during certain activities, so it can still be useful. But using it for intense running would be like entrusting your timekeeping to one of those broken clocks that is still right twice a day.