STANFORD STUDY: FITNESS TRACKERS GOOD FOR TRACKING HEART RATE BUT NOT CALORIES

If you have an Apple Watch or a Fitbit Surge, their measurement of burned calories may be off.

Researchers from Stanford Medicine discovered six out of seven tested devices could not reasonably measure how many calories were being burned during exercise. The most accurate calorie counter was the Fitbit Surge, which was still off by a median error rate of 27 percent; the worst went to PulseOn, a Finnish fitness tracker, which was off by a whopping 93 percent.

In contrast to the devices’ inconsistency in counting calories, their heart rate measurement was accurate enough to be considered clinical quality. Six of seven devices — the exception was Samsung Gear 2 — recorded a median error rate of less than 5 percent. For both heart rate and calorie counting, the researchers were looking for an error rate under 10 percent.

“For the heart rate, we were pleasantly surprised. We are getting to the point where it’s getting valuable for physicians to follow heart rates on these devices,” said Anna Shcherbina, a co-author of the study. “For energy expenditure, we were surprised again but not in a positive way. We want the error rate below 10 percent, and none of the devices are quite there yet.”

Researchers measured both the heart rate and calorie counting for 29 male and 31 female participants while they were sitting, walking, running and cycling. To accurately measure burned calories in contrast to the devices, the researchers counted the participants’ carbon dioxide levels exhaled during the activities while factoring their age, gender and body mass index. From August to November 2015, the researchers independently bought Apple Watch, Basis Peak, Fitbit Surge, Microsoft Band, Mio Alpha 2, PulseOn and the Samsung Gear S2 to make sure they could test the products fresh off the box.

It remains unclear exactly why counting calories was so inaccurate. Shcherbina wrote that gender, body mass index and even body hair and tattoos may affect the results. For example, male participants had a higher error rate than females. One hypothesis is that male body hair lessened the body contact between the devices and the person’s wrist and affected the energy expenditures.

As for why heart rates were measured more accurately, most devices now use a green LED light that shines into the skin to illuminate the blood flow and the shapes of blood vessels to get an image on how much the blood is pumping.