Activity trackers vary in accuracy

(Reuters Health) - Wrist-worn activity trackers, increasingly popular among consumers and in healthcare research, can vary considerably in their accuracy, a study from Iowa State University suggests.

Researchers pitted consumer devices against a gold-standard metabolic monitor and found the wristbands are more accurate for calories burned while resting or jogging, but have higher error rates for activities like weight lifting and crunches.

“We have continued to study these new consumer monitors since the companies provide NO data to support the accuracy of the estimates,” said study author Gregory Welk, an associate professor of kinesiology at Iowa State in Ames, by email.

Consumers are increasingly using the monitors to track healthy behaviors including exercise and sleep, and researchers are adopting the devices for similar uses in the context of health studies, the study team points out in Medicine and Science in Sports and Exercise.

To see if the products provide accurate data on calorie expenditure, Welk and his colleagues recruited 56 men and women to wear each of five consumer fitness trackers and each of two professional research monitors while sitting, doing aerobic exercise and doing resistance exercise.

The consumer products tested were the Fitbit Flex, Jawbone UP 24, Misfit Shine, Nike FuelBand SE and Polar Loop. The professional products were Actigraph GT3X-plus and BodyMedia Core. All were tested against results from a metabolic monitor, Oxycon Mobile, that measures breathing, heart rate and fat burned.

The study participants’ average body mass index - a measure of weight relative to height - was 24, which is on the high end of the normal-weight range. Ages ranged from 18 to 60.

Wearing all of the trackers at once, the participants completed 20 minutes of a sedentary activity such as reading or watching a movie, followed by 25 minutes of aerobic exercise on the treadmill and 25 minutes of resistance exercise. They rested for five minutes between sessions.

On average, participants burned 317 calories during the entire 80 minutes of activity, according to the metabolic monitor. The consumer fitness trackers’ estimates ranged from 275 calories to 396 calories. The two research monitors had the most accurate calorie estimates overall, but the Fitbit Flex, the Jawbone UP 24, and the Nike FuelBand SE provided what the researchers called “reasonably accurate” overall estimates.

The fitness trackers were least accurate for measuring calorie expenditure during resistance exercise. None had less than 25 percent error in that category.

For sedentary activity, two consumer monitors (the Misfit Shine and the Nike Fuelband SE) and one research monitor had error rates under 20 percent. For aerobic activity, one consumer monitor (the Nike Fuelband SE) and one research monitor had error rates under 20 percent.

“These trackers are best for bipedal locomotor tasks, such as walking and running over level ground,” said J. Adam Noah, a psychiatry researcher at the Yale School of Medicine in New Haven, Connecticut, by email.

The further away from this type of activity one gets, the less accurate the calorie estimates are likely to be, said Noah, who was not involved in the study.

Despite error rates, Noah noted, the technology may be useful for people monitoring their own or another person’s behavior. For example, he said, “It could be useful for a care-giver to see if a patient is performing daily activities for weight loss, rehabilitation or prevention.”

“The real value is that the monitors help a person stay accountable and it may remind some to stay active or to get some activity every day,” Welk said.

SOURCE: Medicine and Science in Sports and Exercise, online July 6, 2015.