In the previous post in this blog, I discussed the accuracy of measurement of energy expenditure by fitness machines. In short, it appears that machines that use more data to estimate energy expenditure are likely to be more accurate, but that it is very difficult to know for sure. With wearable devices, which include, but are not limited to, devices made by Fitbit, Garmin, Polar, Timex, Motorola, Nike, and Suunto, however, there is more information available about the accuracy of this measurement. This article will be limited to the accuracy of wearable accelerometer devices like the Fitbit.
These devices work by measuring the movement of the device, which can be on a wrist, upper arm, waistband, or elsewhere. This then leads to estimates of number of steps taken and, by using estimated or measured stride length, estimates of distance traveled. There is a nice article in Berkeley Science Review about the accuracy of steps taken and confounding factors (for example, Fitbits are more accurate with walking at slow speeds and less accurate with walking at faster speeds and running, especially if the devices are kept in a pocket). But we are interested in the accuracy of measurement of energy expenditure. Here is the text from the Fitbit website about how they measure this:
Your tracker and dashboard show an estimated number of calories burned based on your BMR (Basal Metabolic Rate), which we calculate using the height, weight, age, and gender information that you provided when you set up your Fitbit account. If your tracker measures heart rate, the calorie burn estimate also takes heart rate into account. Note that calorie tracking for the following day begins at midnight and incorporates the calories you burn while sleeping.
When you sync your tracker, Fitbit replaces your estimated calorie burn with your tracker’s data. If you manually log activities, the calories burned by during those activities are taken into account as well.
When you haven’t synced your device or logged any activities, Fitbit tries to guess how many calories you have burned if you got out of bed, got dressed, went to your day job, came home, and did nothing much more than walk to your car.
Once you start logging activities, Fitbit stops estimating and uses the data you’ve provided instead. The more you wear your tracker, the more accurate your calorie burn data will be.
In the literature, there have been reports that show that, under controlled conditions, accelerometer devices, such as the Fitbit, can be surprisingly accurate. For example, in a study published in September, 2014, sixty volunteers (average age 26.4 years, 30 male, 30 female), wore eight different types of activity monitors simultaneously while completing a 69-minute “protocol.” An indirect calorimeter was also worn and the measurements of this device were used as the reference. The authors concluded that:
For overall group comparisons, the mean absolute percent error values (computed as the average absolute value of the group-level errors) were 9.3%, 10.1%, 10.4%, 12.2%, 12.6%, 12.8%, 13.0%, and 23.5% for the BodyMedia FIT, Fitbit Zip, Fitbit One, Jawbone Up, ActiGraph, DirectLife, NikeFuel Band, and Basis B1 Band, respectively.
This means that, for most of the devices used, in a controlled test, the inaccuracy of measurement of caloric expenditure was between 9 and 13%. However, there are limitations. For example, in another recent study, 23 subjects were fitted with two Fitbit and Fitbit Ultra accelerometers, two industry-standard accelerometers, and an indirect calorimetry device (which served as the reference). These subjects participated in 6 minute bouts of treadmill walking (3.5 mph), jogging (5.5 mph), and stair stepping. The results of this study indicate that the Fitbit and Fitbit Ultra are reliable and valid for monitoring steps and for determining energy expenditure while walking and jogging without an incline (the Fitbit devices produced estimates of caloric expenditure that were 88-113% of those values that were measured by the indirect calorimetry device). However, with inclined activities, the Fitbit and standard accelerometers under-estimated energy expenditure by 40% or more. Finally, in another recent study,20 participants (10 male, 10 female) wore a Fitbit on the hip and also wore a separate indirect calorimeter as a reference. These subjects performed walking and running trials on a treadmill and a simulated free-living activity routine. The authors found that:
The Fitbit significantly underestimated EE [energy expenditure] for cycling, laundry, raking, TM [treadmill] 3 mph at 5% grade, ascent/descent stairs, and TM 4 mph at 5% grade, and significantly overestimated EE for carrying groceries. Energy expenditure estimated by the Fitbit was not significantly different than EE calculated from the Oxycon Mobile [the indirect calorimeter] for 9 activities.
The authors then conclude that:
The Fitbit worn on the hip significantly underestimates EE of activities. The variability in underestimation of EE for the different activities may be problematic for weight loss management applications since accurate EE estimates is important for tracking/monitoring energy deficit.
My conclusion is that accelerometers, like the Fitbit, can serve a useful role in an individual’s fitness program. These devices do not cost as much as more sophisticated “athlete-grade” devices, but can provide, with limitations, a decent snapshot into “calories burned.” For limited fitness activities, which many people use for their fitness, such as walking the dog around a park or walking with co-workers over the lunch hour, the Fitbit-like devices appear particularly well-suited. If you are planning to use such a device, please discuss it with your health-care provider to best incorporate it into an overall fitness program.
Adam, NJ, Spierer, DK, Gu, J, et al. Comparison of steps and energy expenditure assessment in adults of Fitbit Tracker and Ultra to the Actical and indirect calorimetry. J Med Eng Technol 2013 Oct; 37 (7): 456-462.
Lee, JM, Kim, Y, and Welk, GJ. Validity of consumer-based physical activity monitors. Med Sci Sports Exerc 2014 Sep; 46(9): 1840-1848.
Sasaki, JE, Hickey, A, Mavilia, M., et al. Validation of the Fitbit Wireless Activity Tracker for prediction of energy expenditure. J Phys Act Health 2014 Apr; epub ahead of print.