• News New Blog Banner

  • From PTJ: Common Activity Trackers May Be Inaccurate for Patients With PD

    It's important for physical therapists (PTs) to encourage patients with Parkinson Disease (PD) to stay physically active, and it would seem as though commercially available fitness trackers would be a good way to do that, by allowing the PT and patient to set home goals and track progress through step counts. But new research suggests that PTs may want to think twice about the data they get from the devices.

    In an article published in the August issue of PTJ (Physical Therapy), researchers in Boston analyzed data from 4 fitness trackers—2 worn on the wrist and 2 that attach at the waist—to see how the tracker step reports stacked up against videos that allowed PTs to visually observe and count steps taken. The trackers in question were the wrist-worn Fitbit Surge and the Jawbone Up 2 and the waist-worn Fitbit Zip and Jawbone Up Move.

    A total of 33 patients with mild to moderate PD were recruited for the study, which involved tests of both continuous and discontinuous walking while wearing all 4 activity trackers, with the wrist trackers worn on the less-affected arm. The continuous walking tests involved 2 bouts of 2-minute walks around a 92-meter rectangular track, the first lap at a comfortable speed, and second at a fast speed; the discontinuous walking tests consisted of an "obstacle navigation course" and a "household" course, where patients were required walk to different areas to perform typical household tasks such as taking off and hanging up a coat, washing and drying hands, throwing away trash, and picking up and setting down a glass. In addition to recording tracker data, the tests also were video-recorded so that a pair of PTs could count steps taken. Researchers then compared tracker data with the results of the video monitoring. Here's what they found:

    • Overall, the trackers were reasonably accurate at recording steps taken during continuous walking, with the waist-worn Fitbit Zip showing the highest accuracy, followed by the wrist-worn Jawbone Surge, and wrist-worn Fitbit Surge, and the waist-worn Jawbone Up 2.
    • Tracking discontinuous walking proved to be more problematic, with authors of the study describing all 4 trackers as "generally inaccurate" in both courses. The Jawbone Up Move proved to be the least reliable device, with a mean absolute percent error rate approaching 60% in the household course. The Fitbit Zip was the most reliable, but that's not saying much: its error rate in the household course was close to 30%. The devices fared somewhat better in the obstacle negotiation course but still produced error rates ranging from about 10% to 20%. All devices underreported steps taken.

    Authors speculate that the inaccuracies may have something to do with a lack of tracker sensitivity to steps taken "in environments with greater discontinuity, where starting, stopping, and turning occur frequently." The longer, more symmetrical step lengths associated with continuous walking are better suited to the device's abilities, whereas the "smaller, slower, shuffling steps" taken by participants during the discontinuous walking tests tend to be missed by the devices, they write.

    As for waist-worn versus wrist-worn devices, authors think that the higher accuracy of the waist-worn devices may be due to the fact that the device is closer to an individual's center of mass, which allows for more accurate measurement. Wrist devices worn by patients with PD may be less accurate due to the effects of tremor, dyskinesia, extraneous upper extremity movement, and reduced arm swing often associated with individuals with PD, they believe.

    Another common feature of PD—freezing of gait—may also come into play as a factor affecting device accuracy, according to authors. Although only 1 participant in the study experienced freezing during the tests, that individual's devices produced an aggregate 60% error rate in the household course and 20% error rate in the obstacle negotiation course. "In general, the magnitude of this error exceeded that observed among nonfreezers," they write.

    The overarching problem, according to authors, is that none of the devices studied performed reliably in the setting that arguably would be the most important one for PTs treating patients with PD—the patient's home. "Other mechanisms of monitoring discontinuous walking, such as time spent walking, may be better options when the goal of intervention is focused on increasing physical activity in the home environment," they write.

    APTA members Nicholas Wendel, PT, DPT; Chelsea Macpherson, PT, DPT; Tamara DeAngelis, PT, DPT; and Cristina Colon-Semenza, PT, MPT, were among the coauthors of the study.

    Research-related stories featured in PT in Motion News are intended to highlight a topic of interest only and do not constitute an endorsement by APTA. For synthesized research and evidence-based practice information, visit the association's PTNow website.


    • I am curious as to the direction of the inaccuracies found in this study. I would assume that the trackers underestimated the amount of exercise performed. If that is the case, then unless there is a risk of "over-exercising", then the use of the devices to encourage activity and allow tracking is still valuable. Further, if the inaccuracy is consistent within patients (e.g. always 20% low), then the data obtained is still valid and useful relative to that patient!

      Posted by Jeff Harband, PT on 8/8/2018 4:20 PM

    • Thank you for your comment! It was true that the trackers underestimated the amount of exercise performed but unfortunately based on the design of the study we are not able to say if the inaccuracy was consistent within patients. However, the recommendation for use over longer distance continuous setting is still clinically helpful for many of our patients.

      Posted by Nicholas Wendel -> BMP^>G on 8/15/2018 4:13 PM

    Leave a comment
    Name *
    Email *