Accelerating Outcomes Measurement in Clinical Practice With Fitbit
IT'S A WORLD OF WEARABLES
Wearable sensors, or wearables, seem to be everywhere these days. The concept of using wearables for personal health tracking has become commonplace. Previously, we had to rely on health and fitness professionals to obtain information about our general activity level or health. Now, access to this data is at our fingertips- and, literally speaking, at our wrists, waists, or ankles. The Fitbit® technology is a prime example.
The Fitbit activity tracker is a popular accelerometer-based device intended for personal activity tracking. It can be worn at the waist, wrist, or ankle. The accelerometer within each Fitbit has been designed to capture movement of a typical or normal gait pattern, from walking to running. As such, activity trackers are marketed toward relatively healthy individuals who want to maintain or improve their general health and fitness. Though the promise of personalized health data might pique consumers' interest enough to purchase a wearable, it may not lead to lifelong monitoring of their health. In a 2014 article for Wired magazine, J.C. Herz wrote that about 50 percent of wearable users abandon their devices and about one-third of these individuals do so after only six months.1 His observation speaks to typical wearable users' transient interest.
Unlike typical users of wearables, however, many O&P patients do have a long-term commitment to tracking their health.
We know that users of lower-limb orthotic and prosthetic devices invest more thought and energy into walking than the average healthy person. This is why prosthetists and orthotists encourage the use of devices that lessen the burden of walking and other activities. In most cases, the patients who see us regularly are committed to improving their health and increasing daily engagement in activities. Also in his article, Herz describes wearables as "failing the people who need them most," indicating that wearables have not yet specifically considered the needs of people who do not possess a healthy, normal gait.1 According to his definition, the people who need them most include the majority of our patients.
Nonetheless, this failure on the part of wearable manufacturers may be irrelevant. Without specifically targeting its design or marketing, Fitbit has demonstrated potential to be a simple, effective, and readily accessible clinical tool for O&P patients.
FITBIT REPURPOSED: A CLINICAL TOOL FOR O&P
To use Fitbit as a clinical tool, we need to be aware of what it measures and how it does so.
What does Fitbit measure? This activity tracker measures the wearer's everyday activity patterns in terms of step count, distance walked, floors/stairs climbed, calories burned, and active minutes. These parameters are calculated via proprietary algorithms that take into account two sources of information: acceleration data recorded by the device, and manually entered data specific to the user, such as height, weight, age, and step length. Step count is simply the number of steps taken in a given period of recording time. Distance walked and floors/stairs climbed show the distance traveled horizontally and vertically, respectively. Calories burned represents the total energy expended by the user during the recording time. Lastly, the user earns active minutes for every minute of activity that is considered of moderate to vigorous intensity. Intensity is classified as moderate or vigorous based on the standard metabolic equivalent (MET) of activity type compared to sitting quietly. For example, walking briskly is considered moderately intense and assigned three to six METs, meaning the energy cost is three to six times greater than the cost of sitting. Activities like running and boxing are considered vigorously intense, earning more than six METs. The user's total number of active minutes per day reflects his or her overall activity intensity level.
How does Fitbit work? The easiest way to understand the accelerometer inside the device is to look at how it detects a step. There exists a threshold for the acceleration signal, and each time the threshold is met or exceeded, the accelerometer counts one step. If the threshold is not met, the accelerometer does not register a step. A common source of error arises from a step that does not produce an acceleration signal that reaches the threshold, for example, a step that is taken very slowly may not produce a signal. On the other hand, an acceleration signal can reach or cross the threshold when a step was not actually taken. For instance, stepping or moving around in place or fidgeting in a chair could produce a signal that meets or exceeds the threshold. These sources of error result in undercounting or overcounting steps.
To be considered a clinical tool-and not just a handy device for recreational use-Fitbit must be accurate and robust. It should stand up to existing gold standard methods, which include observation by video recording and research-grade accelerometer systems designed for clinical use.
Is Fitbit accurate? Multiple recent studies attest to Fitbit's accuracy in counting steps for people with abnormal gait patterns and known deviations: individuals following a stroke or who have traumatic brain injuries (TBIs), and the elderly. Table 1 shows that step count accuracy ranged from 94 to 97.5 percent in healthy adults, healthy elderly individuals, and people who have suffered a stroke.2-4 Moreover, step count had excellent correlation (ICC = 0.88-0.99 and r = 0.99) with gold standard methods in healthy adults, elderly community ambulators, and people with TBIs.5-7 Step count had good correlation (ICC = 0.70) in people who had suffered a stroke.7 This data reflects comparison to a visual count or to a research-grade accelerometer.2-7
For studies where walking speed was known, data presented in Table 1 was constrained to walking speeds of 0.4 meters per second (m/s) or greater, with two exceptions. In one study, all participants walked at speeds between 0.3 and 0.36 m/s.8 In another study, the data was organized such that it was constrained by walking speeds between 0.5 and 1 m/s.9
So, why is 0.4 m/s significant?
Do walking speed and body placement matter? It turns out that Fitbit's accuracy depends on the user's walking speed. In 2014, one study showed a substantial decline in accuracy when participants with TBIs and stroke walked slower than 0.58 m/s.7 In the same year, a similar study supported this finding, showing that accuracy dropped when healthy adults walked about 0.5 m/s.9 However, a series of two more recent and more rigorous studies found that accuracy was maintained until the speed fell below 0.4 m/s.3,4 Furthermore, in the only study that recorded poor accuracy of the wearable, all participants walked at 0.36 m/s or slower.8
Hence, the evidence suggests that Fitbit can accurately record step count for user walking speeds of 0.4 m/s or greater. Speeds of 0.9 m/s or less are considered slow, making the accelerometer's high accuracy remarkable at such slow speeds.3,4 As orthotists and prosthetists, the accuracy of an activity monitor at slow walking speeds is of particular interest since our patients likely walk more slowly than the average person without a mobility impairment.
The device's accuracy also depends on its body placement. Contrary to manufacturer recommendations to place Fitbit at the waist, the aforementioned results suggest placing it at the ankle. Moving it to the ankle increases step count accuracy for people who walk slowly.3,4 The rationale behind this has to do with the magnitude of the acceleration signal being detected. Acceleration magnitude is greater at the ankle than at the waist. The greater the acceleration signal, the more sensitive Fitbit is to detecting a step. Hence, placing Fitbit at the ankle increases step count accuracy for slow walkers.
Is Fitbit robust? Fitbit not only has the capacity to yield accurate, clinically relevant data, but it's also robust. For example, one of the studies involving people post stroke noted that despite gait deviations in stance and swing phase, the activity monitor had high step count accuracy (Table 2).4 Together, Tables 1 and 2 indicate that the device's step count is resilient against numerous common gait deviations, which is crucial for it to be considered for clinical use. This data should encourage us to try Fitbit as a means to see beyond the deviations in our patients' gait patterns.
WHAT ABOUT EXISTING CLINICAL TOOLS?
You may have heard about other accelerometer-based devices. Some, like the modus StepWatch™ activity monitor, have been specifically engineered as clinical and research tools. StepWatch was designed to measure various gait parameters for O&P patients, and has received attention from the U.S. Department of Veterans Affairs (VA). Per its new National Contract Template for Provision of Prosthetic Limbs-Utilization Monitors, the VA has mandated the monitoring of activity and prosthesis use among veterans with new amputations.10
Of note, modus health reports that StepWatch is 98 percent accurate or greater at walking speeds as slow as one mile per hour-or 0.447 m/s.11 And as discussed previously, Fitbit is 94 percent accurate or greater and has at least "good" correlation with gold standards at walking speeds as slow as 0.4 m/s. These insights suggest that StepWatch and Fitbit accuracy may be comparably limited by slow walking speeds. However, we have not investigated any clinical significance behind the discrepancy in accuracy values between the two devices.
A major advantage of Fitbit over StepWatch in clinical use is cost-effectiveness. Whereas Fitbit synchronizes to free online software (on a laptop or mobile device), StepWatch requires purchase of a proprietary software package as well as a device docking station. Furthermore, Mark V. Albert, PhD, assistant professor of computer science at Loyola University Chicago, a proponent of Fitbit's clinical applicability, weighed research-grade equivalents like StepWatch against consumer-grade wearables like Fitbit: "Although validated, researchgrade monitors exist for [assessing mobility], the low cost, wide availability and direct-to-consumer design of this class of activity monitors brings distinct advantages in adoption by clinics and individuals."12 As such, for the general clinician, Fitbit may prove itself worthy of investigation.
ACCELERATING OUTCOMES MEASUREMENT
Especially in today's healthcare climate, clinicians must be aware of the tools available to us for measuring outcomes. Outcomes measurement is essential to understanding how our decisions influence real life-how patients are using and benefiting from orthoses and prostheses.
Gait analysis. Observational gait analysis is one way we readily and effectively assess the impact of the devices we provide. Fitbit can supplement our subjective analysis with objective gait data. Step count and distance are known indicators of overall activity level. Cadence and speed denote a person's general functioning and the presence of disability or a movement disorder. Together, the measurements from this accelerometer may be compared before and after prosthetic or orthotic intervention to illustrate effects on activity level and functioning.
K-level evaluation. Additionally, Fitbit may be a useful tool for assessing K-level. In 2014, Albert et al. described a positive relationship between predetermined K-level and Fitbit measurement of activity level in the study cohort, which comprised people with transfemoral amputations.13 The proportion of the device's recordings of "very," "fairly," and "lightly" active minutes correlated with participants' assigned K-levels.13 Those with a higher K-level were tracked as being more active. For instance, the only K2 ambulator in the cohort had the greatest proportion of lightly active minutes whereas the only K4 ambulator in the cohort had the greatest proportion of very active minutes.13
The current evidence reviewed in this article shows that wearable technology is extending its reach to those in need, such as O&P patients. Despite its original design intentions, Fitbit has demonstrated several qualities desirable in a clinical tool: accuracy and robustness, accessibility, and clinical relevance.
Allie Cerutti, BSBE, MPO, is an orthotics resident with Orthotic & Prosthetic Lab, St. Louis, Missouri.
- Herz, J. C. 2014. Wearables are totally failing the people who need them most. Wired, November 6. www.wired.com/2014/11/where-fitness-trackers-fail.
- Storm, F. A., B. W. Heller, and C. Mazza. 2015. Step detection and activity recognition accuracy of seven physical activity monitors. PLoS One 10 (3):e0118723.
- Simpson, L. A., J. J. Eng, T. D. Klassen, S. B. Lim, D. R. Louie, B.Parappilly, B. M. Sakakibara, and D. Zbogar. 2015. Capturing step counts at slow walking speeds in older adults: Comparison of ankle and waist placement of measuring device. Journal of Rehabilitation Medicine 47 (9):830-5.
- Klassen, T. D., L. A. Simpson, S. B. Lim, D. R. Louie, B. Parappilly, B. M. Sakakibara, D. Zbogar, and J. J. Eng. 2015. "Stepping up" activity poststroke: Ankle-positioned accelerometer can accurately record steps during slow walking. Physical Therapy.
- Ferguson, T., A. V. Rowlands, T. Olds, and C. Maher. 2015. The validity of consumer-level, activity monitors in healthy adults worn in free-living conditions: A cross-sectional study. International Journal of Behavioral Nutrition and Physical Activity 12:42.
- Paul, S. S., A.Tiedemann, L. M. Hassett, E. Ramsay, C. Kirkham, S. Chagpar, and C. Sherrington. 2015. Validity of the Fitbit activity tracker for measuring steps in community-dwelling older adults. BMJ Open Sports & Exercise Medicine 1:e000013.
- Fulk, G. D., S. A. Combs, K. A. Danks, C. D. Nirider, B. Raja, and D. S. Reisman. 2014. Accuracy of 2 activity monitors in detecting steps in people with stroke and traumatic brain injury. Physical Therapy 94 (2):222-9.
- Lauritzen, J., A. Munoz, J. L. Sevillano, A. Civit. 2013. The usefulness of activity trackers in elderly with reduced mobility: A case study. Studies in Health Technology and Informatics 192:759-62.
- Fortune, E, V. Lugade, M. Morrow, and K. Kaufman. 2014. Validity of using tri-axial accelerometers to measure human movement - Part II: Step counts at a wide range of gait velocities. Medical Engineering and Physics 36 (6):659-69.
- National Contract Template for Provision of Prosthetic Limbs - Utilization Monitors [press release]. 2015. Orthotic and Prosthetic Service (OPS).
- modus Trex Activity Monitor & Docking Station. Version 3.4 ed: modus Health; 2015. modushealth.com/support/
- C. Tingle. 2014. Use of fitbit activity monitor successfully assessed transfemoral amputee activity levels. O&P News, Fall.
- Albert, M. V., S. Deeny, C. McCarthy, J. Valentin, and A. Jayaraman. 2014. Monitoring daily function in persons with transfemoral amputations using a commercial activity monitor: a feasibility study. PM&R: The journal of injury, function, and rehabilitation (12):1120-7