Counterpoint: Step Activity Monitoring for the Purpose of Personal Information Compared to Health and Scientific Records

Home > Articles > Counterpoint: Step Activity Monitoring for the Purpose of Personal Information Compared to Health and Scientific Records
By M. Jason Highsmith, PhD, DPT, CP, FAAOP, and Joseph A. Miller, PhD, CP
graphic of arrows

This article addresses information presented in the recently published article "Accelerating Outcomes Measurement in Clinical Practice With Fitbit" (Allie Cerutti, The O&P EDGE, March 2016. The premise of Cerutti's article is correct: Access to objective, accurate, real-world patient walking data has clear value in O&P clinical practice. However, we feel the article's conclusion that Fitbit® is accurate and clinically relevant for O&P patients is questionable. The U.S. Department of Veterans Affairs (VA) previously completed evaluations on the Fitbit as an activity monitor in this capacity, which we would like to make O&P practitioners aware of. Additionally, the way in which supporting evidence for Cerutti's conclusions was presented does not provide a thorough analysis of Fitbit as a clinical tool. It is crucial that providers use the most appropriate, evidence-based tools to support delivery of optimal care for our patients.

In 2014, the VA convened a special committee that created specific device accuracy validation requirements and key metrics for utilization monitors for use with prosthetic limbs prescribed to veterans. These included cadence, cadence variability, stance and swing time, and others. The committee then conducted an exhaustive review of commercially available activity monitors (including Fitbit) to identify devices that would meet the accuracy validation and metric requirements. In this process, VA determined Fitbit did not meet the required capabilities nor did it provide the metrics required under the National Contract Template for Provision of Prosthetic Limbs - Utilization Monitors.1 It is worth noting in this context that the Fitbit organization states "the accuracy of the data collected and presented through the Fitbit Service is not intended to match that of medical devices or scientific measurement devices."2 It is our position that using a device to support clinical decisions when the manufacturer says that is not the intent of the device, and where that device is not regulated by the U.S. Food and Drug Administration (FDA), is inappropriate. Beyond the work of the VA, as noted previously, the way in which the supporting evidence was presented may be confusing to practitioners not familiar with each of these studies, which do not adequately support Cerutti's assertion that Fitbit is "94 percent accurate or greater and has at least a 'good' correlation with gold standards at walking speeds as slow as 0.4m/s." Furthermore, we disagree with her conclusion that Fitbit is appropriate as a clinical tool. The following discussion addresses the most significant concerns with the manner in which research is presented in her article. It is important that clinicians have this information so that they can make a fully informed decision-imperative when considering using this or any other physical activity monitor to assist in measuring clinical outcomes.

In Cerutti's article, Table 1 purports to show Fitbit's accuracy across clinical populations at 0.4m/s or greater, however the way in which this data is presented combines studies in which velocity was measured and those in which it was not. Because of the combination of data, lack of clarification in the reporting of some figures, and omission of statistics, taken as a whole, it overstates Fitbit's accuracy in populations with limited mobility.

• In the "Healthy Adults" section, Fortune3 reported on median accuracies, but since Cerutti did not specify this, her data in Table 1 may be construed as representing average accuracy. This is key as medians ignore how poorly the Fitbit measured some people. Moreover, Fortune states Fitbit "didn't detect steps for velocities less than 0.5m/s when located on the waist and detected only approximately 50 percent of steps for velocities less than 0.5m/s and greater than 2m/s when located on the ankle." Thus, an implication that Fitbit is highly accurate for all people when they walk between 0.5 and 0.9m/s is erroneous because Fitbit had large inaccuracies in several subjects when worn on the waist or ankle (< 90 percent accurate) in the walking speed range of 0.5- 0.9m/s. Reporting accuracy across the full range of walking speed is important for clinical decision making.

• Continuing in the "Healthy Adults" section, Ferguson4 did not include gait speed, so including this information in the table skews the perception of Fitbit's accuracy for slower walkers. Further the "excellent correlation" label is misleading, as criterion in this study (research-grade accelerometer, GT3X+) has been reported to underestimate steps by 60 percent at a walking speed of 0.67m/s and by 31 percent at walking speed of 0.90m/s.

• In "Stroke," the combined results of Klassen5 and Fortune indicate the Fitbit will be inaccurate when worn on the waist by post-stroke patients and unpredictable when worn on the ankle. In Klassen, when Fitbit was worn on the waist, it recorded no steps for many subjects walking

• In the "Stroke" and "Traumatic Brain Injury" categories, Cerutti references Fulk.6 In this study, the patients were actually high functioning and walking at an average speed of 0.84m/s in the stroke group and 1.1m/s for the traumatic brain injury group. Cerutti reports the intraclass correlation coefficient (ICC) = 0.70 in stroke rather than reporting the average accuracy of 84 percent, contradicting the author's assertion that Fitbit is 94 percent accurate or greater.6 Also, the ICC = 0.70 does not support the statement that Fitbit has a "good" correlation with gold standards at walking speeds as slow as 0.4m/s, as this "good" correlation was with subjects walking at an average speed of 0.84m/s, not 0.4m/s.

• In "Healthy Elderly," Cerutti only includes the accuracy results in Simpson7 of the Fitbit when worn on the ankle, which, as she mentions in the article, is not in accordance with Fitbit's placement guidelines. She excludes the poor accuracy results of Fitbit when worn on the waist, which is the manufacturer recommended placement, where the Fitbit recorded zero steps for some participants at all but the two fastest speeds (0.8m/s and 0.9m/s), and showed Fitbit was only 2 percent accurate at 0.3m/s and 18 percent accurate at 0.4m/s.

• In "Elderly Community Ambulators," an ICC = 0.88 is cited. Cerutti points out that gait speed was unknown in this study, which, as noted with the Ferguson data on healthy adults, cannot be used to support any particular gait speed ranges, i.e., whether actually at or above the 0.4m/s threshold. Large inaccuracies using Fitbit have already been demonstrated at gait speeds below 0.8m/s by Simpson, Klassen, and Fortune. It is worth pointing out that these inaccuracies apply to both the waist and ankle placements.

Moving beyond Table 1, Cerutti discusses Modus Health's FDA-listed Step- Watch™ and cites a Modus Health support reference that StepWatch is accurate down to 0.447m/s (1 mph). This gait speed is often cited because it has been vigorously supported by peer-reviewed publications where 0.447m/s is the slowest speed that was tested.8 However, high accuracy can be maintained at gait speeds slower than 0.4m/s with the StepWatch given the clinician's ability to adjust settings in the StepWatch software to reflect a patient's specific walking characteristics.9 This is important for clinicians needing to differentiate between patients with no mobility (K0) and very limited mobility (K1) when making care decisions.

Finally, Cerutti cites an article by Albert to support Fitbit's use in K-level determinations, stating that "the proportion of the device's recordings of 'very,' 'fairly,' and 'lightly' active minutes correlated with participants' assigned K-levels."9 However, Albert's feasibility study did not report actual statistical correlations. Further, the total number of subjects in this study was nine and the cited trend was based on two subjects: the sole K2 subject and the sole K4 subject. A trend demonstrating that the sole K2 subject has less activity than the sole K4 subject provides weak evidence for clinical use of Fitbit and specifically use of Fitbit to help justify K-level determinations.

In summary, it is our position that, based on activity monitor evaluations previously completed by the VA, Fitbit's own disclaimer that its data does not match accuracy requirements of medical devices, and the aforementioned issues with supporting data presentation in the article, the conclusion that Fitbit is sufficiently accurate and clinically relevant for O&P patients is not scientifically supported. We applaud the author's effort to contemplate broadening access of consumer activity monitoring technology into clinical practice. Providing more information to individuals regarding their activity habits has obvious benefits and appeal. However, it must be pointed out that use of specific technologies for interpersonal decision making is substantially different than precise activity monitoring as a matter of health information that is part of the medical or scientific record.

Editor's note: This article is offered as a counterpoint to a previously published article to allow experts in the profession an opportunity to present an opposing viewpoint regarding the clinical relevance of Fitbit. The original article was based on a presentation Cerutti authored as part of an American Academy of Orthotists and Prosthetists clinical seminar on gait assessment and was not intended as an exhaustive research study.

M. Jason Highsmith, PhD, DPT, CP, FAAOP, is the deputy chief of research & surveillance for the joint U.S. Department of Veterans Affairs and U.S. Department of Defense Extremity Trauma and Amputation Center of Excellence. He is jointly appointed as an associate professor at the University of South Florida's School of Physical Therapy & Rehabilitation Sciences.

Joseph A. Miller, PhD, CP, is the U.S. Department of Veterans Affairs national program director, orthotic and prosthetic clinical services. Miller, a U.S. Army veteran of Operation Iraqi Freedom, holds a doctoral degree in health science research/healthcare management.


  1. National Contract Template for Provision of Prosthetic Limbs - Utilization Monitors [press release]. 2015. Orthotic and Prosthetic Service (OPS).
  2. Fitbit Terms of Service (
  3. 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.
  4. 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.
  5. Klassen, T. D., L. A. Simpson, S. B. Lim, D. R. Louie, B. Parappilly, B. M. Sakakibara, D. Zbogar, and J. J. Eng. 2016. "Stepping up" activity poststroke: Ankle-positioned accelerometer can accurately record steps during slow walking. Physical Therapy 96(3):355-60.
  6. 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.
  7. 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.
  8. Foster, R. C., L. M. Lanningham-Foster, C. Manohar, S. K. McCrady, L. J. Nysse, K. R. Kaufman, D. J. Padgett, and J. A. Levine. 2005. Precision and accuracy of an ankle-worn accelerometer-based pedometer in step counting and energy expenditure. Preventive Medicine 41(3-4):778-783.
  9. Coleman KL, Smith DG, Boone DA, Joseph AW, del Aguila MA. Step activity monitor: long-term, continuous recording of ambulatory function. J Rehabil Res Dev. 1999; 36(1):8-18.