We commend Toth and colleagues (1) for undertaking this rigorously designed study evaluating the free-living criterion validity of seven wearable step counters relative to direct observation (video recording and manual counting). This study highlights two issues that we believe warrant additional focused discussion and comment.
First, without disputing the superior accuracy of the StepWatch (mean absolute percent error [MAPE] between 4.0% and 5.2%; range dependent upon wear location and settings), its physical design (i.e., relatively large size, ankle wear location), requisite knowledge of wearable device data acquisition and processing methods, and relatively high cost (~$2000 for device, software and accessories) may limit its widespread clinical or real-world adoption. Several of these arguments can also be made for the ActiGraph GT9X (25.1%–119.7% MAPE) and activPAL (22.6%–23.1% MAPE), two devices widely used in the research. By contrast, the Yamax Digi-Walker SW-200 (18.6% MAPE) and New Lifestyles NL-2000 (19.5% MAPE), two relatively simple step counters that cost approximately US $20 and US $70, respectively, performed marginally better than the ActiGraph and activPAL. Such low-cost step counters may be “good enough” to identify behavior patterns and/or track changes in large-scale research studies, translational clinical trials, and frontline consumers.
Second, of all devices tested in the free-living condition by Toth and colleagues (1), only the StepWatch achieved an MAPE close to the suggested error tolerance levels of 3% (2) and 5% (3,4) for laboratory-based studies. The remaining devices achieved 17.5% to 119.5% MAPE. Further research is required to establish acceptable error tolerances for devices evaluated under free-living conditions. Moreover, it is apparent from this study by Toth and colleagues (1) that previous free-living studies using the ActiGraph (as a single illustrative example) may have under or overestimated stepping behavior by up to approximately 30% or 220% (respectively, dependent on wear location and data processing methods). As Toth and colleagues (1) acknowledged, step detection underestimation may be partially attributable to a lack of device sensitivity during slow ambulation and brief bouts of sporadic stepping (e.g., shuffling). Overestimations were likely attributable to the use of the low frequency extension, which is designed to increase sensitivity to low amplitude movement, for example, slow ambulation, but may have inadvertently classified movement artifact (e.g., gesturing, bumps, car travel) as steps. These issues generate several questions: Does every step “count”? Do steps accumulated at low intensities contribute meaningfully toward health benefits? Is 20% to 30% MAPE an acceptable error tolerance for free-living conditions? Finally, in light of this measurement error, can we realistically provide step-based physical activity recommendations for the general public?
Our aim is to stimulate further discussion and research regarding the cost performance and practicality trade-off of step counters across their potential uses in research, clinical practice, the real world. In addition, evidence-based standards are needed to establish appropriate and realistic error tolerances for step counters used in both laboratory and free-living settings. Given the recent formal acknowledgment of step counting in the United States 2018 Physical Activity Guidelines Advisory Committee Scientific Report (5), we believe these issues are important to address to advance the field of physical activity measurement.
Elroy J. Aguiar
Christopher C. Moore
Scott W. Ducharme
Department of Kinesiology
University of Massachusetts Amherst
Aston K. McCullough
Department of Biobehavioral Sciences
Columbia University Teachers College
and Department of Neurology
Columbia University Medical Center
New York, NY
1. Toth LP, Park S, Springer CM, Feyerabend MD, Steeves JA, Bassett DR. Video-recorded validation of wearable step counters under free-living conditions. Med Sci Sports Exerc
2. Hatano Y. Prevalence and use of pedometer. Research J Walking
3. Feito Y, Bassett DR, Thompson DL. Evaluation of activity monitors in controlled and free-living environments. Med Sci Sports Exerc
4. Vincent SD, Sidman CL. Determining measurement error in digital pedometers. Measurement in Physical Education and Exercise Science
5. 2018 Physical Activity Guidelines Advisory Committee. 2018 Physical Activity Guidelines Advisory Committee Scientific Report
. Washington, DC: U.S. Department of Health and Human Services; 2018.