JMIR Publications

ABSTRACT

Background:

Fitness trackers like those offered by Fitbit and research devices like the ActiGraph GT3X+ are both used to measure "moderate-to-vigorous physical activity" (MVPA), the kind of activity health guidelines recommend getting 150 minutes of per week. But these devices don't measure the same thing, even though they use similar words and labels. ActiGraph estimates activity based on how much a person physically moves with everyone held to the same standard of MVPA. Fitbit largely estimates activity based on heart rate, which reflects how hard your body is working relative to your own personal range. One asks "how much did you move?" The other asks "how hard is your body working?" These aren't always the same.

Objective:

To test whether a consumer wearable (Fitbit) and a research-grade accelerometer (ActiGraph) measure a common latent construct or instead operationalize fundamentally distinct constructs of physiological effort–based intensity versus mechanically derived movement to assess MVPA, intensity. We further assess how this construct divergence affects estimates of MVPA accumulation under free-living conditions.

Methods:

Adults (N=356; mean age 47.1 years; BMI 27.5 kg/m²) participating in the YourMove trial (NCT05598996) wore a Fitbit Versa and an ActiGraph GT3X+ three times for one week each. Analyses were restricted to matched wear time, with Fitbit Intensity and Active Zone Minutes compared with Freedson based cutpoints of Actigraph data. Agreement was assessed using confusion matrices, Cohen’s kappa, and Bland–Altman analyses. Subgroup analyses examined differences by sex, age, Body Mass Index (BMI), and fitness.

Results:

Overall agreement was high (~0.94), but agreement beyond chance was low (κ=0.18–0.30), indicating limited concordance in classifying higher-intensities. Both Fitbit metrics systematically estimated higher MVPA with greater differences observed at higher activity levels. Agreement on weekly guideline attainment was moderate and depended on the Fitbit-derived reference metric. Discrepancies increased in males, older adults, and individuals with higher BMI.

Conclusions:

Fitbit and accelerometer-derived estimates of MVPA are not directly interchangeable in free-living conditions. Observed differences reflect systematic divergence in how physical activity intensity is operationalized. These construct-level differences have meaningful implications for estimates of guideline attainment and subgroup comparisons, underscoring the need to align metric selection with the intended interpretive and analytic purpose of the measurement. Clinical Trial: Clinicaltrials.gov NCT05598996