The Feasibility of Using Consumer-Level Activity Trackers in Population Monitoring of Physical Activity: Comparing Representativeness and Measurement Quality With Self-Report and a Professional Research-Grade Accelerometer

Abstract

Consumer-level activity trackers can potentially be used for population monitoring of physical activity, without suffering from the recall and social desirability bias that occur in self-report and at lower costs and effort compared to research-grade devices. However, other drawbacks may be at play when using personal activity trackers. The current study compares response, representativeness and measurement quality of consumer-level activity trackers to a research-grade accelerometer (ActivPAL) and self-report (the SQUASH survey). The study existed of a questionnaire on physical activity, possession and usage of a personal activity tracker and willingness to share data or wear a research-grade device. Subsequently, a smaller follow-up study was conducted in which respondents wore an ActivPAL and their own personal tracker to allow comparison of the different methods. The results showed a loss of respondents in each step of the process. Additionally, the representativeness of the response was influenced by both demographics and physical activity level, introducing nonresponse bias. The measurements of personal trackers had a decent agreement with the ActivPAL for number of minutes MVPA and steps, while both objective measures differed substantially from self-report on all indicators. It is concluded that consumer-level trackers are not a feasible replacement for self-report in large-scale studies for estimation of the exact physical activity level of a population due to representation and measurement issues, but could be used, possibly together with research-grade devices, to supplement or calibrate survey results.