Monitoring occupational sitting, standing, and stepping in office employees with the W@W-app and the metawearc sensor: Validation study

Background: Replacing occupational sitting time with active tasks has several proposed health benefits for office employees. Mobile phones and motion sensors can provide objective information in real time on occupational sitting behavior. However, the validity and feasibility of using mobile health (mHealth) devices to quantify and modify occupational sedentary time is unclear. Objective: The aim of this study is to validate the new Walk@Work-Application (W@W-App)-including an external motion sensor (MetaWearC) attached to the thigh-for measuring occupational sitting, standing, and stepping in free-living conditions against the activPAL3M, the current gold-standard, device-based measure for postural behaviors. Methods: In total, 20 office workers (16 [80%] females; mean age 39.5, SD 8.1 years) downloaded the W@W-App to their mobile phones, wore a MetaWearC sensor attached to their thigh using a tailored band, and wore the activPAL3M for 3-8 consecutive working hours. Differences between both measures were examined using paired-samples t tests and Wilcoxon signed-rank tests. Agreement between measures was examined using concordance correlation coefficients (CCCs), 95% CIs, Bland-Altman plots (mean bias, 95% limits of agreement [LoA]), and equivalence testing techniques. Results: The median recording time for the W@W-App+MetaWearC and the activPAL3M was 237.5 (SD 132.8) minutes and 240.0 (SD 127.5) minutes, respectively (P<.001). No significant differences between sitting (P=.53), standing (P=.12), and stepping times (P=.61) were identified. The CCC identified substantial agreement between both measures for sitting (CCC=0.98, 95% CI 0.96-0.99), moderate agreement for standing (CCC=0.93, 95% CI 0.81-0.97), and poor agreement for stepping (CCC=0.74, 95% CI 0.47-0.88). Bland-Altman plots indicated that sitting time (mean bias -1.66 minutes, 95% LoA -30.37 to 20.05) and standing time (mean bias -4.85 minutes, 95% LoA -31.31 to 21.62) were underreported. For stepping time, a positive mean bias of 1.15 minutes (95% LoA -15.11 to 17.41) was identified. Equivalence testing demonstrated that the estimates obtained from the W@W-App+MetaWearC and the activPAL3M were considered equivalent for all variables excluding stepping time. Conclusions: The W@W-App+MetaWearC is a low-cost tool with acceptable levels of accuracy that can objectively quantify occupational sitting, standing, stationary, and upright times in real time. Due to the availability of real-time feedback for users, this tool can positively influence occupational sitting behaviors in future interventions.