Using Wearable Devices and Smartphones to Track Physical Activity: Initial Activation, Sustained Use, and Step Counts Across Sociodemographic Characteristics in a National Sample

Background: Interest in using wearable devices and smartphones to monitor daily health behaviors, such as physical activity, is growing (1, 2). Many large employers are using these technologies in workplace wellness programs (3). The precision medicine initiative has described how data collected by these technologies can be used to better target interventions. However, the characteristics of persons who use these devices are poorly understood. Objective: To describe rates of initial use of activity trackers, sustained use after 6 months, and step counts across different sociodemographic characteristics from a wellness program offered across the United States. Methods and Findings: Data on activity tracker use, mean daily step counts, and sociodemographic characteristics between 2014 and 2015 were obtained from Humana for insured persons with access to HumanaVitality (now Go365), a wellness program offered across the United States. Median household income from U.S. Census data was linked using ZIP code. Data were received deidentified and deemed exempt from review by the University of Pennsylvania Institutional Review Board. The program supported more than 60 wearable devices and smartphone applications. Activity trackers needed to be connected to the wellness platform once, and then data were transmitted automatically as the device was used. The program had a daily goal of 10000 steps and used gamification with points and levels. Points were earned for reaching goals or logging workouts. Commercial insurance plans offered additional points for the first and fifth workout each week. Achieving higher levels made points redeemed for gift cards or other prizes more valuable. The maximum expected daily incentive value ranged from approximately $0.25 to $0.40. Initial activation rates were evaluated during the 2-year period. To allow for 6 months of follow-up for sustained use and step counts, we evaluated persons who activated by 30 June 2015. We estimated the proportion of persons still transmitting step data at 6 months and their mean daily step counts and the proportion who had achieved mean step count goals (10000 steps per day). The top and bottom first percentiles of step counts were removed as outliers. All analyses were conducted using SAS, version 9.4 (SAS Institute). The sample (n= 4483853) was 53.0% female and 37.4% elderly (aged65 years) (Table 1). During the 2 years, 1.2% of persons activated a device (0.2% in 2014 and 1.0% in 2015). Initial activation was done by 1.4% of women and 0.9% of men, 2.8% to 3.1% of younger adults (aged 23 to 49 years) and 0.1% of elderly persons, and 1.2% to 1.6% of those with a median annual household income of $50000 or higher and 0.7% to 1.0% of those with a lower income. Among those who activated a device, 69.2% (84.1% among elderly persons) used a Fitbit and 13.7% (14.3% to 17.3% among younger adults) used an Apple product. Table 1. Initial Activation of Activity Trackers, by Sociodemographic Characteristics and Device* Six months after activation, 80.0% overall, 90.4% of elderly persons, and 85.9% of Fitbit users had sustained use of the activity tracker (Table 2). The mean daily step count was 7683 overall, 8420 among men, 7291 among women, and 8085 among Fitbit users. Table 2. Sustained Use and Step Counts at 6 Months Discussion: This study had 3 main findings. First, activity tracker activation, sustained use, and step counts varied across sociodemographic characteristics. Second, initial activation was low, particularly among older and lower-income persons; however, overall activation rates increased between 2014 and 2015. Programs should consider ways to better engage older persons and those who may be less able to afford these devices. Third, sustained use and mean step counts were high among those who initially activated their devices, perhaps partly because of the program's use of gamification and incentives. We have previously shown that these approaches can be effective in other settings (4, 5). Programs should consider testing these types of engagement strategies to improve device use and physical activity outcomes. This study has limitations. Data were from a single insurer, incentives and program promotion could vary by insurance and employer, race/ethnicity was unavailable, and data from persons who used a device but did not activate it with the program were not captured. Sustained use over longer periods needs further study. To our knowledge, our study is 1 of the first national evaluations of activity tracker use among a large, diverse sample. Our findings offer new insights to better design interventions using wearable devices and smartphones.