Getting Moving Together; a field experiment on workplace physical activity

Importance: Wearable devices are widely used in an effort to increase physical activity and consequently to improve health. The evidence for this is patchy, and it does not appear that wearables alone are sufficient to achieve this end. Objective: To determine whether social comparisons in a workplace setting can increase the effectiveness of wearables at promoting physical activity. Design: A four week randomized controlled trial conducted in November 2015 with employees of a large firm. Participants were randomised to one of two treatment conditions (control vs social comparison) at team level, and teams are formed into ‘leagues’ based on their activity levels before the study. Impact is measured through wearable devices issued to all participants throughout the study duration. Setting: Offices of a large Australian employer. Participants: 646 employees of an Australian employer, issued with wearable activity trackers prior to the beginning of the study. Intervention(s) (for clinical trials) or Exposure(s) (for observational studies). Participants used a wearable device to track steps. Participants had been wearing these for at least four weeks at the outset of the trial, establishing a baseline level of activity. Teams (n=646, k=49), were randomly assigned to either control (k=24), or a social comparison (k=25) treatment. All participants took part in a step-count competition between their team and others at their employer, in which their team’s ranking within a mini-league of five teams, as well as their own activity was communicated each week. The control group had access to the usual features of the wearable, while the social comparison group received additional information about the performance of the other teams in their league, including how far behind and ahead their nearest rival teams were. Main Outcome(s) and Measure(s): Number of steps taken per day on average, measured by the wearable devices issued to all participants. Results: A total of 646 participants were included in the study. Compared to the control, participants in the social comparison group took significantly more steps per day during the trial period (an additional 620 steps, 8.2%, p<0.001). These effects are largest in both relative and absolute terms for people whose prior steps were in the bottom quartile of steps (an additional 948 steps, 40%, p<0.001), while the effect on people with highest levels of activity was a precisely estimated null (an additional 6 steps, 0.01%, p=0.98). Conclusions and Relevance: Social comparison increased the effectiveness of wearables at improving physical activity, particularly for those with the lowest baseline activity.