JMIR Publications

ABSTRACT

Background:

Engagement with online health behaviour-change programs is positively associated with their effectiveness. However, it is unclear whether tracking devices that sync data automatically (e.g., Fitbit) result in different engagement levels compared with manually entering data.

Objective:

This study examined how different methods of logging steps in the freely-available 10,000 Steps physical activity program influence engagement with the program.

Methods:

A subsample of users (n=22,142, 71% female) of the free 10,000 Steps physical activity program (www.10000steps.org.au) were classified into one of five user-groups based on the method used to log steps: Website-only (n=14,617; 66.0%); App-only (n=2,100; 9.5%); Fitbit-only (n=1,705; 7.7%); Web-and-App (n=2,057; 9.3%); and Fitbit-Combination (combination of web-app-and-Fitbit; n=1,663; 7.5%). Users’ engagement was assessed using generalised linear regression models to examine differences between user groups in website sessions, minutes per session, page views, pages per session, daily step count and total step log entries. Binary logistic regression was used to measure associations between user groups in participation in Challenges and Tournaments and receiving and sending friend requests. Time to non-usage attrition was assessed using Cox proportion hazards regression.

Results:

All outcomes are relative to the Website-only group. The App-only group had significantly fewer website sessions (-6.9, 95% CI -7.6--6.2), and the Fitbit-only (10.6, 95% CI 8.8-12.3), Web-and-App (1.5, 95% CI 0.4-2.6) and Fitbit-Combination (8.0, 95% CI 6.2-9.7) groups had significantly more. The App-only (-0.7, 95% CI -0.9--0.4) and Fitbit-only (-0.5, 95% CI -0.7--0.2) groups spent significantly fewer minutes, whilst the Fitbit-Combination group (0.2, 95% CI 0.0-0.5) spent significantly more minutes on the website each session. All groups, except the Fitbit-Combination group, viewed significantly fewer website pages per session. The mean daily step count was significantly lower for the App-only (-201.9, 95% CI -387.7--116.0) and Fitbit-only (-492.9, 95% CI -679.9--305.8) groups, but it was significantly higher for the Web-and-App group (258.0, 95% CI 76.9-439.2). The Fitbit-only (5.0, 95% CI 3.4-6.6), Web-and-App (7.2, 95% CI 5.9-8.6), and Fitbit-Combination (15.6, 95% CI 13.7-17.5) groups all significantly logged a greater number of step entries. The App-only group was less likely (OR 0.65, 95% CI 0.46-0.94) and all other groups were significantly more likely to participate in Challenges. All groups, except the App-only group, had a significant difference in time to non-usage attrition relative to the Website-only group (HR range=0.55 to 0.75; p <.001). The mean time to non-usage attrition was 35±26 days (Website-only=32±22; App-only=33±23; Fitbit-only=40±29, Web-and-App=39±27; Fitbit-Combination=50±40).

Conclusions:

The use of a Fitbit in combination with the 10,000 Steps app and/or website enhanced engagement with a real-world physical activity program. Integrating tracking devices that sync data automatically into real-world physical activity interventions is one strategy to improve engagement.