Increasing Physical Activity in Inactive Adults: A Randomized Crossover Trial Comparing Two Highly Popular Apps

BACKGROUND: Despite widespread use, smartphone apps for physical activity (PA) lack rigorous evaluation. This study examined the impact of two top PA apps through a crossover trial. OBJECTIVE: To assess the feasibility, acceptability, and effectiveness of two smartphone apps in increasing physical activity among inactive UK adults. METHODS: A randomized crossover trial was conducted with inactive UK smartphone users. After a 1-week baseline period, participants were randomly assigned to one of two sequences: App A followed by App B (A/B) or App B followed by App A (B/A), with a crossover to the alternate app occurring after the initial 2-week intervention period. App A was a 7-minute workout, and App B was a Couch to 5k program. Feasibility was assessed based on recruitment, retention, and adherence rates. Physical activity was measured objectively using accelerometry at baseline, post-baseline (week 1), week 3, and week 5. Self-reported PA levels, sedentary behavior, exercise self-efficacy, and intentions were collected at week 1 and at the end of each intervention period (weeks 3 and 5)._ _The primary analysis assessed changes in PA from baseline to the first intervention period (week 3); secondary analysis compared the two apps. Trial registration: ClinicalTrials.gov NCT03565627. RESULTS: 209 participants accessed the screening survey. 104 were eligible and consented; 63.5% (66/104) were enrolled and randomized. 87% completed the trial. For accelerometer-measured outcomes, there were no significant differences in mean change. 16/51 participants (31.4%) increased their time in moderate to vigorous PA (MVPA) by 20% from baseline following the introduction of the intervention (weeks 3 and 5) (95% CI= 19.1% to 45.39). Self-reported PA outcomes showed significant increases: total time spent in PA (LSM= 32.52, p<.005), moderate PA (LSM= 113.68, p<.024), walking (LSM= 375.0, p<.007), and total PA (LSM= 489.46, p<.010). Sedentary behavior decreased (LSM= -123.23, p<.001). Exercise self-efficacy (LSM= 41.78, p<.0001) and intentions increased (LSM= 5.23, p<.0001). Lower baseline activity was associated with a larger increase in PA (p< 0.03 for all measures). There were no significant differences between the two apps. CONCLUSIONS: A crossover trial is a feasible and acceptable method to study apps and can be used to accelerate the evidence generation for digital health. The two PA apps showed promising results, with an impact observed for a 20% increase in MVPA, self-reported PA, intentions, and exercise self-efficacy. The biggest improvements were in the participants with low baseline PA, who have the greatest unmet need. The study detected no differences between the apps.