Effect of Different Financial Incentive Structures on Promoting Physical Activity Among Adults: A Randomized Clinical Trial

Abstract

Importance: Few adults engage in recommended levels of physical activity. Financial incentives can promote physical activity, but little is known about how the structure of these incentives influences their effectiveness (eg, how incentives are disbursed over time).

Objective: To determine if it is more effective to disburse fixed total financial incentives at a constant, increasing, or decreasing rate to encourage physical activity.

Design, setting, and participants: A 2-week randomized clinical trial was conducted from June 2 to 15, 2014, using an online platform that automatically records daily steps of pedometer-wearing users and awards points redeemable for cash. The study population comprised 3515 adult users of the online platform in the lower 70th percentile of steps taken among all users before treatment. Data analyses were performed from August 20, 2014, to February 1, 2018. Analysis was performed on an intent-to-treat basis.

Interventions: Participants were randomized to either a control group or to 1 of 3 intervention groups during the 2 weeks of the study. Participants in the control group received a constant daily rate of $0.00001 per step. The 3 intervention groups received a 20-fold incentive increase ($0.00020 per step) distributed differently during the 2 weeks of the study: at a constant, increasing, or decreasing rate. Reminder emails explaining incentive schedules were sent the day before the intervention and halfway through the 2-week intervention.

Main outcomes and measures: Change in mean daily steps during the 2-week intervention and 3 weeks after the intervention. The study had 80% power to detect a difference of 280 steps per day during the intervention at α = .05.

Results: The study included 3515 participants (879 in the control condition, 879 in the constant incentive condition, 881 in the increasing incentive condition, and 876 in the decreasing incentive condition). During the intervention, compared with participants in the control group, participants receiving constant incentives logged 306.7 more steps per day (95% CI, 91.5-521.9 steps; P = .005), those receiving decreasing incentives logged 96.9 more steps per day (95% CI, 15.3-178.5 steps; P = .02), and those receiving increasing incentives logged no significant change in steps per day (1.5 steps per day; 95% CI, -81.6 to 84.7 steps; P = .97). One week after the intervention, compared with participants in the control group, only participants receiving constant incentives logged significantly more steps per day (329.5; 95% CI, 20.6-638.4; P = .04). Two and 3 weeks after the intervention, there were no significant differences compared with participants in the control group. Overall, for each $1 spent, participants in the constant incentives group logged 475.4 more steps than those in the increasing incentives group and 429.3 more steps than those in the decreasing incentives group.

Conclusions and relevance: This study found that financial incentives for physical activity were more effective during a payment period when they were offered at a constant rate rather than an increasing or decreasing rate. However, this effectiveness dissipated shortly after the incentives were removed.

Trial registration: ClinicalTrials.gov identifier: NCT02154256.

Conflict of interest statement

Conflict of Interest Disclosures: Dr John reported receiving grants from Foundation for Human Behavior during the conduct of the study. Dr Foschini reported receiving grants from Humana Inc and grants and personal fees from Evidation Inc during the conduct of the study. Dr Milkman reported serving on the Humu Academic Advisory Board (for which she receives stock options); serving on the HealthPrize Academic Advisory Board (which was a paid position); serving on the Idomoo Academic Advisory Board (which was a paid position); and serving as a paid speaker or consultant for World 50, Interbrand, BMGI, Planned Parenthood, Morningstar, O’Reilly Media, 1st Global, Google, Educause, Gibson Dunn & Crutcher, SEI, DE Shaw, Investments and Wealth Institute, Geisinger Health, Condé Nast, and Charles Schwab. No other disclosures were reported.

Figures

Figure 1.. Study Flow Diagram

aData on step count were captured through and contingent on daily pedometer use, which varied between 82.8% and 84.2% among conditions before the intervention and 76.4% to 79.7% during the intervention. To account for missing data, we used an intent-to-treat approach in which we replaced missing pedometer data with a participant’s preintervention daily step count of more than 2000 steps stratified by day of week.

Figure 2.. Relative Incentive Rates Offered per Step

Incentive rate on each of 14 days of intervention by experimental condition, where the y-axis represents a multiplier of the control incentive rate ($0.00001/step). For example, in panel B, the constant incentive condition has a multiplier of 20 for each of 14 days, and so the daily incentive rate would be 20 × $0.00001/step = $0.00020/step.

Figure 3.. Unadjusted Mean Differences in Step Count Compared With Control by Experimental Condition