A discrete choice experiment for financial incentive-based physical activity programs among adults with type 2 diabetes

Abstract Background Physical activity (PA) is essential for type 2 diabetes mellitus (T2D) management, yet many adults with T2D are inactive. Financial incentives can increase participation in PA programs, but few studies have investigated which incentive program attributes are preferred by patients. The purpose of this study was to identify the preferred financial incentive program design for PA program participation among patients with T2D. Methods A discrete choice experiment (DCE) was conducted with adults aged 18–75 years, with a physician’s diagnosis with T2D, and receiving care through UC San Diego Health Family Medicine. Participants completed an online DCE with fifteen choice sets of two hypothetical PA programs presented side-by-side, each with experimental design-generated combinations of different levels of seven incentive attributes: incentive frequency (e.g., daily, program completion), collaboration (individual or team performance), direction (gain or loss framed), size, competitiveness (compared to others’ activity levels), type (cash or voucher), and recipient (self or charity). For each pair, participants selected which program they preferred and indicated if they would enroll in that program (yes/no). Part-worth utilities of each attribute level were estimated using hierarchical Bayes. Results Two hundred seventy-seven participants completed the study (mean age = 60.2years ± 11.51; 58% female). Incentive size had the strongest effect on preference (24%), followed by the financial incentive recipient (18%). Incentive type had the smallest effect on preference (5%). Programs offering incentives that are larger (B = 1.12, 95% CI [0.96, 1.29]), given at PA program completion (B = 0.42, 95% CI [0.25, 0.60]), based on individual performance (B = 0.73, 95% CI [0.61, 0.86]), paid to the individual (B = 1.10, 95% CI [0.88, 1.25]), and structured as non-competitive (B = 0.52, 95% CI [0.41, 0.64]) were preferred. A simulation model estimated that a majority (84%) of participants would enroll in a hypothetical incentives-for-PA program with these preferred attribute levels, compared to 15% enrolling in a program with the least preferred levels of each attribute. Conclusions The majority of patients with T2D would be projected to enroll in an incentive-based PA program if designed based on all preferred attribute levels. Testing such PA programs in real world settings will determine actual program uptake and adherence.