Effectiveness and cost-effectiveness of a loyalty scheme for physical activity behaviour change maintenance: results from a cluster randomised controlled trial

Background: We evaluated the effectiveness and cost-effectiveness of a loyalty scheme based intervention involving rewards for increasing physical activity in public sector employees. Methods: A cluster randomised wait-list controlled trial in public sector organisations in Northern Ireland. We randomly assigned clusters (1:1) using a computer generated random sequence. Researchers were masked to allocation, but participants were not. Employees aged 18-65 years with no self-reported medical contraindications to physical activity were included. The Physical Activity Loyalty Scheme (PAL) intervention was based on high-street loyalty cards where participants earned points for minutes of activity that could be redeemed for rewards, complemented by evidence-based behaviour change techniques. The primary outcome was objectively measured mean steps/day at 6 months using a validated pedometer (Yamax Digi-Walker CW-701) over 7 days, assessed with intention to treat analysis. Secondary outcomes included health, mental wellbeing, quality of life, work absenteeism and presenteeism, and use of healthcare resources. Cost-effectiveness, cost-benefit and mediation analyses were conducted. Trial registered with Current Controlled Trials, number ISRCTN17975376. Results: Between September 2014 and October 2015, we recruited and randomly assigned 37 clusters (from nine organisations; mean clusters per organisation=four) and 853 participants to the intervention (n=19 with 457 participants) or control group (n=18 with 396 participants). Primary outcome data were available for 249 (544%) intervention and 236 (596%) control participants. Mean steps/day were significantly lower in the intervention vs control group (adjusted mean difference=-336, 95% CI: -612 to -60, p=002) at 6 months. Participants redeemed only 39% (SD 43%) of their earned points. Using the Quality Adjusted Life Year outcome, the intervention was not cost effective from an NHS/PSS perspective. A net cost analysis from an employer perspective demonstrated the intervention group was associated with a mean of 297h less absenteeism over a 4 week period (p=062), which could result in net savings ranging from pound 66 to pound 735 depending on the wage rate employed. At 4-weeks post-baseline there were significant increases in identified regulation, integrated regulation, intrinsic motivation, social norms and intentions in intervention compared to control participants. Conclusions: Our mixed results pose challenges that are too infrequently exposed in public heath intervention trials. Although the intervention successfully altered several hypothesised mediating constructs it did not translate into long-term behaviour change. Our incentive level may have been too low to incentivise change, despite being designed a priori by a Contingent Valuation Survey. There were also major re-structuring of several organisations which presented significant implementation challenges, and technical limitations.