|A lesson in business: cost-effectiveness analysis of a novel financial incentive intervention for increasing physical activity in the workplace
|Dallat MA, Hunter RF, Tully MA, Cairns KJ, Kee F
This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.
This study evaluated the cost-effectiveness of the Physical Activity Loyalty (PAL) card, financial incentive scheme, to increase physical activity, in the workplace. The authors concluded that the PAL scheme was potentially cost-effective, but further research was of value to reduce the uncertainty. These conclusions were appropriate, with emphasis appropriately placed on reducing the uncertainty.
Type of economic evaluation
Cost-effectiveness analysis, cost-utility analysis
This study evaluated the cost-effectiveness of the Physical Activity Loyalty (PAL) card, financial incentive scheme, to increase physical activity, in the workplace.
All participants were given a PAL card to objectively measure their physical activity at work. Cards could be swiped at receivers placed, within their workplace grounds, along designated walking routes. All participants were given the opportunity to receive real-time feedback on their activity levels, by logging on to the trial website.
Participants were randomly assigned to one of two groups. The incentive group collected points and received rewards for their minutes of physical activity, and the no incentive group did not.
The evaluation was based on an randomised controlled trial (see Other Publications of Related Interest) conducted at Northern Ireland's main government offices, in a 300 acre park. The follow-up was six months. The authors stated that a health services perspective was taken for the cost-utility analysis, and an employer perspective was taken for the cost-effectiveness analysis.
The main effectiveness parameter was physical activity level, from the swipe card data. Participants were aged 16 to 65 years, working in the office at least four days per week, for at least six hours per day, and able to complete 15 minutes of moderate walking daily. All those in building A were allocated to incentive, and all those in building B were allocated to no incentive. Researchers were blind to allocation. The intervention lasted 12 weeks. Health-related quality-of-life and self-reported workplace absenteeism were recorded at six months. The two groups were comparable at the start for age, gender, body mass index (BMI), activity levels measured by the General Physical Activity Questionnaire (GPAQ), EQ-5D, work absenteeism, and productivity. Productivity was estimated from the EQ-5D scores, using a referenced method. Physical activity was compared at six and 12 weeks, and health-related quality-of-life and productivity were compared at six months.
Monetary benefit and utility valuations:
The utility scores were derived from the participants using the EQ-5D questionnaire.
Measure of benefit:
The primary measure of benefit, for the health services perspective, was quality-adjusted life-years, and for the employer perspective, it was the percentage point gain in productivity.
The costs were those needed to deliver each intervention, which included website development, software, hardware, sensor maintenance, and PAL card delivery for both groups. For the incentive group, the costs of negotiating incentives from local businesses, and the delivery of vouchers, were included. All the resource use and costs were from the trial. All costs were reported in UK £.
Analysis of uncertainty:
Parameter uncertainty was explored, using probabilistic sensitivity analysis. The costs were varied by ±5%, except for the cost of a research fellow, which was varied to the minimum and maximum salary. All costs were drawn from triangle distributions. Bootstrapping was used to draw the utility and productivity values from the trial sample. The results were presented in cost-effectiveness acceptability curves, and used for a value of information analysis.
The activity levels at six and 12 weeks favoured the incentive group, but the differences were not statistically significant. The incentive group had greater gains in utility at six months (0.03, 95% CI 0.01 to 0.05), compared with no incentive (0.02, 95% CI 0.00 to 0.04), but the difference was not statistically significant. The findings for productivity were similar.
The total costs were £30,800 with incentives and £26,700 with no incentives; a difference of £4,100. The QALYs gained were 2.4 with incentives, and 1.2 without; a difference of 1.2 QALYs. The incremental cost-effectiveness ratio was £2,900 per QALY gained, with a 85% likelihood of cost-effectiveness, at a threshold of £30,000 per QALY gained. The value of information analysis indicated that the cost of reducing uncertainty through additional research was far less than the value of the information gained by reducing the uncertainty.
In the employer perspective cost-effectiveness analysis, the cost of an additional percentage point of productivity was estimated to be £2,700. The authors indicated that whether an employer would be willing to pay that much for an improvement in productivity would depend on the size and success of the business, if a business had revenue of at least £270,000 they would get a return equivalent to their investment.
The authors concluded that the PAL scheme was potentially cost-effective, from both perspectives, but further research to reduce uncertainty was of value.
The interventions were sufficiently described. The authors acknowledged that they did not include no intervention or self-monitoring, as a comparator, and this could have underestimated the cost-effectiveness of the intervention. They stated that there was some evidence to support the effectiveness of self-monitoring. The park environment might not be replicable for all work settings.
The trial measured increased physical activity over six and 12 weeks, and quality of life at the start and at six months. Initial characteristics were not statistically different, but some differences were apparent, and it was not clear if there was any attempt to adjust for these differences. The intervention aimed to increase physical activity, which should improve health-related quality-of-life, but factors besides the intervention could affect health and quality of life, which could over- or under-estimate the benefits. The six-month analysis restricted the ability to judge how long the physical activity increases lasted, and to evaluate the long-term potential benefits, such as reduced disease incidence. The trial did not assess whether increasing physical activity at work increased activity in general.
The costs were presented clearly with sufficient detail, but they were limited for the health service analysis. Relevant costs, such as initial and follow-up health service use were not included. The authors acknowledged that omitting health care use was a limitation. As with the benefits, increasing physical activity could lead to decreased incidence of disease, and the six-month time horizon was too short to capture all the cost benefits of reduced disease. The year in which the costs were collected was not reported.
Analysis and results:
The results and analytic methods were transparently reported. It was unclear whether the short time horizon was sufficient to capture the full impact of the intervention, and whether the behavioural change could be maintained. The probabilistic sensitivity analysis used appropriate bootstrapping techniques to incorporate uncertainty into the utility estimates. Triangular distributions were assigned to the costs, with assumptions for the minimum and maximum values. These limited the analysis and the uncertainty is unlikely to have been fully characterised. Appropriate techniques were used to assess the value of information, and the authors' conclusions for the value of reducing uncertainty were accurate. A comprehensive discussion of the uncertainty in productivity measurement was presented, with the acknowledgement of some limitations.
The authors acknowledged that the main limitation of their study was that not all relevant interventions were included, which could inflate the cost-effectiveness ratio estimates. Further research to reduce the high levels of uncertainty was required.
Funded by the National Prevention 384 Research Initiative (NPRI) and their funding partners, and the Department for Employment and Learning, Northern Ireland.
Dallat MA, Hunter RF, Tully MA, Cairns KJ, Kee F. A lesson in business: cost-effectiveness analysis of a novel financial incentive intervention for increasing physical activity in the workplace. BMC Public Health 2013; 13: 953
Other publications of related interest
Hunter RF, Tully MA, Davis M, Stevenson M, Kee F. Physical activity loyalty cards for behavior change: a quasi-experimental study. American Journal of Preventive Medicine 2013; 45: 56-63.
Subject indexing assigned by NLM
Adolescent; Adult; Aged; Cost-Benefit Analysis; Female; Health Behavior; Humans; Ireland; Male; Middle Aged; Motor Activity; Occupational Health Services; Reward; Workplace
Date bibliographic record published
Date abstract record published