Providing patients with financial incentives has been shown to promote exercise, weight loss, adherence to medical advice, and abstinence from smoking and other drug use. However, research on incentives to date has generally viewed them as if they were all-or-nothing, comparing behaviors in the presence versus the absence of incentives, and assuming that they worked similarly among different types of people. Yet, concepts from behavioral economics suggest that how and to whom incentives are delivered may affect their impact substantially. Thus, comparing incentive structures (e.g., should we create potential for gains vs. losses, or target individuals vs. groups?), and identifying patients for whom they work particularly well, may elucidate mechanisms by which incentives alter behavior and inform the design of behavior-modifying interventions. This study seeks to compare 4 incentive structures that are based upon principles of behavioral economics, contingency management, and social network theory. We will examine how well these 4 incentive structures promote smoking cessation because smoking is the leading cause of preventable death in the United States, roughly three-quarters of American smokers wish to quit, and yet only 2-3 percent actually do quit each year. In a 5- arm randomized trial among 2,185 Walgreens employees who smoke, we will compare a usual care group consisting of smoking cessation classes and web-based education with 4 incentive structures: (1) individual rewards (fixed payments for an individual's success), (2) individual deposits (fixed losses for an individual's failure), (3) collaborative rewards (payments to successful group members that increase with increasing group success rates), and (4) competitive deposits (redistribution of deposited money from group members who fail to group members who succeed). Our established partnership with Walgreens and our recent development of an NIH-supported, web-based infrastructure for behavioral research make such a large trial feasible. This study seeks to answer several questions: (1) How do incentives based on the potential for loss compare with those based on the potential for gain in terms of their acceptance (uptake rates), efficacy (success rates among those who accept them), and effectiveness (success rates among all people to whom incentives are offered)? (2) How do incentives with group-oriented versus individual-oriented payouts compare in terms of their acceptance, efficacy, and effectiveness? (3) What individual characteristics - such as income, strength of present-biased preferences, or number of environmental substitutes for smoking - modify these effects? We will use an innovative design to answer these questions in tandem by adapting randomization probabilities to each arm's uptake rate and using instrumental variable analyses to detect mechanisms by which incentives work. Thus, this work has a high potential for advancing the science of behavior change and for improving public health by enhancing strategies to counter behaviors that cause most preventable deaths in the U.S.
Financial incentives have been shown to motivate a broad range of health-promoting behaviors, but the optimal ways to incentivize behavior modification are unknown. In a randomized clinical trial among 2,185 Walgreens employees who smoke, the investigators will compare how well 4 different methods for providing incentives work to promote smoking cessation, and how broadly these approaches are accepted. This work will both clarify mechanisms by which incentives alter behaviors, and will identify better ways to reduce smoking and other important causes of preventable death.
|Halpern, Scott D; French, Benjamin; Small, Dylan S et al. (2016) Heterogeneity in the Effects of Reward- and Deposit-based Financial Incentives on Smoking Cessation. Am J Respir Crit Care Med 194:981-988|
|Halpern, Scott D; French, Benjamin; Small, Dylan S et al. (2015) Randomized trial of four financial-incentive programs for smoking cessation. N Engl J Med 372:2108-17|
|Kahan, Brennan C; Harhay, Michael O (2015) Many multicenter trials had few events per center, requiring analysis via random-effects models or GEEs. J Clin Epidemiol 68:1504-11|
|Harshfield, Eric; Chowdhury, Rajiv; Harhay, Meera N et al. (2015) Association of hypertension and hyperglycaemia with socioeconomic contexts in resource-poor settings: the Bangladesh Demographic and Health Survey. Int J Epidemiol 44:1625-36|
|Brown, J C; Harhay, M O; Harhay, M N (2015) Physical function as a prognostic biomarker among cancer survivors. Br J Cancer 112:194-8|
|French, Benjamin; Small, Dylan S; Novak, Julie et al. (2015) Preference-adaptive randomization in comparative effectiveness studies. Trials 16:99|
|Harhay, Michael O; Wagner, Jason; Ratcliffe, Sarah J et al. (2014) Outcomes and statistical power in adult critical care randomized trials. Am J Respir Crit Care Med 189:1469-78|
|Brown, Justin C; Harhay, Michael O; Harhay, Meera N (2014) Walking cadence and mortality among community-dwelling older adults. J Gen Intern Med 29:1263-9|
|Kerlin, Meeta Prasad; Halpern, Scott D (2012) Twenty-four-hour intensivist staffing in teaching hospitals: tensions between safety today and safety tomorrow. Chest 141:1315-1320|
|Halpern, Scott D; Asch, David A; Volpp, Kevin G (2012) Commitment contracts as a way to health. BMJ 344:e522|