Pilot Randomized Control Trial of Financial Incentives for Smoking Cessation

Overview

Financial incentives may be more effective than other approaches to smoking cessation, but research is needed to identify the optimal structures of incentives. The investigators will conduct a pilot randomized trial comparing different incentive structures for smoking cessation. Collaborating with Walgreens leadership, the investigators will enroll their employees in this RCT using the investigators new web-based research infrastructure (called the Way to Health platform, and approved as a Prime protocol by the IRB as reference # 811860). Hypothesis: providing patients with financial incentives can promote healthier behaviors (e.g.: quit smoking).

This study is designed to explore the potential efficacy of using internet-based incentive programs to promote smoking cessation. We also hope to document that we can recruit participants through a web site, as well as investigate what forms of incentives show the most promise. We seek to achieve three specific objectives: (1) assess our ability to recruit participants for several different types of intervention-based smoking programs; (2) document the feasibility of using the internet to accrue and disburse incentive payments; (3) obtain preliminary assessments of which incentive structures show the most promise for future study. Potential participants will be directed to the Way to Health research portal and will be assessed for eligibility. The web application will automatically verify eligibility based on the results of the survey. If a person is deemed eligible, he or she will be invited to consent to participate in a non-pharmacologic smoking cessation study lasting 3 months. Participants who consent will be randomized automatically to one of the 5 intervention arms (including the no-incentive arm) by the web application. Once the participant consents and submits the necessary financial information, he will gain access to the participant interface. At this point, they will receive detailed descriptions of their assigned incentive structures (or usual care provisions), will indicate their acceptance of their assigned structure and, for those in arms 3, 4, or 5, make their required deposits. Also, he will be prompted to take an initial survey on his baseline smoking habits. The participant will be able to log into this interface at anytime and see a graph of his self-reported smoking habits along with a graph of approximately how much money he has saved compared to his baseline smoking levels. Participants will indicate their preferred method of contact: email, text-messaging, or both. Every week the participant will receive an email or text-message reminder to log in and complete a weekly smoking report. Participants can also log in at any time and report on their smoking habits for a given day or days to improve accuracy. After three months all subjects will be e-mailed to prompt them to log into the web application to take a final smoking habits survey. Participants who report having remained smoke free for the last month of the study will submit saliva (or urine, for participants using any form of nicotine replacement therapy) samples to test for cotinine (or anabasine) to verify that they have quit. Once this has been verified, the participant will be sent whatever financial incentives he has earned and any deposit will be returned. The design of this study differs from a traditional RCT in two important ways. First, because the goal of this RCT is to focus on mechanisms of behavior change, we include a primary per-protocol analysis designed to determine incentive structures' efficacy by analyzing only participants who accept their assigned incentive structure at the time of randomization. In this analysis, the randomization arm will be entered into the analytic model as an instrumental variable to mitigate potential selection bias. As in a traditional RCT, we will also analyze each incentive structures' effectiveness using an intention-to-treat analysis that includes all randomized participants, regardless of whether they accept the randomly assigned arm. The second difference is that because acceptance rates of the different structures may differ (as we will test in Aim II), we will adapt the probabilities of being randomized to each arm on a weekly basis during the enrollment period to achieve the target numbers of participants in the per-protocol analysis.