Adapt2Quit – A Machine-Learning, Adaptive Motivational System: Protocol and Baseline Demographics for a Randomized Controlled Trial for Socio-Economically Disadvantaged Smokers (Preprint)

BACKGROUND People who are socioeconomically disadvantaged (SED) have high smoking rates and face barriers to participation in smoking cessation interventions. One strategy to reach and engage these individuals is by delivering computer-tailored messages via text messages. Computer-tailored health communication, which is focused on finding the most relevant messages for an individual, has been shown to effectively promote behavior change. Our team has developed a machine learning approach (the Adapt2Quit recommender system) to further enhance the relevance of computer-tailored messages. Conceptually, the approach is similar to those used by companies like Amazon, which tailor their content to continuous user feedback. Our pilot work has shown that the Adapt2Quit recommender system increased the relevance and potentially the effectiveness of the messages to SED individuals who smoke. However, the Adapt2Quit recommender system has not been tested with SED individuals in randomized trials. OBJECTIVE This study protocol describes our randomized trial to test whether the Adapt2Quit recommender system will increase smoking cessation among SED individuals who smoke. METHODS SED individuals who smoke were recruited from health systems in diverse geographic areas and then randomized. Participants in the intervention received the Adapt2Quit recommender tailored messages for six months. Intervention participants also received the biweekly texting facilitation messages asking if participants were interested in being referred to the Quitline. Interested participants were then actively referred to the Quitline by study staff. Intervention participants also received biweekly messages assessing their current smoking status. Comparison participants did not receive the recommender messages but received the biweekly texting facilitation and smoking status assessment. Study participants were blinded to group assignments. The primary outcome is the 7-day point prevalent smoking cessation at six months, biochemically verified via carbon monoxide testing. RESULTS The Adapt2Quit study was funded in April 2020 and is still ongoing. As of submission of the manuscript, we have completed recruitment of SED individuals (n=757 participants). The sample consists of 64% females, 35% are Black or African American, 51% White, and 16% identified as Hispanic or Latino. Forty-nine participants reported some high school education or were high school graduates. Seventy percent smoked their first cigarette within 30 minutes of waking, and half had stopped smoking for at least one day in the past year, and 13% had visited a smoking cessation website. Most participants reported prior use of Nicotine Replacement Therapies (NRT) or other medications, and 17% had called the Quitline before participating in the study. We anticipate the follow-up will be completed in November 2024. CONCLUSIONS Methods to increase the effectiveness of computer-tailored interventions, which are widely disseminable via texting, are important. We have recruited a diverse sample of SED individuals and designed a rigorous protocol to evaluate the Adapt2Quit recommender system. CLINICALTRIAL NCT04720625