Micro-randomized pilot trial of an app-based smoking urge reduction intervention for young adults.

Abstract

Background: Cigarette smoking remains the leading preventable cause of illness and death in the United States and young adults are a priority population. This study aimed to investigate the feasibility of conducting a pilot micro-randomized trial (MRT) among young adults, using smartphone messages to reduce smoking urges in high-risk situations and combining Ecological Momentary Assessments (EMA) and geofence-based delivery.

Methods: Participants were recruited online and completed surveys on Qualtrics and an EMA app to report smoking situations over 14 days (assessment phase). Assessment phase data were used to generate individual risk profiles [by combining timestamps, Global Positioning System (GPS), and self-reported data] with the highest likelihood of smoking for each participant. We used geofencing to generate geospatial buffers around these high-risk locations and intervention messages were triggered when a mobile device entered a geofence during specific time windows. In the following 30-day intervention phase, participants were prompted to complete up to 3 geofence-triggered EMAs daily. Each geofence-triggered EMA was followed by an intervention message and the type of message (distraction, acceptance, control) was randomized at each time point (within-subject). Urge level was the primary proximal outcome assessed in a follow-up EMA up to 15 minutes after message delivery. Analyses investigated the feasibility of study procedures, including geofence triggers, within-subject randomization, EMA survey completion rates, and smoking urge reports before and after message delivery. Cigarette smoking cessation and reduction outcomes at 45-day follow-up were also investigated.

Results: A total of 8 participants were included in analyses (mean age 26.3 years; 50% male; 37.5% non-Hispanic White). In the assessment phase, participants completed between 4 and 51 real-time smoking reports. At least 2 geofences were created per participant, with a maximum of 9 geofences for one participant. In the intervention phase, between 11 and 90 geofence EMAs were triggered per participant. Compliance with EMAs was high (90.1% of geofence-triggered, 93.9% of follow-up EMAs). Within-subject randomization was successful and urge ratings declined from pre- to post-message assessments for control (k=98; mean difference =0.20), acceptance (k=99, mean difference =0.36) and distraction messages (k=94, mean difference =0.37). At 45-day follow-up, 1 participant (12.5%) reported no cigarette smoking during the past 7 days and abstinence was confirmed remotely using saliva cotinine testing. Half of participants (n=4, 50%) reduced their number of cigarettes per day (CPD) from baseline to follow-up by 50% or more. Findings also demonstrated the need for robust prevention of fraudulent research participant enrollment, as 10 participants were excluded due to GPS locations outside of the United States.

Conclusions: Results indicate potential technical feasibility of an app-based MRT using intervention messages triggered by geofence locations. Findings will inform a fully powered MRT to investigate message efficacy to reduce smoking urges and smoking in young adults.

Trial registration: ClinicalTrials.gov (NCT05991934).