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

Introduction

Tobacco smoking remains the leading preventable cause of illness and death in the United States, with particularly high smoking rates among young adults (18–30 years old) (1). As smoking initiation rates are highest in young adulthood (2), achieving early smoking cessation is needed to realize a substantial decrease in related health risks (3). Unfortunately, current professional cessation support is not optimally employed and is insufficiently utilized among young adult smokers (4,5), despite high interest in quitting smoking (6). This poses a new challenge to smoking cessation efforts, as novel interventions must be developed to reach young adults and efficiently deliver evidence-based cessation tools (7,8). With 98% of Americans aged 18–29 years possessing a smartphone (9), digital delivery of smoking cessation interventions, especially in the form of “in-the-moment” coping strategies, has the potential to increase availability of cessation aids. However, as of now, few digital cessation tools utilize evidence-based interventions (10-12) and most fail to deliver personalized content (13-15). More work on developing efficacious smartphone-based smoking cessation interventions for young adults is needed.

A promising approach of smartphone-based smoking cessation interventions is to target smoking urges, or strong desires to smoke, which have been shown to be major triggers of smoking in our (16-18) and others’ (19-22) prior work. As increasing urge levels increase the risk of smoking (23), especially among light smokers (24), adequate intervention tools are particularly in demand to address the needs of young adults, who predominantly smoke a limited number of cigarettes daily or non-daily (25). Clinical practice guidelines for smoking cessation (26) recommend using cognitive behavioral therapy (CBT) to teach coping strategies for smoking urges (27). Distraction is a key skill in CBT and involves redirecting attention and focus away from maladaptive thoughts toward alternative stimuli or activities. Evidence supports the use of distraction techniques to help smokers manage and alleviate the intensity of smoking urges (28). In contrast to CBT, newer mindfulness-based forms of psychotherapy, including Acceptance and Commitment Therapy (ACT), promote psychological flexibility through accepting negative experiences (29). In the case of smoking urges, ACT-based messages might train and reinforce the willingness to experience smoking urges without the need to act on them. While there is evidence for the efficacy of both CBT (26) and Mindfulness/ACT smoking cessation interventions (30), it is unclear if these approaches are efficacious when implemented in real-time and with young adults.

In order to test the real-time impact of different smartphone-based intervention messages on smoking urges, a trial design is needed that differs from a traditional between-subject randomized controlled trial (RCT). A micro-randomized trial (MRT) (31-33) is a novel factorial design that addresses key limitations of traditional trials in dynamic and context-sensitive settings, including smartphone-based smoking cessation. Unlike static RCTs, MRTs include repeated within-subject randomizations, ensuring that intervention effects are assessed across diverse situations encountered by participants. In order to target unique high-risk locations for individual participants, these interventions can leverage geofence-triggered intervention message delivery [virtual boundaries by Global Positioning System (GPS), that trigger a message when a mobile device enters the area] (34,35). By combining geofences and real-time data collection through Ecological Momentary Assessment (EMA) (16,17,36,37), an MRT can target high-risk locations for smoking urges, provide data on intervention-message fit, support causal inference, and inform optimal strategies for tailoring messages to enhance smoking cessation outcomes among young adults. In this context, geofences can encompass both public and private spaces, including a participant’s home or workplace, if these are identified as high-risk smoking locations based on participant data, allowing the intervention to address urges across a wide range of settings.

Several recent studies have used MRT designs to investigate intervention effects on mental health and substance use outcomes, utilizing intervention content based on CBT and ACT. For instance, CBT text messages significantly improved mood ratings among adults during the coronavirus disease (COVID)-19 pandemic (38). In contrast, an MRT to assess a mobile ACT intervention for individuals with bipolar disorder resulted in increased scores on self-reported manic and depressive symptoms (39). Focusing on cannabis outcomes, another study compared mindfulness and distraction coping messages for managing urges (40). While the intervention messages demonstrated high feasibility and engagement, neither message type significantly reduced urges compared to control messages (40). However, compared to the current work, this previous study did not use geofences to trigger EMA assessments and intervention message delivery and focused on a different substance use outcome. Taken together, these studies highlight the feasibility of MRTs to investigate real-time effects of mobile intervention messages, but also show mixed efficacy results.

Building on this existing research and to address limitations in the literature to date, the current study sought to investigate the feasibility of conducting an MRT to reduce smoking urges among young adults. Specifically, this study aimed at testing evidence-based distraction (CBT) and acceptance messages (Mindfulness/ACT) for specific high-risk situations based on geofence-triggered intervention delivery. We present this article in accordance with the CONSORT reporting checklist (available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-25-17/rc).

Methods

Procedure

Participants were recruited through Facebook, Instagram, Reddit, and X advertisements that linked to a Qualtrics screening survey. After consenting to participate but before being enrolled in the study, participants were required to email or text study staff a picture of a valid identification (e.g., driver’s license) that had their name, picture, and birthdate to validate their age and identity. All participants then completed a baseline survey on the online survey platform Qualtrics; and the study team created an account for each participant on the MetricWire Catalyst platform that was used to collect EMA and GPS data and conduct the MRT. This platform has the capability to develop, deploy, and conduct MRT studies using an app on participants’ own devices that participants download from the Apple App Store or Google Play Store. MetricWire is Health Insurance Portability and Accountability Act (HIPAA) compliant, thus minimizing participant risk for loss of privacy. We have successfully used their system in our previous EMA studies (41-47). After completion of the MRT (45 days after the baseline survey), participants completed a final follow-up survey on Qualtrics and were invited to participate in a telephone interview to share their user experience (UX) with the study. Participants received up to $228 for participating in the study: $10 for completing the baseline survey and starting data collection using the smartphone app; $2 for each day of participation in the EMA surveys ($2 × 44 days = $88) plus an extra incentive of $90 if they completed at least 75% of the prompted EMAs; and $40 for participating in the interview. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. All study procedures were approved by the Institutional Review Board of the Johns Hopkins Bloomberg School of Public Health (No. IRB00013413) and the pilot trial was registered on ClinicalTrials.gov (NCT05991934). All participants who met eligibility criteria had to provide online consent prior to study involvement. All intervention messages, assessments, and the interview guide are available on OSF (https://osf.io/4ux8q/).

Assessment phase—EMA data collection (14 days)

EMA data collection

Before EMA data collection, participants received detailed instructions on how to use the EMA study app. Participants used their own smartphones and the study app to collect data over the course of 14 days (EMAs and smartphone GPS location sensor data). Participants completed 3 randomly prompted EMA surveys per day and were instructed to report every time they smoked a cigarette. A random subset of these cigarette reports triggered up to 3 EMA smoking survey prompts per day. Participants also completed one daily diary per day, prompted in the morning, which assessed cigarette and other tobacco product use, alcohol, cannabis, and drug use on the previous day. The 14-day time interval was chosen since smoking patterns among young adults are frequently characterized by social smoking (25,48) and patterns differ between weekdays and weekends, with especially light smokers smoking more heavily on weekends compared to weekdays (49).

Purpose of the assessment phase

The assessment phase served multiple purposes. First, it allowed participants to practice using the study app that also delivered the intervention messages during the subsequent intervention phase, and second, the collected data allowed us to generate an individual risk profile with regard to the time of day and location (by combining timestamps, GPS data, and self-reported data) associated with the highest likelihood of smoking for each participant. We used geofencing to generate geospatial buffers around these high-risk locations. In combination with time-of-day information to target high-risk time periods for smoking, geofences can trigger delivery of intervention messages when a mobile device (i.e., a participant) enters a specific area (i.e., a smoking location) during one of the high-risk time windows. We used our established protocol for determining geofences to trigger intervention messages that we have developed in previous work (34).

Intervention phase—MRT with intervention message delivery (30 days)

Intervention messages came from several previous studies (35,50,51) and were refined and combined with image content from free stock photo websites (Pexels, Unsplash) to appeal to young adults. We used a total of 124 intervention messages (58 CBT, 66 Mindfulness/ACT). Intervention message acceptability results have been reported previously (52). See Figure 1 for intervention message examples.

Figure 1 Intervention message examples for distraction and acceptance messages, including image and text content.

The micro-randomized pilot trial (31,53,54) was conducted to test all procedures in preparation for a fully-powered MRT, which will determine if CBT and Mindfulness/ACT messages are superior to control messages in reducing the primary outcome momentary smoking urges, and will follow the same procedures as outlined here. Based on participants’ EMA and GPS data collected in the initial 14-day assessment phase, intervention messages were delivered during time periods and at high-risk locations for smoking. In the intervention phase, participants were prompted to complete up to 3 geofence-triggered EMAs per day for a total of 30 days. Each EMA was followed by an intervention message and the type of message (CBT, Mindfulness/ACT, control) was randomly selected at each time point with equal likelihood (1:1:1; within-subject randomization; see Figure 2). Randomization of messages was automatically conducted by the MetricWire platform. Control messages thanked participants for completing an assessment and did not include any image content. Proximal outcomes were assessed in another EMA prompted randomly between 5–15 minutes after message delivery and included urge levels; smoking or other tobacco product use since the initial geofence-triggered EMA; affect; stress; and an evaluation of the last message (e.g., perceived usefulness of the message, completion of suggested activity/intervention). In addition, participants continued completing one brief retrospective daily diary each morning. Similar to the assessment phase, this retrospective daily diary assessed the previous day’s cigarette and other tobacco product use, alcohol, cannabis, and drug use. The 30-day time interval was chosen opportunistically, striking a balance between maximizing data collection opportunities for the MRT design and limiting participant burden. Several recent MRTs targeting health behavior change have used the same or similar intervention lengths (40,55).

Figure 2 Delivery of intervention messages. ACT, Acceptance and Commitment Therapy; CBT, cognitive-behavioral therapy; EMA, Ecological Momentary Assessment.

Follow-up survey and interview

At the end of the intervention phase (45 days after the baseline survey), participants completed a final follow-up survey on Qualtrics and were invited to participate in a brief UX telephone interview. The interview topics explored participants’ experiences with the study and included questions on participants’ ease of responding to EMA prompts, their ability to report real-time cigarette use, and barriers they encountered. Feedback on intervention messages was also assessed, focusing on their relevance to participants’ behavior, mood, and environment, as well as message impact on smoking, stress, and daily life.

Participants

Participants were young adult men and women who: (I) lived in the U.S.; (II) read English; (III) were between 18 and 30 years of age; (IV) owned a smartphone with an iOS or Android operating system; (V) had smoked ≥100 cigarettes in their lives; (VI) currently smoked at least 1 cigarette per day on 3 or more days of the week; and (VII) were planning to quit smoking within the next 30 days.

A total of 289 participants were screened, of which 54 participants were eligible based on our inclusion criteria and provided informed consent (Figure 3). Study staff verified the identity of 31 participants, 23 completed the baseline assessment, and 22 initiated EMA data collection. Of those initiating EMA, 14 were excluded because they were either fraudulent participants who did not reside in the US based on their GPS data collected on the EMA app (n=10) or were removed after the initial assessment phase because they did not provide enough data for our algorithm to successfully construct geofences of high-risk situations for smoking that were required for the intervention phase (n=4). Eight participants completed follow-up surveys on Qualtrics, and brief user-experience telephone interviews were completed by a total of 6 participants (length 10–23 minutes). Thus, 8 participants were included in the quantitative analysis for the current pilot study, including EMA and follow-up data, and 6 participants were included in UX analyses of interviews. The 4 participants who were removed before entering the intervention phase due to a lack of sufficient data for geofence generation did not significantly differ from the analytical sample on baseline characteristics, with two exceptions: Those removed were younger [t(10)=4.8; P<0.001] and reported fewer smoking days in the past 30 days [t(10)=3.7; P<0.01], compared to the analytical sample.

Figure 3 Study flow diagram. EMA, Ecological Momentary Assessment.

Measures

Outcomes

The main objectives of this pilot trial were to investigate the feasibility of creating multiple individualized geofence locations per participant and assess the number of geofence-triggered EMAs and intervention message deliveries for each participant. The goal was to trigger up to three geofence EMAs and intervention messages per participant per day based on the geofence locations and times of day established during the initial assessment phase. Additionally, we analyzed participant compliance with geofence-triggered and subsequent follow-up EMAs, and calculated differences in the proximal outcome of self-reported smoking urge from pre- to post-message delivery. Smoking urge was recorded with a single item (“Craving a cigarette or tobacco product?”) with responses recorded on a 5-point Likert scale (1 “very low” to 5 “very high”), in line with previous research of our group and the work of others (16-18,21,24,45,56).

The distal outcome of smoking abstinence (7-day point prevalence abstinence) was assessed at 45-day follow-up using self-report and saliva confirmed (assessed using NicAlert saliva cotinine test strips mailed to participants and photo confirmation) (57-60). The follow-up survey also assessed self-reported number of smoking days and cigarettes per smoking day in the past 7 days to calculate the number of cigarettes per day (CPD) at baseline and follow-up for the distal outcome of reduction in CPD, including 50% or greater reduction.

Baseline variables

The baseline assessed basic demographics, including age, gender, sexual orientation, race/ethnicity, and education. Moreover, the baseline assessed current cigarette smoking (daily, non-daily), number of smoking days out of the past 30, number of cigarettes per smoking day, smoking within the first 30 minutes of waking as a measure of nicotine dependence (60,61), as well as any current (past 30-day) use of other tobacco products (e.g., e-cigarettes, cigarillos, smokeless tobacco, and hookah).

Statistical analysis

Geofences were defined algorithmically for each participant based on self-reported real-time cigarette reports during the 14-day assessment phase. These reports were tagged with timestamped GPS coordinates, which were extracted and processed to identify clusters of smoking behavior. For each participant, the convex hull was computed around their set of valid smoking locations to create individualized geofence polygons. These geofences were linked to specific two-hour time intervals when smoking was most likely to occur, based on the temporal distribution of cigarette reports in the assessment phase. This personalized, spatiotemporal geofencing approach ensured that intervention messages were delivered at locations and times empirically associated with increased smoking risk for each participant. The full R script used for this procedure is available on OSF (https://osf.io/4ux8q/), and builds on protocols outlined in our prior work (34).

All statistical analyses were performed using Stata SE 18 and R v 4.4.1. Descriptive statistics were used to summarize baseline characteristics, smoking behaviors, and compliance metrics. The primary analysis focused on assessing the feasibility of study procedures, including how many geofences were created for each participant, the number of geofence-triggered and follow-up EMAs, and within-subject randomization of intervention messages. Compliance with geofence-triggered and follow-up EMAs was summarized as percentages and reported for each participant.

For the intervention phase, differences in smoking urge ratings between pre- and post-message delivery were calculated for each message type (CBT, Mindfulness/ACT, Control). Mean differences and standard deviations (SD) were computed and compared descriptively, as the pilot study was not powered to detect statistical significance.

Smoking cessation outcomes, including self-reported 7-day point prevalence abstinence and reduction in CPD, including a 50% or greater reduction, were evaluated at follow-up using descriptive statistics.

Coding of brief user-experience telephone interviews was conducted at first deductively, based on interview guide, then inductively with sub-themes identified from participant responses. Authors J.T., J.J.C.W., and J.J.H. coded all interviews, with one author conducting the initial coding and the other author verifying accuracy. Author J.T. then reviewed all codes and excerpts and wrote an overall summary of interview findings. Codes addressed pre-determined themes based on sets of interview questions and included the following: overall participant experience in the study and app; experience with reporting cigarettes; experience with geofence-triggered EMAs; experience with intervention messages; and experiences with smoking cessation. These themes are presented following established thematic analysis recommendations (62). The sample size for follow-up interviews was based on UX design recommendations that a sample size as small as 5 users will detect 85% of usability errors in a given system (63,64).

Results

Sample description

Participant characteristics are reported in Table 1 (N=8). The mean age of participants was 26.3 years old (SD =1.2), 50.0% identified as male, and 75.0% identified as straight. In this sample, 37.5% were non-Hispanic (NH) White, 25.0% were NH Black/African American, and 25.0% were Hispanic or Latino, and 50.0% of the sample had a bachelor’s or master’s degree. Most participants (87.5%) reported daily cigarette smoking. The average number of days of cigarette smoking (out of the past 30 days) was 28.1 (SD =3.2), with an average of 7.4 (SD =5.7) cigarettes per smoking day. Smoking within the first 30 minutes of waking as a proxy for nicotine dependence was reported by 12.5% of participants and almost all participants (87.5%) reported current (past 30-day) use of tobacco products other than cigarettes (e.g., e-cigarettes, cigarillos, smokeless tobacco, hookah).

MRT results

Table 2 displays the number of cigarettes per smoking day reported at baseline (range, 2–20) and the number of real-time cigarette reports each participant contributed during the initial 14-day assessment phase (range, 4–51). Based on these data, geofences were created using our algorithm (34). At least 2 geofences were created per participant, with a maximum of 9 geofence locations. The number of geofence-triggered EMAs per participant over the entire 30-day intervention phase ranged from 11 to 90, with between 11 and 29 days during which any geofence EMAs were triggered.

Participant compliance with geofence-triggered as well as follow-up EMAs is displayed in Table 3. Compliance was high among all participants and between 81.3% and 100.0% of geofence-triggered EMAs were submitted. As for follow-up EMAs, compliance ranged from 84.6% to 100.0%.

Table 4 displays participant self-reported smoking urge ratings in geofence-triggered and follow-up EMAs by message condition (CBT, Mindfulness/ACT, Control). Due to the space-time dynamics of the participants entering multiple constructed geofences (e.g., if they entered several geofences in a short period of time), participants were able to submit assessments out of order (i.e., submission of first geofence-triggered EMA, followed by submission of second geofence-triggered EMA, followed by submission of the first follow-up EMA). Therefore, only EMA pairs (geofence-triggered and subsequent follow-up EMA) submitted in succession were included in this analysis (k=15 assessments excluded; k=291 assessments analyzed). The assessments reflect urge ratings before and after the intervention or control messages were delivered. Urge ratings numerically declined in both control (k=98, mean difference =0.20) and combined intervention conditions (k=193, mean difference =0.37), with a numerically larger reduction in urge ratings in intervention conditions. There was a similar decrease in mean smoking urge ratings for CBT (k=94, mean difference =0.37) and Mindfulness/ACT (k=99, mean difference =0.36) messages. Of the analyzed assessments, 94 (32.3%) were for CBT messages, 99 (34.0%) were for Mindfulness/ACT messages, and 98 (33.7%) were for control messages, demonstrating the balance obtained by randomization.

Follow-up results

At 45-day follow-up, 1 participant (12.5%) reported no cigarette smoking at all during the past 7 days. This participant was sent a NicAlert saliva cotinine test strip and results confirmed nicotine abstinence. The average number of self-reported CPD reduced from a baseline of 7.1 (SD =5.8) to 3.2 (SD =2.2), which was not statistically significant [t(7)=1.6; P=0.15]. A 50% or greater reduction in number of CPD from baseline to follow-up based on participant self-reports was calculated for 4 participants (50.0%).

UX follow-up interviews with participants after study completion

UX follow-up interviews were conducted with six participants and analyzed thematically following Braun and Clarke’s approach (62). Five themes were identified that collectively captured participants’ experiences with the study and geofence-based intervention delivery.

Overall UX with the study and app

Most participants (5/6) described the app as straightforward to use when their phone was readily accessible, with logging and responding to prompts becoming routine over time. As one participant noted, “It was easy to log in, to log the cigarette” (ID2). All participants found at least some benefit in taking part, whether by prompting reflection or introducing new strategies to manage smoking. However, five reported technical issues, including delayed notifications or repeated prompts and questions.

Experience with reporting cigarettes

All participants said they occasionally forgot to log cigarettes, typically when distracted or away from their phone. One participant explained: “I used to forget sometimes” (ID1). For some, logging was effortless when the phone was nearby, but others found it challenging to remember in the moment.

Experience with location-triggered assessments

Half of the participants (3/6) wanted messages triggered in more or different locations, while the others felt the locations used were adequate. One participant appreciated when prompts aligned with smoking contexts, saying they often arrived “right when I needed them” (ID5). One participant (ID4) noted missed high-risk areas, such as a friend’s house or a hospital where they often smoked, likely because they visited this location frequently during the intervention phase due to an acute hospitalization of a family member, but the participant had not reported smoking events at that location during the initial assessment phase.

Experience with intervention messages

Every participant reported at least some messages as helpful, especially those offering actionable strategies like breathing exercises or urge delay techniques. As one participant recalled, “There was one about breathing and counting to 10 before I smoke or to let the urge pass […] I thought that was really good advice” (ID2). Four participants also described certain messages as irrelevant or repetitive, such as suggestions to read blog posts, which they were unlikely to follow. Timing was another issue mentioned, with some messages arriving after the urge to smoke had passed, limiting their usefulness.

Experience with smoking cessation

Four participants felt they had reduced their smoking during the study, with one participant (ID1) reducing smoking from a pack a day to half a pack, while another (ID6) achieved complete cessation. On the other hand, two participants reported little or no change. For some, both experiences coexisted: A temporary reduction in certain situations, but no overall shift in daily totals. Three participants described becoming more aware of their smoking patterns, such as realizing they often smoked at home.

Discussion

The current study investigated the feasibility of conducting a pilot MRT among young adults, using smartphone messages to reduce smoking urges at specific high-risk situations and combining EMA and geofence-based intervention delivery. This pilot trial showed encouraging technical feasibility data of a smartphone app-based MRT delivering distraction messages (based on CBT) and acceptance messages (based on Mindfulness/ACT) to reduce smoking urges in young adult smokers. We were able to track individual smoking patterns among participants to establish locations with a high risk of inducing smoking urges. These locations were used to inform geofence-triggered delivery of “in-the-moment” urge reduction intervention messages that could help reduce smoking urges and ultimately prevent smoking events. Our study demonstrated high participant compliance with EMA surveys, with 90.1% completion of geofence-triggered EMAs and 93.9% completion of follow-up EMAs. Within-subject randomization was successful and urge ratings declined from pre- to post-message assessments for all message conditions, with numerically greater reduction for distraction and acceptance messages, compared to control messages. However, the study also demonstrated challenges with the prevention of potential participant deception and research fraud at enrollment that future studies need to be aware of. Moreover, the dose of geofence-triggered intervention message delivery was lower than initially targeted for almost all participants, which will need to be addressed in future studies.

Our study obtained good feasibility regarding EMA compliance, which is an important outcome at this stage of the research. Our study also extends the recent MRT literature in several ways. While previous studies used time-based intervention message delivery (38-40), our study demonstrated the feasibility of location-based interventions (34,35), which adds a novel intervention delivery approach for MRTs. Compared to some previous studies that obtained EMA survey completion rates of approximately 65% (39,40), our study achieved completion rates of 90% and greater. Geofence-triggered EMAs in our study may have improved situational relevance of assessments and intervention messages for participants compared to the time-based prompts in other studies, which might explain the higher compliance rate in the current study. Thus, our study underlines the feasibility of MRT-based methods to address mental health and substance use outcomes. Moreover, this research expands the literature by adding geofence-triggered EMA and intervention delivery tailored to the unique cigarette smoking risk profiles of individual participants.

Moreover, our study showed promise for both distraction (CBT) based as well as acceptance (ACT) based messages and their ability to reduce cigarette smoking urges among young adults. Both CBT (26) and ACT (30) interventions for smoking cessation are supported by evidence and our findings extend their promise to potential “in-the-moment” effects on smoking urges. In the current pilot study, there did not appear to be a difference in message impact across CBT and ACT conditions, which is in line with previous literature that has demonstrated similar efficacy across these smoking cessation interventions (65). Additional research with a larger sample is needed to further investigate the question of comparative efficacy of these different intervention approaches and their momentary impact on smoking urges and behaviors.

Lessons learned and updates to be made before a fully-powered trial

The current study also provided important insights and lessons learned that will help us inform a fully powered MRT. We had to exclude 10 participants after it became evident that their geolocation was outside of the US based on the GPS data collected on the MetricWire app. This underlines the need for robust identity verification and other approaches to mitigate potential participant deception and fraud in online studies, as outlined in previous work (66). During the initial 14-day assessment phase of this pilot trial, only a few real-time cigarettes were reported. This resulted in the creation of few geofence locations for some participants, leading to fewer than the target of three daily geofence-triggered EMAs, thus falling short of our target intervention dose. There are several potential approaches to address this issue: For example, inclusion criteria may need to be tightened for our future trial to ensure that participants smoke more frequently at baseline, allowing for the creation of more geofence locations. In addition to the spatial algorithm used to create geofences in this pilot trial, we could adjust the app to enable participants to self-report smoking locations at baseline, which could be used to supplement geofences. Finally, discrepancies between number of cigarettes reported in real time and in daily diaries could trigger feedback, including reminder messages and implementation intentions, to improve participant reporting.

Limitations and strengths

This study has several limitations. The small sample size and pilot design limit the generalizability of the results and the statistical power to detect significant differences between message conditions. For example, individuals who were willing to enroll and remain in the study through the intervention phase may have been more motivated, technologically comfortable, or responsive than the broader population of young adult smokers, which may be reflected in their high compliance with EMA prompts. Additionally, some participants reported a few smoking events during the initial assessment phase, resulting in a reduced number of geofence locations and fewer intervention opportunities than anticipated. For example, previous work used a different approach and set geofences around locations where a participant reported smoking more than once (35,67), which may result in a greater number of geofences compared to the algorithmic approach chosen in the current study. Moreover, limited cigarette reports during the assessment phase could have affected the accuracy of geofence-triggered EMAs and intervention messages and we may have missed high-risk locations for certain participants. If participants have a significant change in their smoking locations during the intervention phase, those locations would not be set up with a geofence in the current study design. For example, one participant reported in the follow-up interview that they frequently smoked at a hospital due to an acute hospitalization of a family member, but no location assessments were triggered for this location. Some participants also reported technical challenges, such as missed notifications and delays in loading intervention messages on the study app, which may have affected their engagement with the app. We also did not assess readability scores of intervention messages, which may have impacted participant engagement and message impact. While we conducted biochemical verification of remote saliva cotinine to confirm smoking cessation, CPD were self-reported, and those data may be prone to social desirability and self-report bias. Moreover, almost all participants reported use of other nicotine and tobacco products in addition to cigarettes, which may impact saliva cotinine results, and future studies may need to address multiple tobacco product use. Finally, the 45-day period from baseline to follow-up limited our abilities to observe longer-term changes in smoking behavior or sustained effects of the intervention, including seasonal variation in weather, social activities, and mobility patterns that could influence both smoking behavior and the contexts in which urges occur.

The current study also has strengths. The use of an innovative MRT design enabled the assessment of real-time intervention effects, a novel approach in this field. The feasibility of geofence-triggered EMAs and intervention messages was successfully demonstrated, with high compliance rates for geofence-triggered and follow-up EMAs, reflecting strong participant engagement. The integration of EMA provided detailed insights into smoking behaviors and urges in participants’ natural environments, enhancing the ecological validity of the findings. Additionally, the study assessed both proximal and distal outcomes related to smoking behavior, offering a comprehensive evaluation of intervention effects.

Conclusions

In summary, the results of this pilot study indicate potential for the technical feasibility of an app-based MRT using intervention messages triggered by geofence locations using GPS, and generated valuable insights into methodological refinements and technological improvements. Specifically, this study highlighted the importance of robust prevention of potential research fraud at enrollment and needs to increase intervention message dose in future studies. Moreover, this pilot provides support for conducting a larger trial to rigorously evaluate this intervention approach. Findings will inform a fully powered MRT to investigate message efficacy to reduce smoking urges in young adult smokers. A larger MRT will also allow us to investigate the effects of different intervention messages to examine “strategy-situation fit”, i.e., whether intervention messages are differentially effective under specific momentary circumstances.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the CONSORT reporting checklist. Available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-25-17/rc

Trial Protocol: Available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-25-17/tp

Data Sharing Statement: Available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-25-17/dss

Peer Review File: Available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-25-17/prf

Funding: This work was supported by the National Cancer Institute (NCI; R01 CA246590 to J.T.) and the National Institute on Drug Abuse (NIDA; T32 DA007292 to J.J.C.W.).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://mhealth.amegroups.com/article/view/10.21037/mhealth-25-17/coif). F.N. is an unpaid member of the scientific committee for the Smoke Free app, an app unrelated to this study. He also serves as an unpaid Trustee/director and Honorary Secretary of Society for Research on Nicotine and Tobacco Europe (SRNT-E) – registered charity in UK. Neither of the above roles include any financial conflict and both are scientific roles. M.M. conducts research not related to the current manuscript funded by NIH, FDA and Burroughs Wellcome Fund. M.M. has received an honorarium and travel costs for a guest lecture delivered at Ohio State University. M.M. is a senior editor at Health Communication and receives compensation for this role. M.M. has received honoraria for NIH grant reviews. J.T. reports membership on the scientific advisory board of MindCo Health, which offers a smoking cessation program. This arrangement has been reviewed and approved by the Johns Hopkins University in accordance with its conflict-of-interest policies. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. All study procedures were approved by the Institutional Review Board of the Johns Hopkins Bloomberg School of Public Health (No. IRB00013413) and the pilot trial was registered on ClinicalTrials.gov (NCT05991934). All participants who met eligibility criteria provided online informed consent prior to study involvement.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. National Center for Chronic Disease Prevention and Health Promotion (US) Office on Smoking and Health. The Health Consequences of Smoking—50 Years of Progress: A Report of the Surgeon General. Atlanta (GA): Centers for Disease Control and Prevention (US); 2014.
2. Thompson AB, Mowery PD, Tebes JK, et al. Time Trends in Smoking Onset by Sex and Race/Ethnicity Among Adolescents and Young Adults: Findings From the 2006-2013 National Survey on Drug Use and Health. Nicotine Tob Res 2018;20:312-20. [Crossref] [PubMed]
3. Jha P, Ramasundarahettige C, Landsman V, et al. 21st-century hazards of smoking and benefits of cessation in the United States. N Engl J Med 2013;368:341-50. [Crossref] [PubMed]
4. Curry SJ, Sporer AK, Pugach O, et al. Use of tobacco cessation treatments among young adult smokers: 2005 National Health Interview Survey. Am J Public Health 2007;97:1464-9. [Crossref] [PubMed]
5. Watkins SL, Thrul J, Max W, et al. Cold Turkey and Hot Vapes? A National Study of Young Adult Cigarette Cessation Strategies. Nicotine Tob Res 2020;22:638-46. [Crossref] [PubMed]
6. Solberg LI, Asche SE, Boyle R, et al. Smoking and cessation behaviors among young adults of various educational backgrounds. Am J Public Health 2007;97:1421-6. [Crossref] [PubMed]
7. McClure LA, Arheart KL, Lee DJ, et al. Young adult former ever smokers: the role of type of smoker, quit attempts, quit aids, attitudes/beliefs, and demographics. Prev Med 2013;57:690-5. [Crossref] [PubMed]
8. Watkins SL, Thrul J, Max W, et al. Real-World Effectiveness of Smoking Cessation Strategies for Young and Older Adults: Findings From a Nationally Representative Cohort. Nicotine Tob Res 2020;22:1560-8. [Crossref] [PubMed]
9. Pew Research Center. Mobile Fact Sheet. 2024 [cited 2024 Dec 27]. Available online: https://www.pewresearch.org/internet/fact-sheet/mobile/
10. Rajani NB, Weth D, Mastellos N, et al. Adherence of popular smoking cessation mobile applications to evidence-based guidelines. BMC Public Health 2019;19:743. [Crossref] [PubMed]
11. Regmi D, Tobutt C, Shaban S. QUALITY AND USE OF FREE SMOKING CESSATION APPS FOR SMARTPHONES. Int J Technol Assess Health Care 2018;34:476-80.
12. Micalizzi L, Mattingly DT, Hart JL, et al. Smartphone Apps Targeting Youth Tobacco Use Prevention and Cessation: An Assessment of Credibility and Quality. Curr Addict Rep 2023;10:649-63. [Crossref] [PubMed]
13. Hoeppner BB, Hoeppner SS, Seaboyer L, et al. How Smart are Smartphone Apps for Smoking Cessation? A Content Analysis. Nicotine Tob Res 2016;18:1025-31. [Crossref] [PubMed]
14. McKay FH, Wright A, Shill J, et al. Using Health and Well-Being Apps for Behavior Change: A Systematic Search and Rating of Apps. JMIR Mhealth Uhealth 2019;7:e11926. [Crossref] [PubMed]
15. Robinson CD, Seaman EL, Grenen E, et al. A content analysis of smartphone apps for adolescent smoking cessation. Transl Behav Med 2020;10:302-9. [Crossref] [PubMed]
16. Thrul J, Bühler A, Ferguson SG. Situational and mood factors associated with smoking in young adult light and heavy smokers. Drug Alcohol Rev 2014;33:420-7. [Crossref] [PubMed]
17. Nguyen N, McQuoid J, Ramo D, et al. Real-time predictors of smoking among sexual minority and heterosexual young adults: An ecological momentary assessment study. Drug Alcohol Depend 2018;192:51-8. [Crossref] [PubMed]
18. McQuoid J, Regan T, Devkota J, et al. Situations and roles of cannabis versus cigarette use: Integrating ecological momentary assessment with qualitative mapping interviews. Health Place 2024;89:103314. [Crossref] [PubMed]
19. Cronk NJ, Piasecki TM. Contextual and subjective antecedents of smoking in a college student sample. Nicotine Tob Res 2010;12:997-1004. [Crossref] [PubMed]
20. Ferguson SG, Shiffman S. The relevance and treatment of cue-induced cravings in tobacco dependence. J Subst Abuse Treat 2009;36:235-43. [Crossref] [PubMed]
21. Shiffman S, Paty J. Smoking patterns and dependence: contrasting chippers and heavy smokers. J Abnorm Psychol 2006;115:509-23. [Crossref] [PubMed]
22. Dunbar MS, Scharf D, Kirchner T, et al. Do smokers crave cigarettes in some smoking situations more than others? Situational correlates of craving when smoking. Nicotine Tob Res 2010;12:226-34. [Crossref] [PubMed]
23. Shiffman S, Gwaltney CJ, Balabanis MH, et al. Immediate antecedents of cigarette smoking: an analysis from ecological momentary assessment. J Abnorm Psychol 2002;111:531-45. [Crossref] [PubMed]
24. Shiffman S, Dunbar MS, Li X, et al. Craving in intermittent and daily smokers during ad libitum smoking. Nicotine Tob Res 2014;16:1063-9. [Crossref] [PubMed]
25. Schane RE, Glantz SA, Ling PM. Nondaily and social smoking: an increasingly prevalent pattern. Arch Intern Med 2009;169:1742-4. [Crossref] [PubMed]
26. Tobacco Use and Dependence Guideline Panel. Treating Tobacco Use and Dependence: 2008 Update. Rockville, MD: U.S. Department of Health and Human Services; 2008:276.
27. Perkins KA, Conklin CA, Levine MD. Cognitive-behavioral therapy for smoking cessation: A practical guidebook to the most effective treatments. New York, NY, US: Routledge/Taylor & Francis Group; 2008.
28. Ploderer B, Smith W, Pearce J, et al. A mobile app offering distractions and tips to cope with cigarette craving: a qualitative study. JMIR Mhealth Uhealth 2014;2:e23. [Crossref] [PubMed]
29. McCallion EA, Zvolensky MJ. Acceptance and Commitment Therapy (ACT) for smoking cessation: a synthesis. Curr Opin Psychol 2015;2:47-51.
30. Bricker JB, Mull KE, McClure JB, et al. Improving quit rates of web-delivered interventions for smoking cessation: full-scale randomized trial of WebQuit.org versus Smokefree.gov. Addiction 2018;113:914-23. [Crossref] [PubMed]
31. Klasnja P, Hekler EB, Shiffman S, et al. Microrandomized trials: An experimental design for developing just-in-time adaptive interventions. Health Psychol 2015;34S:1220-8. [Crossref] [PubMed]
32. Qian T, Walton AE, Collins LM, et al. The microrandomized trial for developing digital interventions: Experimental design and data analysis considerations. Psychol Methods 2022;27:874-94. [Crossref] [PubMed]
33. Liu X, Deliu N, Chakraborty B. Microrandomized Trials: Developing Just-in-Time Adaptive Interventions for Better Public Health. Am J Public Health 2023;113:60-9. [Crossref] [PubMed]
34. Luken A, Desjardins MR, Moran MB, et al. Using Smartphone Survey and GPS Data to Inform Smoking Cessation Intervention Delivery: Case Study. JMIR Mhealth Uhealth 2023;11:e43990. [Crossref] [PubMed]
35. Naughton F, Hope A, Siegele-Brown C, et al. A smoking cessation smartphone app that delivers real-time 'context aware' behavioural support: the Quit Sense feasibility RCT. Public Health Res (Southampt) 2024;12:1-99. [Crossref] [PubMed]
36. Thrul J, Bühler A, Ferguson SG. An Internet-based ecological momentary assessment study relying on participants' own mobile phones: insights from a study with young adult smokers. Eur Addict Res 2015;21:1-5. [Crossref] [PubMed]
37. McQuoid J, Thrul J, Ling P. A geographically explicit ecological momentary assessment (GEMA) mixed method for understanding substance use. Soc Sci Med 2018;202:89-98. [Crossref] [PubMed]
38. Arévalo Avalos MR, Xu J, Figueroa CA, et al. The effect of cognitive behavioral therapy text messages on mood: A micro-randomized trial. PLOS Digit Health 2024;3:e0000449. [Crossref] [PubMed]
39. Cochran A, Maronge JM, Victory A, et al. Mobile Acceptance and Commitment Therapy in Bipolar Disorder: Microrandomized Trial. JMIR Ment Health 2023;10:e43164. [Crossref] [PubMed]
40. Stanger C, Anderson MAB, Xie H, et al. Momentary mindfulness versus distraction coping messages to reduce cannabis craving among young adults: A microrandomized trial. Psychol Addict Behav 2025;39:200-11. [Crossref] [PubMed]
41. Adams L, Igbinedion G, DeVinney A, et al. Assessing the Real-time Influence of Racism-Related Stress and Suicidality Among Black Men: Protocol for an Ecological Momentary Assessment Study. JMIR Res Protoc 2021;10:e31241. [Crossref] [PubMed]
42. Adams LB, Watts T, DeVinney A, et al. Acceptability and Feasibility of a Smartphone-Based Real-Time Assessment of Suicide Among Black Men: Mixed Methods Pilot Study. JMIR Form Res 2024;8:e48992. [Crossref] [PubMed]
43. Czaplicki L, Barker HE, Thrul J, et al. Quantifying Cigarette and e-Cigarette Marketing Exposure Among Chinese Adolescents Using Ecological Momentary Assessment. Nicotine Tob Res 2024;26:1480-8. [Crossref] [PubMed]
44. Luken A, Rabinowitz JA, Wells JL, et al. Designing and Validating a Novel Method for Assessing Delay Discounting Associated With Health Behaviors: Ecological Momentary Assessment Study. JMIR Form Res 2024;8:e48954. [Crossref] [PubMed]
45. Thrul J, Nordeck CD, Devkota J, et al. 'I crave not to feel uncomfortable' - investigating craving for opioids and cannabis among individuals with chronic pain. Drugs 2024;32:243-53. (Abingdon Engl). [Crossref] [PubMed]
46. Dauber S, Beacham A, Hammond C, et al. Adaptive Text Messaging for Postpartum Risky Drinking: Conceptual Model and Protocol for an Ecological Momentary Assessment Study. JMIR Res Protoc 2022;11:e36849. [Crossref] [PubMed]
47. Dauber S, Beacham A, West A, et al. Ecological momentary assessment of heavy episodic drinking in the early postpartum period: A feasibility study. Drug Alcohol Depend Rep 2023;7:100146. [Crossref] [PubMed]
48. Song AV, Ling PM. Social Smoking Among Young Adults: Investigation of Intentions and Attempts to Quit. Am J Public Health 2011;101:1291-6. [Crossref] [PubMed]
49. Shiffman S, Kirchner TR, Ferguson SG, et al. Patterns of intermittent smoking: An analysis using Ecological Momentary Assessment. Addict Behav 2009;34:514-9. [Crossref] [PubMed]
50. Vilardaga R, Rizo J, Palenski PE, et al. Pilot Randomized Controlled Trial of a Novel Smoking Cessation App Designed for Individuals With Co-Occurring Tobacco Use Disorder and Serious Mental Illness. Nicotine Tob Res 2020;22:1533-42. [Crossref] [PubMed]
51. Bricker JB, Watson NL, Mull KE, et al. Efficacy of Smartphone Applications for Smoking Cessation: A Randomized Clinical Trial. JAMA Intern Med 2020;180:1472-80. [Crossref] [PubMed]
52. Hamoud J, Devkota J, Regan T, et al. Smoking cessation message testing to inform app-based interventions for young adults - an online experiment. BMC Public Health 2025;25:1852. [Crossref] [PubMed]
53. Liao P, Klasnja P, Tewari A, et al. Sample size calculations for micro-randomized trials in mHealth. Stat Med 2016;35:1944-71. [Crossref] [PubMed]
54. Dempsey W, Liao P, Klasnja P, et al. Randomised trials for the Fitbit generation. Signif (Oxf) 2015;12:20-3. [Crossref] [PubMed]
55. Coughlin LN, Campbell M, Wheeler T, et al. A mobile health intervention for emerging adults with regular cannabis use: A micro-randomized pilot trial design protocol. Contemp Clin Trials 2024;145:107667. [Crossref] [PubMed]
56. Regan T, Devkota J, McQuoid J, et al. Real-time antecedents of cannabis use among young adults: An Ecological Momentary Assessment study. Exp Clin Psychopharmacol 2025;33:396-406. [Crossref] [PubMed]
57. Thrul J, Howe CL, Devkota J, et al. A Scoping Review and Meta-analysis of the Use of Remote Biochemical Verification Methods of Smoking Status in Tobacco Research. Nicotine Tob Res 2023;25:1413-23. [Crossref] [PubMed]
58. Thrul J, Meacham MC, Ramo DE. A novel and remote biochemical verification method of smoking abstinence: Predictors of participant compliance. Tob Prev Cessat 2018;4:20. [Crossref] [PubMed]
59. White JS, Toussaert S, Thrul J, et al. Peer Mentoring and Automated Text Messages for Smoking Cessation: A Randomized Pilot Trial. Nicotine Tob Res 2020;22:371-80. [Crossref] [PubMed]
60. Ramo DE, Thrul J, Delucchi KL, et al. A randomized controlled evaluation of the tobacco status project, a Facebook intervention for young adults. Addiction 2018; Epub ahead of print. [Crossref]
61. Fagerström K. Determinants of tobacco use and renaming the FTND to the Fagerstrom Test for Cigarette Dependence. Nicotine Tob Res 2012;14:75-8. [Crossref] [PubMed]
62. Braun V, Clarke V. Using thematic analysis in psychology. Qual Res Psychol 2006;3:77-101.
63. Nielsen J. Nielsen Norman Group. 2000 [cited 2023 Apr 5]. Why You Only Need to Test with 5 Users. Available online: https://www.nngroup.com/articles/why-you-only-need-to-test-with-5-users/
64. Nielsen J, Landauer TK. A mathematical model of the finding of usability problems. Proceedings of the INTERACT '93 and CHI '93 Conference on Human Factors in Computing Systems. New York, NY, USA: Association for Computing Machinery; 1993 [cited 2023 Apr 5]:206-13. doi: 10.1145/169059.169166.
65. McClure JB, Bricker J, Mull K, et al. Comparative Effectiveness of Group-Delivered Acceptance and Commitment Therapy versus Cognitive Behavioral Therapy for Smoking Cessation: A Randomized Controlled Trial. Nicotine Tob Res 2020;22:354-62. [Crossref] [PubMed]
66. Heffner JL, Watson NL, Dahne J, et al. Recognizing and Preventing Participant Deception in Online Nicotine and Tobacco Research Studies: Suggested Tactics and a Call to Action. Nicotine Tob Res 2021;23:1810-2. [Crossref] [PubMed]
67. Hope A, Naughton F, Notley C. Smoking cessation supported by a smartphone app: A qualitative process evaluation of the Quit Sense feasibility RCT. Br J Health Psychol 2025;30:e70010. [Crossref] [PubMed]