This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mHealth and uHealth, is properly cited. The complete bibliographic information, a link to the original publication on https://mhealth.jmir.org/, as well as this copyright and license information must be included.
Smartphone mobile apps are frequently used in standalone or multimodal smoking cessation interventions. However, factors that impede or improve app usage are poorly understood.
This study used the supportive accountability model to investigate factors that influence app usage in the context of a trial designed to reduce maternal smoking in low-income and predominantly minority communities.
We conducted a secondary analysis of data (N=181) from a randomized controlled trial that included a smoking cessation app (QuitPal-m). Supportive accountability was measured by the number of times a participant was advised by their cessation counselor to use QuitPal-m. Participants reported app use helpfulness and barriers. Investigators tracked reported phone and technical problems that impeded app use.
Most participants rated the app as very helpful (103/155, 66.5%), but daily use declined rapidly over time. App use was positively related to the level of perceived app helpfulness (
Findings show the utility of supportive accountability for increasing smoking cessation app use in a predominantly low-income, minority population, particularly if quit motivation is low. Results also highlight the importance of addressing personal and phone/technical barriers in addition to adding supportive accountability.
ClinicalTrials.gov NCT02602288; https://clinicaltrials.gov/ct2/show/NCT02602288
Mobile apps can offer convenient, low-cost, and on-demand support and intervention for smoking cessation. According to estimates, approximately 63%-76% of smokers own a smartphone [
The investigation was guided by the supportive accountability model, which maintains that adherence to eHealth interventions, including mobile health apps, can be increased through accountability to a supportive, trustworthy person with relevant expertise, such as a health coach or medical provider [
Drawing upon self-determination theory [
This observational study investigates the relations between supportive accountability, motivation, and smoking cessation app use in the context of a clinical trial aimed to promote smoking cessation among low-income maternal smokers. The trial, Babies Living Safe and Smokefree (BLiSS) [
We tested two hypotheses based on the supportive accountability model:
A higher percentage of smokers will use a cessation app on their phone in the 24 hours after receiving prompts about app usage from a cessation counselor (ie, supportive accountability) than in the 24 hours preceding such prompts.
The relation between prompting and app usage will be stronger among participants not planning to quit in the next three months than among participants planning to quit in the next three months (ie, motivation as a moderator).
We also explored correlates and potential barriers to app usage. As the target population is low income, we anticipated some potential technical and phone-related barriers (eg, service disruptions due to late payments, phone sharing) as well as practical barriers (eg, no time, lack of interest). Finally, we explored whether app usage correlated with amount of smoking at end of treatment.
This investigation used secondary data collected as part of the BLiSS trial [
Trial eligibility criteria included the following: received WIC clinic AAR intervention; English-speaking; at least 18 years old; report smoking; own a smartphone; and report their child aged <6 years is exposed to tobacco smoke. Exclusion criteria included the following: currently pregnant; presenting issues that could interfere with their ability to provide informed consent or follow study procedures, such as psychosis, inadequate health literacy, or non-nicotine drug dependence. All BLiSS participants randomized to the AAR + MBI treatment arm (N=199) were potentially eligible for inclusion in this observation study. However, 18 participants were excluded from all analyses, leaving a sample of 181. Reasons for exclusion included technical problems downloading the mobile app (n=6), issues with the back-end software that tracked participants' app usage (n=6), and participant withdrawal from the trial before receiving intervention or app advice (n=6). Comparisons of excluded and nonexcluded participants revealed no statistically significant differences in age, race, marital status, employment status, education level, or phone operating system.
After WIC staff referred mothers to the trial, trained research assistants screened for eligibility, administered informed consent, and collected baseline self-report data using computer-assisted telephone interviews. Participants were then randomized to either the AAR + MBI or AAR + Control condition. The AAR + MBI intervention included messaging about child TSE harms as well as support and guidance with skills training and problem-solving delivered via multiple channels: the project quitline, providing up to 5 telephone counseling sessions over 3 months; cessation mobile app; print materials for the participant and their family; intersession text follow-up, reminders, and support, as well as educational video clips that reinforced telephone session and written materials content; and 8 weeks of nicotine replacement therapy and instructional support.
After randomization, participants in the AAR + MBI condition had an orientation home visit that provided a review of intervention objectives, a binder of intervention print materials, and an illustrated guide to using the mobile app. Research staff also assisted participants in downloading the mobile app and showed them a brief video tutorial on how to use the app.
Telephone counselors delivering the skills training and support intervention received intensive training in tobacco treatment enhanced with support, advice, and problem solving around protecting children from tobacco smoke exposure and creating a smoke-free home and car. Importantly, they were trained in how to use a telephone counseling process that would promote participants' mobile app usage to complement and extend treatment beyond the phone sessions. For example, telephone counseling included guidance on goal setting, building social support, and improving skills (eg, self-monitoring) to reduce child TSE, identify smoking triggers, and manage urges to smoke. The mobile app has tools that support all these processes.
The BLiSS mobile app was a modified version of the National Cancer Institute's QuitPal app [
An innovative feature of QuitPal-m is a web-linked portal that connects telephone counselors to a dashboard that displays participant app usage (
During the home visit orientation to the app, participants were told that the smoking self-monitoring, or tracking, features were the most important and should be used daily. Home visitors demonstrated how to enter data into the app. Participants also were informed about the counselor dashboard and how telephone counselors would routinely monitor app entries to learn more about participants' smoking habits and guide their advice about the participants' behavior change efforts. Finally, they were told that the counselor would remind them or initiate troubleshooting when app use adherence was low. Thus, participants were aware of the expectations, monitoring, and accountability related to app usage from the beginning of the intervention.
QuitPal-m app features and illustration of web-enabled counselor dashboard.
Adherence is defined as the active use of the app as recommended [
Another important outcome to test the supportive accountability model was the proportion of days that participants used the app 24 hours before and 24 hours after receiving advice from their counselor to do so. This usage was calculated by dividing the total number of days the app was used 24 hours after (or before) advice was offered divided by the total number of days advice was offered. For example, if a person received advice on three days and used the app within 24 hours on each of those three days, they would score 100 (3/3). If they only used the app within 24 hours on two of the days, they would score 66 (2/3). Thus, scores could range from 0 to 100.
Participants rated the app helpfulness on a 4-point scale (1=not at all, 2=a little helpful, 3=somewhat helpful, 4=very helpful) postintervention. They also reported (no/yes) whether they experienced any of the following barriers to app usage during the intervention: forgetting, lack of interest, lack of time, and confusion/difficulty using the app. In addition, we tracked phone and technical problems reported throughout the study that interfered with app use (eg, phone not in service, app freezing).
To assess participants' motivation/determination to quit smoking, we included a baseline question about whether they planned to quit smoking in the next 3 months (no/yes). We also included a self-report measure of average cigarettes smoked per day in the past week at the 3-month end-of-treatment period.
The sample (N=181) was comprised of mothers who were mostly single (113/181, 62.4%), Black (123/181, 68.0%), and unemployed (105/181, 58.0%). The highest level of education completed for the majority was high school or less (111/181, 61.3%). The most common phone operating system was Android (144/181, 79.6%), followed by iOS (37/181, 20.4%). At baseline, over three-fourths (138/181, 76.2%) of the participants reported that they were planning to quit smoking in the next three months.
On average, participants received advice to use the app 3 times (median 3; mean 2.98, SD 1.58) over the course of the intervention. A total of 10 of the 181 participants (5.5%) received no advice: 9 because they could not be reached for phone intervention sessions and 1 because the interventionist did not have an opportunity to bring it up during the single phone session the participant completed. Patterns of app usage are shown in
Patterns of app usage (N=181).
Variable | Mean (SD) | Range |
Days app used (out of 91) | 16.48 (17.52) | 0-87 |
Times tracking feature used | 63.64 (81.68) | 0-443 |
Times savings feature used | 1.72 (3.09) | 0-26 |
Times graph feature used | 1.36 (2.43) | 0-16 |
Times tips feature used | 0.92 (1.85) | 0-16 |
Times summary feature used | 1.23 (2.46) | 0-15 |
Times friend alert feature used | 0.87 (1.69) | 0-14 |
Times my health feature used | 1.48 (2.17) | 0-12 |
Times quitline phone number used | 0.39 (0.87) | 0-14 |
Percent of participants using QuitPal-m app by intervention week (N=181).
Examining the relation of days of app usage to sociodemographic factors revealed a single significant association, with education: participants used the app on significantly more days if they had more than a high school education (mean 21.49, SD 21.37) than if their highest education was high school or less (mean 13.32, SD 13.77;
Of the 155 participants who completed the postintervention survey, most rated the app as very helpful (103/155, 66.5%) or somewhat helpful (26/155, 16.8%), a minority (23/155, 14.8%) rated it as not at all/a little helpful, and a few did not answer (3/155, 2%). Participants who rated the app as very helpful tended to open the app on significantly more days (mean 20.56, SD 19.44) than their counterparts who rated it less helpful (mean 13.56, SD 13.26;
A sizeable proportion (68/181, 37.6%) of the participants reported phone and other technical problems that occasionally interfered with using the app at some time during the intervention. This is in addition to the 12 who were excluded from the study due to app download and back-end issues from the start. The most frequently reported problem was intermittent service disruptions due to running out of minutes or being unable to pay bills on time (33/181, 18.2%), followed by deleting the app due to software problems (eg, app freezing) or insufficient memory for the app (23/181, 12.7%), followed by getting a new phone and having difficulty downloading the app again (12/181, 6.6%). Certain problems, such as phone service disruption, affect some app features and functions, such as communication between the app and the counselor dashboard, but not other features, such as tracking, savings, and tips. Participants who experienced phone/technical problems used the app on significantly fewer days (mean 11.40, SD 13.04) than their counterparts who did not experience these problems (mean 21.49, SD 19.86;
Finally, we analyzed patterns of usage following advice from the counselor and whether motivation/determination to quit moderated the effects of counselor advice on app usage. These analyses excluded the 10 participants who received no advice. Paired
As shown in
Regression models predicting proportion of days app was used within 24 hours after receiving app advice from a counselor (N=171).
Predictor | |||
Constant | .56 (.05) | 12.10 | <.001 |
Highest education levela | .06 (.05) | 1.12 | .27 |
App ever unavailable due to phone/technical problemsb | –.16 (.06) | –2.82 | .005 |
Reported any barriers to app useb | –.16 (.05) | –3.00 | .003 |
Total counselor advice to use app | .05 (.02) | 3.28 | .001 |
Plan to quit in next 3 monthsb | .05 (.06) | .79 | .43 |
Total advice × plan to quit | –.08 (.04) | –2.25 | .03 |
a0=high school or less, 1=more than high school.
b0=no, 1=yes.
Relation between counselor app advice and proportion of days app was used within 24 hours after advice as a function of motivation to quit smoking (N=171).
Overall adherence to daily app use goals was low and declined steadily over the course of the intervention. Consistent with the supportive accountability model [
The current findings suggest that supportive accountability is a promising method to improve adherence to mobile app use, especially among users with low levels of motivation to change. The interaction between accountability and motivation suggests that increased motivation is a primary mechanism of supportive accountability in app adherence. Intervention models, such as BLiSS, that incorporate mobile apps and interaction with counselors can build accountability directly into the counseling protocol. Folding in supportive accountability processes and messaging into treatment-goal setting and skills-training elements of behavioral interventions can maximize the chance that participants benefit as much as possible from app engagement. For example, getting participants to engage in self-monitoring/tracking as “homework” and taking stock of their smoking patterns, including triggers and consequences of smoking that sustain the behavior, may boost intervention efficacy. Emphasizing accountability messaging with nonadherent users could have the twin virtues of supporting users with the greatest need and of not alienating users who are already highly engaged.
A final noteworthy finding was the relatively high participant ratings of the helpfulness of the app. Further, the more satisfied participants were with the app, the more they tended to use it. Thus, increasing user satisfaction could help to promote more engagement. These findings appear to conflict with the overall low engagement and rapid drop-off in app use during the intervention. It is possible that participants were able to extract the primary value from the app in less time than we originally predicted. In retrospect, we realize that it would not take more than a week or two of tracking for users and counselors to recognize smoking behavior patterns and, on that basis, make smoking avoidance/cessation strategies. In addition, after several days of using the tracking feature, the app generates automated notifications about health improvements and personalized tips for avoiding smoking without the user having to launch the app daily and manually input data. Apps that use other strategies to address smoking, such as stress management, may be helpful over longer periods of time when relapse prevention is important. However, for initial quit attempts, when it is important to identify and address factors that trigger and sustain smoking, an app with a more circumscribed time of use may be sufficient.
An interesting question for future research is whether the medium and source of supportive accountability influences adherence levels. For example, is an SMS text message from a counselor as effective as direct contact and collaborative problem solving that can occur during a live counseling session? Is a supportive accountability approach as effective at improving app adherence when it is automated or delivered by an embodied conversational agent (eg, a computer-generated avatar of a counselor) versus a live human? Unlike common reminder software, embodied agents can be verbally expressive and mimic human gestures, which could elicit social responses from people that parallel human-human social interaction. If effective, such an approach would reduce some of the burden on the interventionist.
A related question is how much encouragement or nudging from a coach is beneficial for promoting app usage? The supportive accountability model suggests that once an individual has internal motivation to change a behavior, ongoing accounting may backfire, or at least show diminished returns. In this study, the amount of direct advice from the health counselor was deliberately modest and it was front-loaded to the early weeks of the intervention. However, as we did not experimentally manipulate frequency and intensity of supportive accountability-driven app advice and feedback, it raises the question of whether more advice might have resulted in greater adherence.
In thinking broadly about the challenges of app engagement, it is important to consider the social and economic contexts that influence user engagement. In this study, participants were drawn from predominantly minority and low-income communities. As reflected in some of the observed barriers (eg, phone service disruptions, phone sharing, time constraints), participants' life circumstances would have undermined adherence even among those motivated and otherwise engaged in the overarching multimodal intervention. Linking the QuitPal-m app to wearable smoking sensors [
This study has some noteworthy limitations. The primary one is the lack of experimental data. Instead of manipulating levels of supportive accountability, we measured it and observed how it related to app use behaviors. This correlational design is subject to internal validity threats, including the possibility that something other than accountability increased participants' app use after receiving counselor advice. For example, maybe the call itself served as a reminder and did not require the counselor to specifically discuss the app. Future research will need to add randomization and control conditions to rule out alternative explanations of findings. Another limitation is that participants used their own cell phones and service plans. This introduced extraneous factors that interfered with app use independent of users' intentions/desires to use the app. A controlled study would be able to isolate the effects of supportive accountability by providing a device and service, ensuring equitable access for all participants.
The findings show the potential utility of supportive accountability for increasing use of a smoking cessation app in a low-income, predominantly minority population. Consistent with the model advanced by Mohr and colleagues [
Ask, Advise, Refer
Babies Living Safe and Smokefree
multimodal behavioral intervention
modified QuitPal app
tobacco smoke exposure
Special Supplemental Nutrition Program for Women, Infants, and Children
The authors thank the research assistants and staff and administrators from North Inc who assisted with the study. This project was supported by a grant from the National Institutes of Health/National Cancer Institute (R01CA188813).
None declared.