Introduction
Mobile technologies have recently emerged as highly accessible support tools for health behavior change and mental health promotion. This has been particularly true for smoking cessation, with notable increases in the use of [1,2] and referral to [3] mobile technologies designed to support quitting. User engagement with smartphone apps, however, represents a critical challenge. It has been estimated that less than 5% of apps continue to be used 15 to 30 days after the initial app download [4]. Note that to our knowledge, a standardized operational definition of app engagement has not yet been established [5]; throughout this paper, we use the term “app engagement” to denote user behavior, that is, an app user interacting with the app’s user interface.
App engagement is important for several reasons. First, it is through interaction with an app that app users engage in therapeutic activities. Such engagement can take different forms. An app may present information, provide a tool or behavior change strategy, assign homework, or prompt the app user to take specific actions (eg, call a clinician) [1]. It can be assumed that the more app users interact with the app, the more they will engage in thoughts, feelings, and actions that are believed to be beneficial to their smoking cessation goal. This logic is consistent with findings across the eHealth literature demonstrating that greater engagement with eHealth tools (eg, websites) is associated with more favorable outcomes related to smoking cessation [6-8].
A critically useful feature of smartphone apps is their potential utility in providing sustained support over time. Several studies suggest that extending behavioral support helps smokers remain abstinent in the long term, with longer treatments lasting 8 to 12 weeks [9-11]. If longer treatment is better, the next question is what app users should be doing within a smoking cessation app. The vast majority of publicly available smoking cessation apps focus on simple tools: calculators to track money saved and health benefits accrued or calendars to track the days until or since the chosen quit day [1]. The apps provided on Smokefree.gov, a recommended mHealth referral site for treating smokers in health care settings [3], are more sophisticated. They offer a variety of tools and trackers (eg, time-based and GPS-based reminders to abstain from smoking at high-risk times or in high-risk locations and daily tips and milestone achievement badges) while also providing tailored feedback for overcoming urges to smoke due to cravings and mood states. These apps are able to capture the initial engagement of a very large number of smokers (there are over 25,000 and 13,000 users to date for QuitGuide and quitSTART, respectively) [12]. It is unclear, however, to what extent and how these apps sustain engagement over time.
A new generation of smoking cessation apps is emerging. In such apps, there are additional ways that smoking cessation is supported. The nature of these approaches varies widely from app to app, including contingency management [13,14], motion-sensor detection of smoking [15], carbon monoxide level monitoring [16], using gaming to engage smokers in skills practice [17] or other activities promoting smoking cessation [18], promoting nicotine replacement therapy adherence [19], prescribing bouts of physical activity [20], mindfulness training [21], and acceptance and commitment therapy (ACT) [22,23]. Most of these apps are in the early stages of development, and study protocols have been published, but the studies are ongoing. A handful of pilot feasibility studies exist, only a few of which offer insight into app usage over time. App usage ranged from 7 days for a smartphone app using a gamification approach to smoking cessation [18], to 34 and 32 days, respectively, in a pilot randomized control trial that compared an ACT-based app to the National Cancer Institute (NCI)’s app QuitGuide [22]. The largest randomized trial to date reported 24 days of app use for the app iCanQuit, an ACT-based smoking cessation app, compared to 7 days of use for an earlier version of NCI’s QuitGuide, a version that was largely text-based and did not include trackers or tools [23].
In this paper, we present a feature-level analysis of app usage over time of the Smiling Instead of Smoking (SiS) app. To our knowledge, this is only the second research project that links feature-level app use to subsequent smoking cessation. The first such project provided a feature-level analysis of the app SmartQuit [24], an early version of iCanQuit [23]. The feature-level analysis of SmartQuit [24] indicated that the use of 3 features was prospectively linked to smoking cessation at follow up; 2 of these features were ACT focused (ie, tracking ACT skills practice and tracking the practice of letting urges pass); the remaining feature was a traditional US Clinical Practice Guidelines (USCPG) feature (ie, viewing the quit plan) [24]. In that analysis, the team noted little overlap between the popularity of the app’s features and their subsequent link to smoking cessation success.
The app we are examining in this paper, the SiS app, was developed specifically for nondaily smokers [25-27]. Nondaily smoking is a widespread, increasingly prevalent pattern of smoking. Currently, 24.3% of all adult smokers smoke on a nondaily basis [28], which constitutes a 27% increase over the past decade [29]. Despite such prevalence, particularly in ethnic minority groups [30-34] and vulnerable populations, such as persons with mental health and substance use challenges [35], behavioral and pharmacological recommendations for nondaily smoking remain unaddressed in clinical practice guidelines [36]. Nicotine replacement therapies have been tried but so far have failed to show efficacy in achieving smoking abstinence in nondaily smokers [37,38], in line with a lack of interest among nondaily smokers in pharmacotherapy for smoking cessation [39,40]. Nondaily smokers are, however, highly motivated to quit smoking. Compared to daily smokers, they have greater current intentions to quit smoking [31,41,42] and more recent and planned cessation efforts [42-45]. These factors point to the utility of behavioral support, which can be delivered effectively via smartphone technology.
The therapeutic goal of the SiS app is to maintain positive affect while smokers undergo a quit attempt. Positive affect often decreases during a quit attempt [46], as smokers struggle with cravings and adjusting to a smoke-free life. Indeed, recently, a decrease in positive affect has been suggested as a new symptom of tobacco withdrawal, based on data from 24 trials involving 2054 participants showing a medium effect size (Cohen d=–0.40) for an overall decrease in positive affect [47]. Maintaining positive affect during this time, however, may be especially beneficial, because greater positive affect is associated with increased self-efficacy to quit smoking [48], decreased desire to smoke [49,50], and greater readiness to process self-relevant health information [51], all of which are constructs highlighted in dominant health behavior theories as causal agents in successful behavioral change [52-56].
Smoking cessation apps can deliver a wide array of content, but users must engage with the information to benefit. The SiS app engages users in short, daily exercises to boost their positive affect. This boost in positive affect is intended to increase their readiness to engage in the smoking cessation materials provided in the app [51]. Moreover, completing happiness-boosting exercises is intrinsically rewarding, in and of itself. Having rewarding experiences while using the app may, in turn, entice app users to return to the app on subsequent days. This dynamic is evident across a diversity of settings, and positive psychology interventions have been found to be highly appealing to patients [57], resulting in better treatment adherence [58,59] and engagement [60].
In the initial study of version 1 of the SiS app (N=30), the “happiness exercises” appeared to be a driving factor in app engagement [26]. In this paper, we are examining app usage of participants in a larger trial, using version 2 of this app. The overall app usage was quite high, with smokers using the SiS app for an average of 24 days within the prescribed period of 49 days [27]. Using these data, our goals were to (1) describe how nondaily smokers used the app and (2) test if app usage patterns (ie, any app use, use of happiness content, and use of smoking content) during the prescribed treatment period predicted smoking abstinence (ie, self-reported 30-day point-prevalence abstinence [PPA]) at the end of treatment and at a 6-month postquit follow up.
Discussion
Key Findings
This secondary data analysis of participants enrolled in a smartphone app–based smoking cessation study provided insight into how feature-level app usage behaviors relate to smoking cessation outcomes. The app under study, the SiS 2 app, is part of an emerging generation of smartphone apps that offer a treatment framework beyond standard USCPG content. In our feature-level analysis of the SiS 2 app, we found that overall greater app usage predicted higher chances of subsequent smoking cessation. This finding is in line with the feature-level analysis of the app SmartQuit [24], which showed that greater app use was positively related to subsequent smoking cessation. It suggests that there was a potential therapeutic effect of engaging with the SiS 2 app, though a causal relationship could not be established in this observational study.
Divergent from SmartQuit findings, for SiS 2 the popularity of the app’s features aligned with smoking cessation success, where greater usage of the happiness components of the app predicted greater chances of 30-day PPA at both the end of treatment and at the 6-month follow up. This finding suggests that the positive psychology components of SiS 2 are an important factor in supporting smoking abstinence. This is in line with findings from in-person treatment studies that indicate the value of positive psychology in smoking cessation [66,67]. Our findings here suggest this value may extend to the smartphone app environment. Particularly noteworthy is the high level of engagement with the SiS 2 app (ie, sustained use over 49 days, with multiple uses per week), largely driven by the positive psychology content. In mHealth research, touch point frequency is an emerging area of investigation, with some apps focusing solely on reminding app users not to smoke at key timepoints [15], as simple reminders can be powerful tools in smoking cessation [68]. The high touch point frequency observed for the SiS 2 app speaks to its ability to remain present in smokers’ minds as they navigate smoking cessation.
Prescriptive Clarity
Our results highlighted a rather stark difference in the completion rates of assigned versus ad libitum tasks. Happiness exercises were completed on 45% of days, and 57% of assigned behavioral challenges were completed. These completion rates are in line with the completion rates reported for a mindfulness smoking cessation app for adolescents; participants completed 13 of 22 (61%) of the assigned mindfulness modules [69]. The ad libitum tools, on the other hand, were sparsely used: they were used on only 7 days for the most popular ad libitum tool. This finding suggests that prescriptive clarity may be of critical importance in driving app usage, and therefore in achieving an app’s therapeutic effect. In the SiS 2 app, there was prescriptive clarity: clear expectations were set about treatment length (ie, 49 days), and which specific actions to complete (ie, daily happiness exercises and 15 behavioral challenges). These expectations were reinforced with proactive push notifications. The iCanQuit app [70] also had prescriptive clarity: a set number of modules were required to be completed. Both apps had high app engagement over time. By contrast, the NCI app QuitGuide lacks prescriptive clarity. Many potentially useful tools are offered by QuitGuide, but it is not clear which tools to use, when, and for how long. Future research that experimentally tests whether prescribed content is more engaging would be useful to inform the development of health behavior apps.
To date, text messaging has shown greater smoking cessation benefits than smartphone apps [71], potentially in part due to prescriptive clarity. Text-messaging interventions have prescriptive clarity (ie, there is a set number of days in the program; information is provided on specific days, in proactive, succinct fashion; and actions to be taken are clearly spelled out), while many apps do not [1]. Our data show that participants are willing to complete assigned tasks much more than use ad libitum tools. Data from a randomized trial conducted in the United Kingdom show that assigning daily tasks within a smoking cessation app versus offering the same content without the specific daily tasks led to improved smoking cessation rates [72]. Combined, these findings lead us to believe that prescriptive clarity is a critically important feature in the development of smartphone apps targeting smoking cessation.
Long-term Engagement
Smartphone app technology has the potential to provide ongoing support for smoking cessation over long periods of time. To date, this potential has been largely unexplored, including in our own work. To our knowledge, few studies have examined app engagement; these studies have focused on factors contributing to initial app use [73] or have tested the value of push notifications in enhancing engagement [74]. These studies have not provided insight into the content features and app parameters that promote long-term engagement. In-person smoking cessation interventions typically provide 8 to 12 weeks of support. The SiS 2 app provided assigned tasks for 49 days and continued use of the app’s ad libitum tools as needed, a treatment length in line with the support offered via the NCI’s Smokefree text-messaging interventions, and similar to the UK Smoke Free app, which provides assigned daily tasks for 31 days [72]. This treatment length roughly covers the time from preparation to action according to the transtheoretical model of change and does not address maintenance [55]. In fact, originally, the SiS app spanned only 21 days [26], a length that was specifically chosen to provide support during the acute “cessation” phase [75] of the process of smoking cessation. Based on user feedback, we increased treatment length to 49 days in version 2. Our app usage data demonstrate that this increased treatment length was well tolerated, opening the door to potentially further increasing treatment length to provide support during the maintenance phase of smoking cessation. To date, however, very little research exists to guide the intervention content of smartphone apps to support sustained user engagement and long-term abstinence from smoking.
Limitations
This secondary data analysis was based on a single-arm trial, and therefore causal inferences about the observed effects cannot be drawn. Our analyses were exploratory. They hint at the value of positive psychology to engage app users in smoking cessation over time and the value of prescriptive clarity, but we did not design this study, nor the SiS 2 app, to address these questions. In considering these effects, it should be kept in mind that the participants were asked to complete happiness content on a more frequent basis than smoking cessation content (ie, daily vs every 3 to 4 days) and that the smoking cessation tools had a shelf-life (eg, “smoke alarms” became less useful as cravings diminished; participants could log cigarettes, but not smoke-free days). In terms of generalizability, it should be noted that the SiS 2 study used an interactive onboarding procedure via phone. While this is in line with warm handoff models for smoking cessation [76,77], app usage patterns within the context of a clinical trial are typically higher than real-life app use [78].
Conclusions
In the SiS 2 app, greater app usage predicted greater chances of self-reporting 30-day PPA at both the end of treatment and at a 6-month follow up. This finding strengthens the rationale for testing this app in a randomized trial. Feature-level analysis of app usage patterns suggests that positive psychology content and prescriptive clarity may promote app engagement.