Since 2013, mobile health (mHealth) or the use of mobile computing and communication technologies in health care, public health, and personal wellness has gained significant interest, including the use of SMS through mobile devices. This is because of efforts to engage patients [1] and improve health outcomes at lower costs [2]. mHealth is considered to be a subset of the broader eHealth movement, which involves the digitization of health care processes [3]. Although eHealth and mHealth are potentially transformative for health care users, population health, and health care systems, the benefit of the significant economic investment in these modalities is undergoing scrutiny. For example, a 2011 review of the billions of dollars spent on eHealth systems globally has shown no evidence that implementing these technologies reduces health service costs [4].

When investigating clinical impact, early randomized control trials focused on mHealth feasibility. Current research has shown mHealth to be effective for improving health outcomes, especially when using brief messages such as those delivered by SMS through personal devices (eg, smartphones). Brief messages are cost-effective and can be delivered en masse in low-resource areas and can be personalized to target specific populations [5]. For example, women in low- and middle-income countries who received SMS support during pregnancy showed reduced chances of infant morbidity or mortality [6]. This is congruent with other SMS mHealth initiatives that have shown an increase in smoking cessation [7], improvements in the uptake of sexual health services using SMS reminders compared with controls [8], and reduced viral loads of patients with HIV when reminded to take antiretroviral medication [9].

For mHealth interventions that include a communication component, message content is critical to the effectiveness of the interventions. For example, in a 24-month randomized clinical trial for weight loss, targeted feedback from health care providers by SMS was shown to improve engagement behavior in self-reported energy and fat intake [10]. An important aspect of this study was that messages were personalized using the individual feedback provided by participants. Communication between health care providers and patients holds an immediate opportunity for improved clinical outcomes and patient engagement, emphasizing the importance of identifying suitable message content for this communication [5]. However, one of the key challenges faced by these health care services is how to understand patterns of engagement and keep users engaged [11].

mHealth engagement is complex because of numerous definitions and the technology straddling the disciplines of health care, information systems, and business studies. In this research, engagement is defined as the uptake or use of an mHealth device [12]; however, limiting the exploration of mHealth to one discipline skews our understanding of the phenomena, resulting in a partial view of the impact of the technology and engagement.

Through information systems research, the usability and design of technology play a key role in the uptake of mHealth [13]. It is recommended that the usability and visual appeal of apps should be taken into consideration from the beginning of the design process [13]; however, there is also an opportunity for it to be critiqued based on user feedback throughout the life of the mHealth intervention to support sustained engagement.

The business literature suggests that a user-centered experience is a key factor that contributes to engagement in service contexts, and this can also be applied to mHealth technology [14]. It is evident that there are different views on user experience. This paper defines the user experience as the total experience that a person obtains from all interactions with a health care provider and an mHealth device. This includes the value they receive [15-17]. The total user experience can be evidenced by end-user thoughts, feelings, and behaviors related to their technology use [18,19]. Value is critical to engagement and can be defined as a total assessment of the perceived cost [20] versus benefits [21,22]. If the effort to engage is greater than the total perceived value of the engagement, users may not be interested in an mHealth device.

An interdisciplinary framework that incorporates health, information systems, and business does not exist in the present literature for mHealth engagement. However, O'Brien and Toms [12] developed a linear conceptual framework for defining technology user engagement that was applied in the context of web-based technology apps and incorporates attributes from both business and information systems. Through empirical inquiry, these authors demonstrated 4 stages of engagement, including point of engagement, period of engagement, disengagement, and re-engagement, that can be experienced in a linear sequence and repeated. Figure 1 provides the model of engagement stages, as proposed by O’Brien and Toms [12].

As illustrated earlier, the point of engagement is when the user encounters the technology for the first time, the period of engagement is the periodic use of the technology, disengagement is characterized by a period without using the technology, and re-engagement is a reintroduction of the technology to a previous user. Each stage in the process is characterized by themes and attributes. Attributes are the phenomena experienced by end users when engaging with mHealth technology, and these can be analyzed into themes. Attributes also signify the total user experience and are used as a baseline for exploration in this research. Importantly, this conceptualization provides a standard framework that can be flexibly applied in different contexts to explore the determinants of engagement in an mHealth context [23].

The framework of O'Brien and Toms has been explored in other digital health investigations, often underpinning studies that seek to conceptualize context-specific engagement [24]. Others simply apply the existing stages of the engagement framework [25,26]. Only one known study has explored additional stages by examining the notion of unengagement [11]. In this 2019 study of 486 health services’ users who accessed web-based health services [11], it was recognized that the nonengagement or disengagement were examples of unengagement. Thus, to date, the stages and attributes in the model by O’Brien and Toms require critique and further investigation, specifically within the context of mHealth, where exploration is limited.

With that in mind and using the framework by O’Brien and Toms as a departure point, this study aimed to develop a person-centered approach to generate relevant brief message content relating to engagement stages to promote sustained user engagement. When combined with users’ thoughts and observations of actual use, the framework by O’Brien and Toms can provide a way to investigate the stages of engagement for mHealth devices [12]. This is commensurate with the business literature and other health technology interventions that “demand a user-centered and iterative approach to development, using mixed methods and in-depth qualitative research to progressively refine the intervention to meet user requirements” [27]. Hence, this study asks, “How can knowledge about users’ mHealth device engagement stages inform the development of brief message content intended to promote sustained user engagement?”

To gain rich insight, a qualitative approach was used to investigate the framework by O’Brien and Toms [28]. Qualitative studies allowed the researcher to observe or record behavior in its natural setting [28]. Interviews were used to explore participants’ state of mind in-depth [29], and face-to-face observations provided an understanding of participants’ behaviors [30].

In total, 2 sample groups were used in this research to deepen the understanding of mHealth technology use, avoiding the use of a single case to support findings. Participants were considered eligible if they had used an mHealth technology for at least 2 weeks before the interviews and observations (excluding pregnancy apps), were aged >18 years, were able to communicate effectively in English, and provided informed consent. Interviews and observations across the 2 sample groups were conducted until data saturation occurred [31]. This occurred in a total of 34 participants.

Samples 1 and 2 were contextually different, comparing mHealth engagement with solo (single person) technology use and mHealth engagement in a social (group) context. In both samples, a screening questionnaire, interviews, and structured face-to-face observations were undertaken. Questions included perceptions of use, value, behaviors, and engagement. Participants were asked about their program usage (either an individual program such as a tracking app for sample 1 or the group program for sample 2). Example questions included How has the program impacted your behavior toward wellness?, Can you please show me what you value about the program and explain why you have selected this?, and Why do you think your usage has dropped and/or increased? The full interview protocol is outlined in Multimedia Appendix 1. Ethics approval was obtained from the researcher’s home university (University Human Research Ethics Committee), and the research was conducted in line with standard ethical guidelines and the National Statement on Ethical Conduct in Human Research [32]. No incentives were offered in this study.

Participants who used mHealth technologies (eg, mHealth apps or wellness websites) in an individual context for at least 2 weeks before the first interview were purposively recruited using a promotional flyer and snowball sampling in a large metropolitan city in Australia. Upon recruitment, participants attended in-depth confidential interviews and structured face-to-face observations using their mHealth technology in a private room or in a public location of their choice. Interviews and observations of the participant using the device were recorded by a researcher on a video camera. Participants were then invited to attend a second session 6 to 8 weeks after the first session and a final interview and observations at 8 to 10 months after the first session.

Participants who used mHealth technologies in a group context were purposively recruited from the Global Corporate Challenge (GCC), a 100-day workplace team pedometer competition. The program is designed to promote workplace health and engagement around the world by encouraging employees to work in teams completing a minimum of 10,000 steps per day. After 2 to 4 weeks of engagement, participants attended in-depth confidential interviews and structured face-to-face observations using their mHealth technology in a private room or in a public location of their choice. These were recorded using a video camera by a researcher. All participants were invited to attend a second session 8-10 weeks after the first session and a final interview and observation 3-5 weeks after the GCC was complete.

Interviews and structured participant observations from samples 1 and 2 produced verbal and visual data that were analyzed by the principal investigator. Interviews were the primary source of the research findings, whereas observations were used to confirm the data collected from the interviews and to show how participants engaged with their mHealth technology.

Transcribed interviews were read several times, and deductive-inductive thematic analysis was undertaken using Atlas.ti software [33,34]. The stages of engagement and attributes from the framework by O’Brien and Toms [12] were applied to the interviews and observations. Deductive analysis was used to evaluate the conceptual framework, and a general inductive analysis was used to discover new attributes and important themes based on attributes related to each stage of engagement [35]. When a priori attributes and stages of engagement were not present in the data, they were removed; when new ones emerged, they were added to the schema. To reduce researcher subjectivity bias, 2 experienced health engagement researchers completed inter-rater reliability checks on the data. The inter-rater reliability calculated using the Cohen kappa coefficient was 0.92 in SPSS Statistics 26 (IBM Corp). This result suggests a strong level of agreement between the coders. Finally, mHealth engagement data through face-to-face observations confirmed participant use, thus triangulating the engagement states through both interviews and observations.

To establish message content, the occurrence and salience of themes at each engagement stage were established. Increased repetition of a theme at an engagement stage suggested increased salience of the corresponding message content. Thus, the most common themes and codes at each stage of engagement were identified and arranged in a hierarchy of the most relevant to be used for personalized communication.

A total of 19 participants attended in-depth confidential interviews and structured participant observations using their mHealth technology with a researcher in sample group 1 (individual context) at the first period. Of the 19 participants, 12 (63%) were females and 7 (37%) were males, with age range of 18 to 49 years. The participants were invited to be interviewed and observed at 3 periods. At the second period, of the 19 participants, 15 (79%) participants completed the interviews and observations, with 11 (58%) participants agreeing to a third interview. In sample group 2, (group context) 25 participants were recruited from the GCC. Of the 25 participants, 18 (72%) were females and 7 (28%) were males, with an age range of 18 to ≥70 years. Again, each of the participants was invited to partake in the 3 periods. In total, 96% (24/25) participants completed both the second and third periods of data collection. Full participant details for sample 1 and sample 2 are provided in Table 1. Examples of technologies used across both samples included Fitocracy, Strava, Garmin Watch, Fitbit, or simple pedometers, and these were accessed via smart watches, mobile phones, and advanced trackers.

The framework by O’Brien and Toms [12] explored engagement in the context of web searching, web-based shopping, educational webcasting, and video games, finding 4 stages of engagement: point of engagement, period of engagement, disengagement, and re-engagement. Similarly, sample group 1 identified these 4 types of engagement plus a fifth engagement domain of self-management. Self-management refers to a stage where users had completely disengaged from their technology; however, they were still engaged with their health activity.

Furthermore, sample group 2 indicated an additional stage alongside self-management called limited engagement. This occurred when users had minimal interaction with their mHealth technology but engaged in the health activity to contribute to the group goals. Participants did not want to fully disengage as they felt the desire to stay in the program for others. Although limited engagement was discovered in the group context, it may also apply to solo engagement contexts. Thus, this research extends the framework by O’Brien and Toms [12] to include 6 stages of mHealth technology engagement, as seen in Figure 2. Importantly, these stages are not linear, with participants skipping stages based on motivation and accountability to engage, which are all features of a person’s wider engagement context.

When developing message content for sustained engagement, this research determined 8 themes that represent the main discussions impacting the 6 engagement stages. Each engagement stage is characterized by a different combination of these themes and 49 attributes. Figure 3 indicates the themes relevant at each engagement stage. These themes are presented in a hierarchical order based on their level of importance for each engagement stage. For example, in the limited engagement stage, the 5 themes of interruptions, social support, specific goals, technology features, and use motivations contribute toward shaping the level of limited engagement for the participant. The most frequently discussed theme for limited engagement was interruptions, with the lowest being use motivations. This indicates that interruptions are the greatest consideration for users experiencing limited engagement with mHealth technology. Multimedia Appendix 2 provides a full summary of the themes as they are expressed at each engagement stage.

On the basis of Textbox 1 and Multimedia Appendix 3, the point of engagement is characterized by the user’s attraction to the technical features of an mHealth device that helps them to fulfill their motivations to be healthy. At this stage, it is important that the technology is easy to use and promotes the support of a provider or a peer network. There was recognition that the mHealth device could improve users’ health awareness and their ability to reach their goals; however, these are less important than the features of the technology. Technology aesthetics was not highly relevant at the point of engagement.

Sustained engagement is characterized by the regular continued use of the mHealth device to support health and wellness activities. During this stage, technology features are important to achieve motivation, and this can be amplified through the support of a peer network or through provider messages. The continued use of an mHealth device encouraged users to have more health awareness. Usability and goal setting are still important in this stage but are less important than the features of the technology.

Limited engagement is characterized by users’ interrupted use of the mHealth device, with participants driven by social motivations to continue with the health activity. Social support through a peer network or provider messages was considered by users as influential during this period to achieve their health goals. Interruptions experienced by users were often caused by a feature of the technology; however, social interactions provided the motivation to continue.

Disengagement is characterized by discontinuing the use of an mHealth device and participants stopping the health activity. This was often caused by the mHealth technology not being suitable for the user’s needs and not providing motivation to continue. This mainly occurred because the mHealth device lacked features or was difficult to use. In addition, users did not receive the required social support from their device to develop sustained engagement. This was because of the inability of the mHealth technology to engage the user’s peer network or support provider messages. Finally, other interruptions also caused disengagement.

Self-management is characterized by users who are able to stay motivated and achieve their goals without their mHealth technology. Interruptions caused by the technology features meant that an mHealth device was not appealing and no longer used. In addition, as users were already active and aware of their health status, they did not need their mHealth technology for health awareness.

Re-engagement is characterized by users wanting to be motivated and wanting to achieve their goals. During this period, the technology features and technology aesthetics or the visual appeal of a device were important in the decision to re-engage. When re-engaging with existing or new technology, users were after specific technology features to help reach their health goals (eg, heart rate monitor and step counter). In addition, users were looking to re-engage with a mHealth technology that provides social support and helps them achieve health awareness.

A total of 8 themes were established in interviews across the 2 samples: technology usability, features, technology aesthetics, use motivations, health awareness, goal setting, social support, and interruptions when engaging with mHealth technology. To develop message content, an understanding of the themes, their attributes, and the relevant stage of engagement is considered below.

On the basis of current evidence, brief messages are cost-effective and can be personalized and delivered at scale in multiple settings to improve sustained health engagement. This research demonstrates that end-user mHealth engagement is complex, nonlinear, can be social or solo, and is characterized by 6 stages. Each stage is defined by themes and attributes that signify the total user experience. Fundamentally, this research calls for the consideration of interdisciplinary frameworks that incorporate health, information systems, and business to avoid the partial view of this phenomenon and to enhance engagement over time.

The limitations of this research are the small sample size from a single first-world country. Future research could measure the existing, and explore for additional, engagement stages using a larger sample and in different cultures. This would allow for more detailed investigations between groups, such as those users at different stages in the engagement journey, those displaying different demographic characteristics, or those using different types of technologies.

Further research could also include larger quantitative investigations that could be used to overcome the more exploratory and perhaps limited representative sample in this study. Future quantitative research could investigate interventions at each engagement stage or use experimental design to test the suitability of message content across these stages. The development of a short identification survey could also be undertaken, focusing on the 6 engagement stages. This would enable quick categorization of users and the stage they are at in their engagement journey, thus limiting the reliance on more longer time-intensive interviews for organizations that could result in more easily obtained targeted communication.