Diabetes has increasingly become a major burden for industrialized societies, with rising health care costs and mortality rates pressuring governments globally to address this problem [1,2]. These governments have finally started to recognize the problem’s seriousness. Singapore, the first country that we examined, has the second-highest diabetes prevalence rate after the United States [3], and it launched a war on diabetes in 2016. Germany, the second country that we examined, ranks third for diabetes-related health expenditures worldwide [2,4].

With the rapid growth and ubiquitous availability of mobile phones, mobile health (mHealth), that is, “the use of mobile communications for health information and services” [5], can potentially contribute to improving health promotion, disease prevention, and disease self-management [6-10]. In the diabetes context, functions such as messaging and chatting with health care providers, connections to external devices (eg, heart-rate measurement and monitoring of blood glucose or blood pressure), and support of medication, as well as tracking physical activity or nutrition behavior via mobile apps are discussed [11,12].

Up to this point, research has mainly focused on mHealth’s effects, indicating not only promising results overall [13,14] but also contradictory empirical evidence [15-19]. Even a recent systematic review concluded that research is still too heterogeneous and somewhat too low in methodological quality to “provide reliable evidence of effects for stakeholders” [12]. To gain a clearer picture of mHealth’s potential in the diabetes context, it is necessary to understand potential users’ needs and expectations, as well as the factors determining their mHealth use. In recent years, more and more studies have dealt with the user perspective in the mHealth context [20-39]. However, most of these studies have fallen short in their ability to explain more than just adoption factors.

Our goal is to describe different patterns of everyday life integration that go beyond mere usage metrics or the simple dichotomy of adoption versus rejection [40]. To achieve this goal, we draw on the mobile phone appropriation (MPA) model [41] as our theoretical frame. Empirically, we conducted semistructured interviews with diabetes patients in Singapore (study 1) and Germany (study 2). Study 1 identified relevant functional, normative, symbolic, and restriction evaluations tied to diabetes app use, whereas study 2 complemented these evaluations for mHealth appropriation and identified supplemental patterns of evaluation, use, and meta-communication.

Current extant research on mobile devices’ and services’ effects on health outcomes often focuses on SMS text message interventions [14,42]. Increasingly, studies have concentrated on smartphone apps’ effects on diabetes self-management [12,13,15]. However, their results are diverse, ranging from positive effects on diabetes outcomes, for example, hemoglobin A_1c reduction [43], to limited or no effects [44,45]. Moreover, recent systematic reviews are not consistent but reveal positive effects overall [13,14]. Studies have increasingly tried to specify effects by varying message design and tailoring messages to users’ needs [46-49], with promising results. However, also in this context, a clear picture cannot be drawn so far. This may be explained, at least partly, by an overly simplistic idea of use and effects. Using apps or receiving certain messages does not tell us how users interact with apps and interpret their functions [18].

Thus, the need exists for a better understanding of actual everyday-life use and mobile devices’ integration into diabetes self-management. To increase our knowledge about mHealth’s role in the diabetes context, we need to ask how, why, and for what purpose patients use mobile devices for diabetes self-management and which motives, perceptions, and expectations drive their use.

Primarily, the answers to these questions require that we define what we mean by use. In extant literature, use has been used quite heterogeneously, describing all kinds of processes and subprocesses in decision and implementation phases, as defined by Rogers’ [50] innovation-decision process. These 2 stages help distinguish between 2 broad areas of use: first, Rogers describes the decision stage (adoption), in which the overall question of use versus nonuse (ie, adoption vs nonadoption) is tackled. The second phase of implementation (appropriation) deals with the question of everyday-life integration and actual use patterns [50,51].

In recent years, more and more studies have focused on mHealth adoption, drawing mostly on the technology-acceptance model (TAM) [24,25,27,33,37,51,52] or its successor, as well as the unified theory of acceptance and use of technology (UTAUT) [30,35,38,39,53], to explain influences on the adoption decision. However, only a few studies have considered an implementation or appropriation perspective on mHealth use, mostly focusing on continued use in contrast to the single-adoption decision. These studies fail to consider the multifaceted patterns of everyday-life integration [21,26,37].

In the diabetes context, most studies that concentrate on the implementation process are rather descriptive in nature, examining frequency of use or expectations regarding diabetes apps [20,22,29].

The MPA model [41] provides a theoretical framework with which to analyze the appropriation of mobile media as a process (Figure 1), not only in general but also in specific contexts [54]. The model integrates concepts of technology adoption, for example, diffusion of innovations [50], theory of planned behavior [55], TAM [52], and UTAUT [53], with conceptualizations of the actual use and implementation of technological innovations into users’ everyday lives. On the basis of cultural studies [56,57] and the domestication approach [58], Wirth et al [41] term this process appropriation, emphasizing users’ active co-construction of meaning, thereby overcoming the binary logic of adoption [40].

The MPA model conceptualizes appropriation as a creative and active process, resulting in various use patterns by individual mobile-media users. Consequently, behavior is differentiated in sub-constructs in the MPA model, comprising symbolic and functional aspects. The functional aspects represent the variety of uses, for example, highlighted in research on the basis of the uses and gratifications approach. By adding symbolic aspects such as prestige, the MPA captures the concept and extent of observability [50], making the choice and use of mobile media a continuous statement about oneself in public [59]. Functional, symbolic, normative, and restriction evaluations influence these use dimensions. Functional and symbolic evaluations represent users’ beliefs about the functional and symbolic aspects of their future mobile-media behaviors. Normative evaluations refer to their beliefs and judgments about social norms related to their future behavior. Restriction evaluations—comprising financial, technical, cognitive, and temporal factors—represent users’ beliefs about constraints hindering their future mobile-media behaviors. Restriction evaluations find resonance with the information communication technologies for the health care model used in mHealth studies [60,61], which propose economic, technological, infrastructural, and sociocultural barriers.

In addition, the MPA model integrates meta-communication, that is, the impact of communication on communication technologies. As users communicate about their respective uses of mobile media and observe others’ behaviors, this meta-communication influences their future behaviors. Consequently, the MPA model is conceptualized as a cycle, with the appropriation being a constantly renewed process. Functional and symbolic mobile uses are not only the results of behavioral, normative, or restriction beliefs but they also become the basis of those beliefs [62].

So far, the MPA model has been adapted successfully to the mHealth context in 1 study examining patterns in nutrition-app appropriation [63]. On the basis of a Web-based survey of nutrition app users, the study identified 4 distinct appropriation types: supported, indifferent, health-conscious, and socializer. These types differed mainly regarding (1) the support they received from their social peers for their app use, (2) their personal attachment to their app use, and (3) app use for socializing (and competition). Thus, we see great potential in using the MPA model as a theoretical framework to gain insight into the appropriation processes of mobile services and devices for diabetes self-management and ask these research questions (RQs):

To answer these questions, we carried out 2 related yet independent studies that focused on (1) different aspects of the appropriation process and (2) 2 different cases (Singapore and Germany). Study 1 lays the foundation for investigating app appropriation, and study 2 expands this notion by looking at the broader picture of mHealth for diabetes self-management.

Study 1 (Singapore) focused on the appropriation and use of diabetes-specific apps for self-management (diabetes apps). These apps are designed specifically to support diabetes self-management, including tools for blood sugar monitoring and direct feedback (diabetes log books, eg, Glyco App), diabetes information (eg, MySugr Academy), or food databases (for counting carbohydrates to adjust insulin, eg, the Singaporean Health Promotion Board’s food database). We carried out 21 semistructured, face-to-face interviews [64,65] (approximately 1 hour each, in English) with Singaporean type 1 and type 2 diabetes (and prediabetes) patients between December 2015 and September 2016. Singaporean diabetes-support groups (Diabetes and Diabetic Society of Singapore) were used to contact diabetes patients. The participants were asked to choose the interview locations to make them feel as comfortable as possible throughout the interviews.

The interviewees were recruited in such a way that the greatest possible variance in demographic characteristics—gender, age, diabetes type, period since diagnosis, and form of therapy—could be covered [65,66]. The prerequisites for participation were the following: having an existing diabetic condition and being a Singaporean resident. Using such a broad cross-section of participants was undoubtedly a challenge, but only in this way was it possible for us to grasp the combination of different characteristics as widely as possible to gain insight into the use of mobile media in the diabetes context. Focusing on 1 or 2 characteristics (eg, only 1 diabetes type, 1 age group, or 1 type of treatment) would have restricted this view. Table 1 provides an overview of sample characteristics, and Table 2 provides additional information on participants. Moreover, it should be noted that 8 participants suffered from other diseases in addition to diabetes (eg, heart conditions, high blood pressure, high cholesterol, hypothyroidism, and/or breast cancer), 17 participants received diabetes education at some point, and 15 were part of a diabetes support group.

Both in the construction of the interview guide (see Multimedia Appendices 1 and 2) and in the analysis of the interviews, we followed a theory-driven approach, which differs from classic grounded theory or hermeneutics [67]. The interview guide was based on the MPA model and assessed the diabetes context, general daily diabetes self-management, and the use of Web-based (mobile) devices as part of diabetes self-care. It included 30 flexible questions, that is, if answers to a specific question had been provided before, the question was omitted. All interviews were transcribed. The resulting transcripts each totaled 5000 to 10,000 words and were analyzed following a theory-driven approach on the basis of the research questions and the MPA model [41]. The data analysis followed a thematic-analysis approach as described by Braun and Clarke (2006), who define thematic analysis as “a method for identifying, analyzing, and reporting patterns (themes) within data” (p. 79), and it can be called “theoretical thematic analysis” because the themes are derived (at least partly) from the theoretical background, in contrast to an inductive approach [68]. The analysis was based on the categories’ functional, normative, symbolic, and restriction evaluations as described in the MPA model, and it used the interview data to build themes around these theoretical concepts to identify commonalities and differences, as well as understand the appropriation of diabetes-specific apps for self-management in detail. If important interview extracts did not fit into existing categories on the basis of the MPA model, new subcategories were added. The procedure of creating categories and themes was dynamic and constantly adapted on the basis of the interview content. An institutional review board approval was received by NTU Singapore for the face-to-face interviews.

On the basis of the results obtained from study 1, study 2 asked for patterns of mHealth appropriation that can be found in the context of diabetes self-management. From June 2017 to August 2017, we conducted 16 semistructured interviews with German individuals with diabetes [64,65]. They were recruited through a purposive-sampling approach to cover a variance in the characteristics of age, diabetes type, period since diagnosis, and treatment [69-71]. We recruited interviewees via doctors in private practices and hospitals in Munich and Jena. Tables 3 and 4 provide more information about the sample.

The interview guide was based on our theoretical assumptions [41], study 1’s results, and the interview guide developed for study 1. We adjusted it for language, as well as a broader understanding of mHealth use beyond specific diabetes apps. We asked the participants about (1) their smartphone use, (2) their knowledge about mobile media, (3) their attitude toward the use of mobile media in the context of diabetes, and (4) their self-management of their disease with the support of mobile media. The interview guide covered 21 questions.

The interviews lasted between 30 min and 60 min each, and they were audiotaped and transcribed into written form. The transcripts covered between 5000 and 10,200 words per interview. Our analysis was based on our paper’s theoretical concept (MPA model) and study 1’s results. We analyzed the interviews following the data-analysis process suggested by Creswell [71]. We read all transcripts, marked relevant passages, and abstracted them until we found the dimensions of mediated communication, diabetes self-management, and social prestige and control, ordering the presentation of the results. All responses were allocated to these 3 dimensions, and we identified similarities and differences among participants’ usage patterns and linked study 2’s results to the MPA model. The Research Support Office of University Hospital Jena checked the interview guide and approved it. Information on sample characteristics can be found in Table 3, and the participants are described in Table 4.

Type 1 patients used the Freestyle Libre device by Abbot (n=5), Continuous Glucose Monitoring by Dexcom (n=2), or Accu-Chek by Avia with Contour-App (n=1) to test blood sugar levels. Only 1 patient relied only on test strips. Type 2 patients did not self-test blood sugar levels. Participants used IOS (n=6) or Android (n=10) devices. Only Andreas, aged 71 years and Linda, aged 42 years, used smartphones older than 5 years. In total, 9 participants used smartphones, which they had for less than 2 years.

Our first research question inquired about which specific functional, symbolic, normative, and restriction evaluations were relevant in the context of mHealth appropriation for diabetes self-management. Patients generally differed in their appropriation of apps designed for diabetes, with their use ranging from no previous use and no knowledge about existing diabetes self-management apps to infrequent and short-term app use as well as to long-term app use. The length of app use varied significantly, from a few days to several months or even years. Almost all type 1 diabetes patients reported using diabetes apps, whereas just a few type 2 patients used diabetes apps for their self-management. In addition, the interviews revealed that high-risk diabetic patients, that is, those with critical conditions or insufficient self-management, did not use diabetes apps (eg, Li Ting, aged 49 years; Ming, age unknown; Rei Hong, aged 61 years; Zhen Wei, aged 47 years).

In summary, the interviews in Singapore revealed that evaluations of diabetes apps’ usefulness for self-management differed largely in the sample and across patient types (eg, motivated vs unmotivated patients). Some participants found diabetes apps to be useful for daily self-management, used 1 diabetes app over a longer period, switched among different apps, or used various apps concurrently. Other participants did not perceive diabetes apps to be useful and stopped using them after the first trial or did not try apps for self-management at all. Although most participants tested or used diabetes apps, our results show that other mobile services and devices that are not necessarily diabetes-specific (eg, fitness trackers, dietary apps, and instant messaging) are used in addition to or instead of diabetes apps. Thus, study 1 indicates that Singaporean individuals with diabetes do not use diabetes apps exclusively but rather make use of the broader mobile-media ecosystem to manage their disease. Therefore, in the next study, we broadened our focus beyond specific diabetes apps to other tools in the mobile ecosystem.

We identified individual evaluations and, unlike study 1, further synthesized them into distinct appropriation patterns following the MPA model’s [41] logic to answer our third research question. In total, 3 overarching dimensions of mHealth appropriation for diabetes self-management emerged: mediated communication comprising the functions of information gathering and social connectedness, diabetes self-management, including self-treatment, testing, and lifestyle management, and social prestige and control.

Our objective was to provide a more comprehensive view of mHealth use for diabetes self-management beyond the simple question of adoption versus rejection. Study 1, conducted in Singapore, specifically aimed to identify evaluations of diabetes-app use and appropriation for self-management. Most importantly, study 1 revealed that diabetes patients do not merely use specific diabetes apps for their daily diabetes self-management but rather make use of their whole mobile-media ecosystem, such as other health apps, chat apps, or Web-based databases. Building on these findings, we broadened our perspective in study 2, conducted in Germany, more generally examining not only diabetes-app use but also mobile-media use in diabetes self-management, including information search and retrieval, and monitoring linked to diabetes self-management.

The semistructured interviews revealed several functional, normative, and restriction evaluations that play a role in both studies, whereas symbolic evaluations only appeared in study 2. In particular, the following functional, symbolic, normative, and restriction evaluations were identified (RQ1):

As proposed by the MPA model in the context of mobile phone appropriation, meta-communication also plays a role in diabetes self-management with diabetes apps (RQ2). Interestingly, diabetes patients discuss topics around diabetes management mainly with other diabetes patients, in online diabetes support groups or via WhatsApp. In contrast, physicians only play a minor role, as they are reluctant to recommend or mention diabetes apps.

Regarding RQ3, study 2 revealed 3 overarching appropriation patterns emphasizing that the use and everyday-life integration of mHealth in diabetes self-management are not restricted to simple diabetes-app use:

Thus, studies 1 and 2 complement each other in 3 aspects:

Further research also is needed to address a few limitations. First, we conducted the 2 studies in 2 different countries despite the fact that we did not intend to compare appropriation processes. On the contrary, it was our objective to extend research that we conducted in Singapore both to a broader perspective on mHealth appropriation and to another country to assess recurring patterns in 2 different cases. The 2 countries are comparable regarding the relevance of diabetes, industrialization, smartphone penetration, and the presence of a well-developed health system. However, cultural differences must be acknowledged. This might explain why we did not find symbolic evaluations in Singapore, whereas in Germany, social prestige evolved as a relevant aspect of mHealth appropriation. It also should be noted that even if both countries rank among the top nations regarding technological development, it remains unclear how widely attitudes toward technology differ in Singapore versus Germany. Our results did not reveal any major differences regarding smartphone use for diabetes self-management; nevertheless, future research should consider this aspect in more detail. In addition, we did not systematically control for the influence of socioeconomic status and thus cannot make any statements on this. Finally, because of rather small sample sizes, qualitative research generally lacks generalizability in the results. However, a qualitative approach was a necessary step to gain deeper insights into diabetes patients’ appropriation patterns and everyday lives and to identify evaluations relevant to them [77]. In the next stage of research, our findings can be used to develop standardized questionnaires to gain insights into the distribution of appropriation patterns among a broader population.

The study of mHealth apps for diabetes management is in a nascent stage with not only promising results but also many open questions [12-19]. In this project, studies 1 and 2 revealed that appropriation of mHealth for diabetes self-management is not limited to using specific diabetes apps but rather includes patients’ entire mobile-media ecosystem. Even if diabetes apps play a role, especially for self-treatment and self-testing, diabetes patients use many more digital resources when dealing with their conditions, such as lifestyle apps, messenger apps, traditional health-information websites, or forums accessed from a computer or mobile device. Thus, mHealth is important for diabetes self-management but in multiple ways that go far beyond diabetes-app use. In addition, our findings indicate that mHealth cannot substitute for interpersonal communication, for example, with other patients, peers, or health care providers, but it complements and supports interpersonal communication, especially via messenger apps. However, doctor-patient communication only plays a minor role in this context and can even be a barrier to mHealth use, as doctors are reluctant to recommend using mobile apps. Further reasons for diabetes patients to be reluctant to use mHealth for diabetes (continuously) include financial, technical, cognitive, and temporal issues. Apart from individual constraints (eg, eye problems, technical skills, and use of outdated devices), it appears that patients still cannot find what they are looking for. Patients want a not-yet-existent app that combines everyday-life requirements while computing correct insulin doses, thereby allowing for more freedom and spontaneity. Thus, considering these recurring evaluations, the appropriation of mHealth for diabetes self-management could be enhanced if doctors were less reluctant to recommend self-management apps, if connectivity between devices and apps, including multipurpose apps, was improved, if apps adapted to users’ physical restrictions (eg, font size), and if people already using mobile media for diabetes self-management were integrated into new apps’ development process and used as consultable opinion leaders to facilitate appropriation for others.

However, even if some overarching issues evolve, not every app solution can accommodate each and every patient. This is reflected in the 3 overarching dimensions of mHealth appropriation for diabetes self-management that we uncovered: (1) mediated communication (information-gathering, connectedness, exchange), (2) diabetes self-management in a narrow sense (self-treatment and monitoring, lifestyle-management), and (3) social prestige and control (symbolic aspects of mHealth use for diabetes). Diabetes patients differ highly in respect to these dimensions, that is, how they use mHealth for diabetes self-management, for what purposes they use it, and how they evaluate use. Integrating this knowledge into future mHealth apps’ designs and effect studies might shed more light on mHealth’s great potential for diabetes self-management.