Development and testing of a mobile application to support diabetes self-management for people with newly diagnosed type 2 diabetes: a design thinking case study

We applied the methodology of design thinking, which is an innovative human-centered approach to developing new solutions [15]. Design thinking addresses the needs of the people who will consume a product and the infrastructure that enables it [15]; it takes into account the perspectives of multiple stakeholders. In this case, patients newly diagnosed with type 2 diabetes were the primary users of the app (referred to hereafter as ‘users’), but healthcare professionals were also stakeholders and were involved closely in the design process.

Design thinking focuses on rapid prototyping, which means turning ideas into actual products that are then tested, iterated, and refined, based on user feedback [15]. The study process was inspired by Brown and Wyatt’s three phases of inspiration, ideation, and implementation [15], which are depicted in Table 1. In practice, the phases overlap and iterate. We used qualitative methods such as observations, semi-structured interviews, and interactive workshops to promote participation with users and healthcare professionals in developing and testing the app. Tools and methods from the educational concept ‘Next Education’ (NEED) were applied to promote active involvement, reflection, and dialog in workshops [16‐18].

The inspiration phase focused on understanding users’ needs and challenges in everyday life after diagnosis with type 2 diabetes. Users were recruited in a general practice and a community health center providing patient education in the municipality of Copenhagen. We conducted an interactive workshop with healthcare professionals (e.g., general practitioners [GPs] and diabetes nurse specialists), observations of patient education in a community health center and individual consultations in general practice as well as 12 semi-structured individual interviews with users. We intended to include individuals who had been diagnosed within the previous 2 years, but we broadened this criterion to facilitate recruitment; users’ duration of disease ranged up to 10 years. We asked users who had been diagnosed longer than 2 years to focus on their needs and challenges in the period after diagnosis.

Data collection focused on three themes: everyday life with type 2 diabetes, communication with the healthcare system, and technology knowledge and readiness. An interview guide was compiled, based on recommendations from Brinkmann and Tanggaard [19]. The guide also included questions related to demographics, such as age, education, employment status, and marital status, and questions related to diabetes, such as use of medication, blood sugar monitoring and control, disabilities, and other chronic diseases (Table 2). The semi-structured interviews were inspired by the model ‘The Balancing Person’, which describes patients’ challenges with a chronic condition [20]. The interviews varied in length from 20 to 60 min and were transcribed verbatim; the findings were validated in a workshop with users.

In the ideation phase, we analyzed data and ransformed them into insights about innovative solutions for change [15]. Analysis was inspired by Malterud’s ‘systematic text condensation’ [21]. First, we captured an overall impression of all data and then identified preliminary themes. Secondly, we identified the meaning units relevant to the study question and sorted them into categories representing different themes. Thirdly, the units of meaning were sorted into subgroups and the meaning in each subgroup was refined and condensed. Finally, the content of the subgroups was synthesized to generate descriptions and concepts [21].

We conducted three interactive workshops with users and healthcare professionals. The purpose was to develop, discuss, and prioritize ideas for app content and design. We used tools from the Next Education concept, visualized ideas, and used dialog tools such as a flowchart of a user journey experience inspired by participants’ experiences and challenges. The number of ideas, app content, and design were adjusted and refined throughout the ideation process. Relevant app content was validated by the study collaborators in general practice, the community health center, the hospital and by the Danish Diabetes Association, the Danish Podiatry Association as well as the Eye Clinic at Steno Diabetes Center in Copenhagen.

The app prototype was pilot tested with users for a period of 4 weeks. Fourteen users were recruited from general practice and the community health center (Table 3). Further three users had agreed to test the app, but dropped out of the study before they had downloaded the app due to family circumstances, lack of acceptance and readiness in relation to a diabetes diagnosis, and technical issues. An inclusion criterion was access to an iPhone or iPad because the app was only developed for Apple’s iOS platform. Users agreed to test the app for 4 weeks and participate in an interview regarding usability and usefulness. To assist healthcare professionals in the recruitment process, we created both a script and an information letter to give to users who showed interest in the pilot study.

To access the app, users installed a secure development platform app that required an invitation with a username and a password. We observed the processes of recruitment and introduction to the app to gain insight into potential implementation challenges. For this purpose, we developed an observation guide that contained questions related to how recruitment was conducted in practice, the types of questions that potential users had, and the characteristics of people who declined to participate.

We conducted semi-structured interviews with 14 users. They focused on the participant’s experience with the app and also included data such as duration of disease, age, education, employment status, and marital status (Table 3). The interviews, which lasted 22 to 55 min, were transcribed verbatim. Furthermore, implementation issues were discussed in a workshop with healthcare professionals. The findings from interviews and the workshop were used to create a list of recommended adjustments to the app.

The analysis of data interviews, observations, and workshops from the needs assessment in Phase 1 revealed four themes: 1) diabetes – a real illness?; 2) lack of action competency; 3) non-transparent diabetes journey; and 4) lack of care coordination.

Gaining a deep understanding of users’ needs inspired two solutions for an app prototype. The first solution aimed to strengthen users’ ability to navigate diabetes care activities and coordinate information (health data) by providing an overview of vital diabetes care activities and stimulating reflection about their diabetes-related needs, goals, and challenges. The second solution focused on supporting newly diagnosed individuals in making health behavior changes and maintaining psychosocial health by addressing well-being, stress, and sleep. During discussions about the relative priority of these solutions with users and healthcare professionals, a collaborative decision was reached to integrate both solutions into a single app prototype (Fig. 2). After prototyping was completed, the app comprised five functions:

We identified four themes from the interviews with users and from a focus group with healthcare professionals about implementation. The themes were: 1) a viable tool to support diabetes self-management, 2) patterns of app use, 3) barriers and facilitators of app use, and 4) barriers and facilitators of implementation.

Simple and understandable design, content, and menu navigation are pivotal and seem to encourage app usability [14]. In addition, recent studies have observed a negative correlation between usability and apps comprising several functions [13]. According to Arnhold et al., the majority of apps offer similar functionalities but combine only one or two of them [14]. The most common functions in diabetes apps involve documentation of data, data forwarding, information function, analysis function, reminder function, but usually only one function per diabetes app [14]. Recorded data often concern psychosocial aspects (well-being), health behaviour (diet and physical activity) or diabetes specific data such as HbA1c, but not the interplay between these factors [13]. With the exeption of data forwarding, our app included all the mentioned functions. However, most users used only one or two functions but favored different functions and different types of recorded data. None of the test users mentioned that the app contained too many functions or that multifunctionaly inhibited their app use. In addition, a clear finding emerged in the design process that users preferred a multifunctional model with the ability to choose between different functions focusing on diabetes specific data and knowledge as well as psychosocial and health behavioural aspects. Similarly, Arnhold et al. argues that multifunctional apps combining documentation, reminder, and advisory functions are more suited for newly diagnosed individuals and elderly people with diabetes [14]. Our study also suggests the value of developing multifunctional apps for individuals newly diagnosed with type 2 diabetes, including the possibility of personalizing apps to individual needs.

Individuals who are newly diagnosed with type 2 diabetes are a diverse group in terms of technical skills, age, individual needs, preferences, diabetes knowledge, and interaction with different healthcare professionals. It is also unknown whether the benefit of the app we developed is confined to only those who are newly diagnosed and how long a diagnosis should be considered new. Some users in our study stated they would not have been ready to use an app during the first years after diagnosis because they had not accepted their diabetes diagnosis. In addition, users who had been diagnosed longer than two years found the function providing an overview of diabetes activities valuable because their diabetes care and prevention activities had changed dramatically since diagnosis. We did not find significant differences in terms of app use or preferences for functions when comparing those diagnosed within the previous 6 months or later. However, including more participants might have enabled analyses of the value of the app to different user groups.

The implementation process for an app is crucial for usability and effect [13]. Most health apps are downloaded by patients through online app stores, and some are introduced to patients as part of their contact with the healthcare system. Some apps are stand-alone solutions with the objective of supporting the patient, and others involve some degree of communication between patients and healthcare professionals or other patients.

In our study, it was evident in the preliminary workshops that a potential app would not be integrated with the separate information technology systems of GPs and community health centers because these systems are not interoperable. In addition, no healthcare professionals were interested in a supplementary system that would operate in parallel to their existing technology, nor did healthcare professionals feel competent or ready to use a mobile app in their consultations with patients. This correlates with other studies finding that lack of human and technical skills are barriers for integration of health technology in practice [8, 10, 22]. A survey among 173 health centers and clinics showed that the three main barriers to implementing cell phone interventions were limited human and technical organizational resources to support implementation, lack of external funding sources to finance investment in mobile technology solutions, and challenges to the technical integration of mobile health solutions with electronic health records and other health information technology infrastructure [22]. These barriers suggest the importance of including healthcare professionals in the development, testing, and implementation processes to create a sense of ownership among healthcare professionals and to identify organizational needs and possibilities.

Users in our study considered it crucial that healthcare professionals introduced the app because it created trust. Some users were afraid of using self-selected apps because they might contain out-of-date or incorrect information. This concern is rational because few apps are research-based, and they may not convey guidelines or content that have been validated by experts [8]. Another issue concerns health economic analysis of health apps. There is a need for studies about apps focusing on both benefits and disadvantages in terms of resources [23].

There are some limitations to this study. Only 14 users tested the app, and results cannot be generalized to all individuals with newly diagnosed type 2 diabetes. Three users stated that they would not continue to use the app. Due to the small number of test users, it is difficult to predict how many newly diagnosed that would in fact accept and use the app in a real life setting. Many users included in the study were well informed about diabetes-related activities, because 10 users were recruited from the community health centers, where recruitment was easier than through GPs. In addition, our criterion for identifying users as newly diagnosed was broad. Consequently, the app requires further testing.