For phase I, we used a nominal group technique to develop consensus on the app content and features. This included a number of separate FGDs for HCPs, younger patients (aged 10-15 years) and their parents, and older patients (aged 16-24 years). Following an introductory presentation on the overall goals of the meeting, YP and parents were asked what features and information will help them or their child better follow their treatment plans and cope with arthritis-related symptoms and consequences. The HCPs were asked what data can be collected remotely to help with understanding and assessing patient’s condition and treatment effectiveness. All participants were also questioned regarding the design and aesthetics of the user interface. Lists of questions posed to each of the groups are summarized in Multimedia Appendix 1. Consensus was considered as having been achieved when 75% of the participants endorsed a given answer. Answers that did not reach 75% endorsement were discarded or reformulated through discussion.

Participants’ answers to the questions posed during phase I FGDs were used to generate a catalog of initial system requirements. An app prototype was then developed in collaboration with the Computer Science Department at UCLH. The agreed set of features were iteratively built into the app codebase by a developer from the Computer Science Department. When clarification of the data generated was required, participants were queried by email and phone to provide further feedback (see Figure 1). The prototype was developed using open frameworks for cross-platform compatibility and runs on both the Google Android (version 4.4.1 and above) and the Apple iOS (version 8 and above) operating systems.

The FGDs were again conducted with HCPs and YP with JIA during phase II to vet and refine the requirements of the phase I app prototype. Participants were able to test and navigate through the app’s various features using a smartphone to improve their understanding of the app and focus on their recommendations. Screenshots of important pages in the app were also shown to examine whether YP and HCPs liked or disliked the design and the question format. They were asked to provide further suggestions for improvement, and technical problems with the app were also recorded.

Both HCPs and YP were also asked whether the information collected would be useful during consultations, and how clinically relevant information reported by patients should be imported to, summarized, and displayed in a Web application for HCPs. Also, what data recorded by patients and information on app usage should be stored for research were discussed. Semistructured interview questions that were developed based on the Usefulness, Satisfaction, and Ease of Use questionnaire [57] were used to guide the FGDs (see Multimedia Appendix 1). The design of the app was modified and new screenshots were generated and evaluated until no further changes were suggested. All FGDs were conducted with each participant group until point of data redundancy or point of no proposed new data [58] and lasted between 60 and 90 min. System analysis and final design were done in joint effort with the developer from the Computer Science Department.

Following the development of the final app, the usability and acceptability of the user interface, as well as the registration and data extraction process were tested with YP during phase III. The interviews with YP took place on a one-to-one basis and YP were shown the app on a smartphone. The researcher explained that the purpose of the study was to pilot ease of use and evaluate its features and encouraged YP to input information in all sections of the app at least once. After navigating the various features of the app and using it for 10 to 15 min, YP answered a usability and acceptability questionnaire [57] where they rated on 5-point scales their general impressions of the app, its user-friendliness, clarity of information provided, and whether or not they found the app useful and would recommend it to other YP (see Multimedia Appendix 2). In addition, YP answered qualitative questions regarding the specific features of the app, and how it can help them improve self-management (see Multimedia Appendix 1). Each interview lasted between 25 and 45 min. Discussions were audio-recorded for all phases of the project, and field notes were made for phase III. At all FGDs, RAC and HC were present as interview leader and observer, respectively.

The first theme was self-monitoring, as all participants thought that tracking certain information will (1) help them understand what influences their JIA symptoms and (2) provide better data for HCPs to use during assessments. This theme included six initial categories for monitoring; however, only those for which there was >75% agreement on its importance among YP, parents, and HCPs were included in the app. These were monitoring (1) symptoms, (2) general well-being, (3) activities, and (4) sleep. Categories that did not achieve this concordance were related to tracking (5) weather and (6) nutrition.

All participants identified tracking of joint symptomatology, including stiffness, swelling, and pain as very important, particularly as this varies in intensity, location, and duration over time. They felt that using the app would reduce recall bias and enable both patients and clinicians to monitor symptoms in real time:

One patient stated:

Participants indicated that the app should include continuous visual analog scales (VAS) to indicate the intensity of JIA-related symptoms by moving a bar across a horizontal line in preference to categorical selections. One adolescent stated:

As requested by YP, each scale was accompanied with a cartoon face that changed facial expressions from extremely positive at the left to extremely negative at the right. Psychometric properties of the VAS have been widely explored with children as young as 8 years of age and have confirmed its test-retest reliability, construct validity, and criterion-related validity [62,63].

Additionally, YP wanted to use a body map to indicate how painful and swollen a specific joint is. Body maps can facilitate children’s reports of location of pain [64,65], and electronic versions have been used successfully with children as young as 8 years old [40]. The app will include a simple body map (see Figure 3) with circles at major joint areas in the body, mirroring what is used in routine clinical assessments [66]. YP can click on a joint that is painful or stiff and use a VAS to indicate the intensity of their pain and stiffness, and the joint’s color will then change accordingly. This is important to include, as YP often experience different pain intensities from different joints, and their stiffness intensity also vary by location and time of day. Both YP and HCPs thought that recording this will help them better explain their symptoms and assess whether treatment is required (eg, physiotherapy exercises targeting specific joints). Lastly, YP requested an open text area where they can input additional details regarding their symptoms and add comments about the type of pain, with one of the patients stating the following reason:

Being able to monitor general well-being such as mood, fatigue, and functional health were also identified as important. For example, YP indicated:

Three positive (excited, energetic, happy) and three negative emotions (anxiety/fear, sad, angry) were chosen together with YP and HCPs, which were adapted from the Positive and Negative Affect Schedule for Children [67]. These were assessed using the VAS and the horizontal line ranged from “not at all” on the left to “extremely” on the right. A stress scale was also included, but a verbal rating scale for stress (VRSS) was preferred by YP instead of a VAS. The VRSS was found to be more reliable than the VAS and can be easily understood and completed by children as young as 7 years old [68]. It measures stress on a scale from zero to five with responses describing an increase in stress levels. In addition, YP wanted an electronic version of a JIA-related childhood health-assessment questionnaire (CHAQ) [69], which is a widely used functional health status measure in YP with JIA and is administered during routine clinic consultations. They recognized that completing it once every 3 to 6 months during clinic consultations may provide unreliable responses, as it does not reflect variations over time:

Digital CHAQs have been developed and tested in previous studies [63] and were found to yield similar outcomes when compared with the paper form of CHAQ.

Recording activities such as participation in school and physical exercise and sports was another important area for YP. One YP said,

Another stated,

YP suggested that the app should provide visual feedback of their data so that they can see how different factors impacted their condition, and what techniques were helpful in managing their symptoms.

Participants were interested in using the app to improve treatment adherence by setting reminders for when to take regular medications, request repeat prescriptions, complete physiotherapy exercises, and attend hospital appointments for important assessments and tests (eg, regular blood tests). One of the patients said,

However, some participants felt that these reminders should be optional to avoid any disengagement with the app because of annoyance from compulsory reminders. Furthermore, YP were interested in recording when medication doses were missed, reasons for missing doses, side effects, medication change, and whether they took any additional medicines such as analgesics or antibiotics and anti-emetics (see Figure 4). Regarding physiotherapy exercises, YP could select which ones they were assigned or would like to perform from a list of possible exercises recommended by physiotherapists at GOSH and could input the number of sets and repetitions. Additionally, both patients and HCPs wanted a section where patients could input their blood test results, with an indication of whether these results were normal and what the tests measured:

Besides wanting to learn more about their condition, YP wanted to have a discussion forum in the app with other patients. Also, YP from all age groups highlighted the need to have easy access via the app to condition- and treatment-related information, news about research, and social support:

Additionally, YP said that learning different ways of managing their stress and pain from peers, and knowing that they are not alone, will make them feel more empowered and confident. Moreover, YP mentioned that their arthritis symptoms tend to worsen during times of stress and exams and would like information on how to better manage stressful situations and pain. Therefore, the app will include general information regarding JIA, medications, physiotherapy exercises, and practical coping advice, with links to more detailed and trusted Web resources (see Figure 4). It will also signpost users to information regarding support groups that are reputable and age-appropriate, as well as possible involvement activities organized by local organizations and charities.

Older patients (18-24 years) requested additional information on how to become more independent as they become an adult and what support is available for higher education and work. The app will provide general information covering these topics, with links to how other YP with JIA overcame challenges of growing up. In addition, there is a diary section where YP can write down any questions or comments that they have and can easily refer to during consultations.

It was observed that YP wanted to be rewarded for using the app, and they preferred a reward-based rather than a punitive system. One of the patients stated:

Therefore, the app will incorporate a point-based reward system and include encouraging video messages from celebrities and athletes. However, this was more appealing to the younger group rather than the older group. Older group enjoyed choosing reward pictures (eg, of landscapes, animals, and sports) to use as their wallpaper and motivational quotes that they could display on the home page.

In terms of designing the user interface, three aspects were important for YP and HCPs: (1) data security, (2) visual appeal/clarity, and (3) convenience. First, as the app will contain patient’s personal details, it needs to be password-protected:

Also, YP wanted the option of diarizing information that would not necessarily be shared with clinicians:

Second, YP wanted a colorful visual display, easy-to-read fonts, and pictures and were keen on being able to personalize colors and backgrounds. One participant stated,

Participants also suggested that scales should include faces to represent different pain or fatigue levels:

Moreover, YP wanted more space on the home page to display reward videos and pictures on the bottom; therefore, the home page displays a dial (instead of a drop-down ladder) to count down the number of days that are left until the next weekly deadline (see Figure 4).

Lastly, there was a consensus among YP that it should typically take less than 10 min to input all required information:

Moreover, YP felt that there needed to be an agreed minimum input to provide clinicians with interpretable data. One adolescent stated:

It was agreed by both YP and HCPs that the most important data to collect were arthritis-related symptoms, CHAQ, and mood, which should be provided at least once a week. The frequency of completion of other sections should be decided by the YP themselves to give them control over the process.

Ease of use was thus improved in several important ways. First, YP were now able to access the three main monitoring components of the app (symptoms, CHAQ, and mood) directly from the home page rather than from the menu bar. Second, after submitting each input, YP were presented with a pop-up showing how many days ago was their last input for each of the other section, which allowed them to quickly select another section that had not been completed recently. Similarly, hyperlinks were added within the app itself to facilitate access to relevant tips and advice. YP confirmed that these changes made inputting data and finding relevant information faster and more convenient. Lastly, YP requested a calendar display of entries, and the ability to input data retrospectively. One patient said,

All the themes and their associated app functionalities were mapped onto the COM-B model (see Table 2).

The HCPs discussed how JIApp can be integrated seamlessly into clinical practice and how reports from patients will be received. Clinically relevant data collected from the app will be automatically aggregated and presented in table and graphical summaries (see Figure 4) on a clinician Web interface. The interface was optimized for usability and data are presented in an intuitive format such that any changes over time or associations between various factors (eg, pain and mood) are visually presented to aid comprehension. Figure 6 illustrates the system, where raw data recorded by YP are stored and clinically relevant information can be summarized and displayed on the clinician interface.

The within-application workflow was also aligned with clinician’s regular work streams. For example, HCPs can access the Web interface via a patient’s existing electronic health records, which can increase the likelihood of integration and access from all multidisciplinary members of the rheumatology team. Information can also be downloaded by HCPs and used to populate clinic letters.

The HCPs also identified possible barriers to integration such as accessibility to all patients and demands for additional resources. For example, HCPs proposed that a desktop version of the app will be beneficial to allow access to YP without a mobile phone. In addition, HCPs were concerned about the additional time needed to monitor remote symptom deteriorations and interpret additional information collected from the app. It was agreed that HCPs are only expected to review the data during or right before a clinic consultation. Therefore, before downloading the app, participants will be informed that data will not be monitored in real time for this stage of the project but will be reviewed during clinic consultations. This is also explicitly explained on the app with a disclosure statement. The HCPs suggested that the effect of app use on HCP-patient interactions and cost-effectiveness of the intervention should be formally assessed in the next stages of the project during a prospective pilot study.