Community-Driven Priorities in Smartphone Application Development: Leveraging Social Networks to Self-Manage Type 2 Diabetes in a Low-Income African American Neighborhood

Social networks have the potential to enhance Type 2 Diabetes Mellitus (T2DM) self-management. We used qualitative methods to study if and how mobile application (app) functions that mobilize social resources to improve T2DM management would be desired in a low-income African American community. Data were collected through community discussions and in-depth interviews with 78 participants in 2016–2018. Participants included individuals with self-reported pre-diabetes, T2DM, close family members or friends of a T2DM patient, and healthcare providers. Open-ended questions solicited information about challenges with T2DM management and gathered ideas on features of a mobile app that could address them. Data were transcribed and thematically coded by two coders using Atlas-ti. Regarding types of app functions, main themes included: (1) the importance of having support in diabetes self-care; (2) using informal networks to help to each other; and (3) monitoring one another through an app. Suggested app features included reminders for and transportation to medical visits, sharing information and exercise companionship, and providing opportunities for monitoring by friends/family members, especially in case of emergencies. Participants viewed an app as a potential vehicle for reinforcing accomplishments in T2DM self-management. Future research should implement and test an app with these features in this or similar communities.


Introduction
Diabetes mellitus is a chronic metabolic disease that requires ongoing glycemic control and medical care [1]. Affecting 30.3 million, or 9.4%, of the U.S. population, diabetes was the seventh leading cause of death in 2016 and contributed to 80,058 U.S. deaths [2]; an additional 33.9% or 84.1 million American adults are pre-diabetic [2]. Type 2 Diabetes Mellitus (T2DM) accounts for 90-95% of all diabetic cases and its prevalence has been rising among low-income minority communities [3][4][5]. Poor glycemic in-depth interviews. Twenty health care providers participated in either in-depth interviews or a focus group discussion. After app design concepts were created from community forum and provider data, additional patients and family/friends participated in pre-usability forums to discuss design concepts. Recruitment was spaced over the course of two years for two main reasons: (1) difficulties in recruitment and (2) simultaneous development of an app prototype that was iteratively based on feedback. Many community members had multiple part-time jobs, and little access to affordable transportation, which made it difficult to find times when people were available and to find locations that were easily accessible for people who were interested in participating. Simultaneous technological development of the app was also conducted in parallel to, and in response to, the collection and analysis of the data. Therefore, data collection was paced to coordinate with the development of app features.
Consistent with the high incidence of T2DM in the community, some non-provider participants with pre-diabetes or T2DM were also friends/family members of other participants with pre-diabetes or T2DM. For categorization, such participants are reported here as adults with pre-diabetes or T2DM. All but one non-provider participant self-identified as African American. Most (n = 59) participated in one of six discussion group forums (averaging in size n = 9 participants), but a few non-provider individuals participated in one-on-one interviews (n = 4). Four individuals from early forums also participated in later pre-usability forums.
Data from providers were collected through nine individual in-depth interviews (n = 9), one focus group (n = 9), and one paired interview (n = 2). Providers included diabetes educators, pharmacists, pharmacy student, nurses, and a physician's assistant, as well as physicians (primarily from health care facilities in the study area).
The study took place from February 2016 to May 2018, and included 7 forums and 15 in-depth interviews. Participant recruitment was done in collaboration with community organizations (e.g., residential buildings, a neighborhood supermarket), primarily through distributing flyers and describing the project to community members attending meetings, community events, and shopping at a local supermarket. Forums generally lasted 3 h, while interviews lasted approximately 1 h. Forum participants were given a $50 or $40 gift card, depending on the length of the forum, and $30 for participation in interviews.
Forums consisted of a series of sequential, interactive group discussions around experiences and barriers/problems to T2DM self-management, from either a first-person perspective or that of a close family member or friend, with the stated goal of creating a smartphone app that could help with T2DM self-management. A moderator (with qualitative research training at the graduate level) used a semi-structured discussion guide and was accompanied by a research team (representing the fields of public health, anthropology, and computer interface design) that took notes. The open-ended discussion guides asked about common problems with diabetes self-management and solicited ideas regarding features of the mobile app that could be used to help with such challenges.
The principle goal of the first set of forums was to gather ideas about features of the app that would be useful to users, while later 'pre-usability' forums aimed to achieve this goal while also reviewing and refining design concepts developed from earlier forum data. We used active moderation as well as structured record keeping based on audio recordings, verbatim transcription of all recordings, and digital photographs of artifacts (e.g., group-generated lists of desired functions, constraints on the design, drawings, sketches of how a diabetes app should be organized). Forums were supplemented by in-depth interviews with four community residents with T2DM. Interview topics were similar, focusing on neighborhood barriers and facilitators to T2DM self-management. Interviewers also solicited participants' ideas about which features of an app could be useful for them.

Data Analysis
Data analysis was based on thematic content analysis [25]. We primarily used an inductive approach to explore: (1) major areas and sub-areas of problems in diabetes self-management; and (2) ideas/solutions that could be implemented on a mobile app.
Data transcription was concurrent with data collection, and themes were reviewed and discussed regularly by the research team, so that later forums and interviews could build on earlier findings [26]. Coding was undertaken by two study team researchers, with initial coding done with Atlas-ti computer software (Version 7, Scientific Software Development GmbH, Berlin, Germany), followed sequentially by further code refinement by a second team member. Additional themes evolved throughout the analytic process, as more patterns were identified. Multiple coders allowed for the discussion of differences to reach consensus. Memos were used throughout the analysis to document and share intermediate results, and the constant comparative method was used to facilitate theme development [27,28].
The Johns Hopkins Bloomberg School of Public Health (FWA 00000287; IRB X 00000758)) determined the study protocol to be Exempt, Category 2 (45 CFR 46.101(b)). This designation was made based on the fact that participants were not identified in recorded data and that disclosure of human subject responses would not place the participants at risk. All participants provided informed consent prior to participation.

Results
We conducted qualitative research with 78 participants to inform the development a mobile phone app for people with T2DM. Fifty-eight were self-identified as pre-diabetic, as having T2DM, or self-identified as being a close family member or friend of a T2DM patient who helps with diabetes care.
Participants included adults with self-reported pre-diabetes or T2DM (n = 31), close family/friends of people with T2DM (n = 27), and healthcare providers in the study community (n = 20) (see Table 1). An overarching theme was the importance of social support for diabetes self-management in this community. Themes also emerged around ways that an app could be used; these focused mainly on motivating healthy behaviors, providing support to each other, and monitoring each other's progress (see Table 2 for all themes and subthemes).

Importance of Having Support in Diabetes Self-Care
A major theme was the importance of social support for T2DM management. Many participants highlighted how having social support helped them maintain healthy behaviors around both exercise and nutrition.
"Everyone can piggyback off one another and develop not just healthy eating, but exercise together . . . Knowing you have a network is huge, because then you are not alone." (Forum 4) Most participants viewed an app as a potential way to foster support to help them take part in diabetes management activities like exercise. Other participants stated how exercising with friends can be motivating and enjoyable.
"It fun, you don't even realize you're working out, doing it for your health...we need that support to help motivate us, to get some of the stress, to exercise, and to be in better health." (Forum 1) Some mentioned the positive impact of social support on food choices. Several providers reinforced this idea that an app could motivate self-care of T2DM by enhancing social support through companionship. One provider explained: "The other thing that anecdotally that patients tell me is that 'I'm not motivated to go by myself,' but if they have a partner or buddy or somebody [things would be different.]" (Provider Interview)

Using Informal Networks to Help Each Other
A number of participants provided ideas for how a support network could facilitate reminders and transportation for going to healthcare visits. As one provider suggested: "Like any chronic condition, if you have people around you who can help you, remind you to 'do this,' 'this thing's bad for you,' [and provide] some family support. Some people have no one to take them to see the doctor when they need to see the doctor, and that could be a problem." (Provider Interview) Motivation through contact with friends and family was echoed throughout the data. As two participants illustrated: P1: "I find, myself, when I do things with someone [then] I really stick to it, but if it's just me myself I just lay on the couch." P2: "That's why I was saying [about] the app-this app would be good because everyone who's in the group and others who come along-we could connect through the app." (Forum 1)

Monitoring Each Other Through the App-Checking Blood Sugar
Participants described how an app could be used to check on others. As one woman stated: "That's how we know who's cheating, and who's not checking their sugars and things like that, in keeping a record, you can see when I did my last finger stick, so I know what it was. So I know if that I know at 12 o'clock your last finger stick was 80 I'm gonna send you a text like 'Look girl, I know your last finger stick was 80, how you doing? Is everything okay? . . . This app, it can go so many places.'" Another participant described challenges with finger sticks and suggested how a webcam might be used to connect people with T2DM to each other. For example, the app could be used to virtually share experiences: "So maybe you have this device where I can come on my phone and I go, 'You ready to do your finger stick? Okay, let's do them together.' She has her phone. I have mine. We got the camera. 1, 2, 3...stick. Now we got our numbers. I know she did it. She knows I did it because we seen each other doing it." (Pre-Usability Forum)

Monitoring Each Other Through the App-Taking Medication
Much discussion in the forums and in-depth interviews focused around use of the app as a way to monitor each other's T2DM self-care in terms of medication. One idea was that it could be used to coordinate reminders among network members.
It would be nice if it [my brother's phone] were shared with my phone so that even if he gets a reminder on his smartphone it also sends a reminder to my phone, so if he doesn't do what he needs to do then I can call him and check on him that 'Hey, I just got a reminder that you're supposed to take your medicine, did you do that?' (Forum 4) Some participants suggested that they could use the app to get support for equipment needed to aid in diabetes care.
"Let's say like if I ran out of test strips and the kind that I use I can't buy over the counter, although you can, but something of that nature. I'm in an emergency, I am having an emergency. What if there's some type of app or within your app that I could reach out in my community and say, 'Hey I'm out of testing strips, I'm out of insulin,' or whatever. Is there anyone that I can go to for like a borrow? Not personally but like are there resources-like are there pharmacies that will do something?" (Pre-Usabilty Forum) Others expanded on potential use of a buddy system to help T2DM patients and their contacts keep track of medications.
"Seeing as is you take your medicine every day-how you gonna keep track? With the app, instead of keeping track on a notebook or something, you can keep track on the app so as though your family members, or whoever you have as an emergency contact, you know can see you is taking your medicine and as I said you can get a reminder sent to your smartphone, like 'oh they haven't responded to the reminder but you see online that they have tracked in that they took their medicine and did everything they're supposed to do, they just didn't respond to the reminder.'" (Forum 4)

Sharing Information about T2DM with Others Through the App
Participants also suggested sharing information through the app. As one woman suggested "Postings of foods that are low in fat. I screenshot that-because I think that's important. It's more of, like, a sharing." Others suggested taking advantage of social networks through the app to "send recipes back and forth," along with posting information about activities, community groups and resources, to increase physical activity. One woman described: "With the app would it be possible so like you could drop like a pin if you went to something helpful. So instead of just having the calendar and the addresses in-it might make it more easier." (Pre-Usability Forum) Others suggested posts could include diabetes-friendly dishes that they have prepared.
"The group can discuss, 'Well this is what I cooked for dinner, what I put together from this,' and that would be a good idea." (Provider Interview) Healthcare providers however cautioned about the possibility for misinformation: "I think that's great both for self-efficacy and I think that's a huge way to increase motivation, but I think I would just be cautious about that with the potential for misinformation. I don't know how that's monitored. I think it's okay to share recipes, but from what I understand, here is such a knowledge deficit of what is truly healthy." (Provider Interview) However, participants also discussed ways to circumvent misinformation.
"Maybe you could have a toll free number on the bottom" in order to "call someone who you know who actually has experience." (Pre-Usability Forum)

Using the App to Help Each Other in Case of Emergency
Many themes from discussions in the forums and interviews revealed how support could be mobilized in an emergency. Another popular potential use of the app included developing a system of automatic alerts to be sent to emergency contacts.
"You can keep in the application . . . someone who is very close to you's contact number like email or phone number and we can send a message to them . . . and in emergencies we can send a message." (Pre-Usability Forum) In the context of worrying about someone to help in a crisis, one woman suggested: In developing the app, like you're talking about, each and every one of us in this community in this forum and we could exchange information so that we know 'Okay well I haven't heard from Babette she usually online at 7 am she gets off at 11 am to watch The Price is Right, she checks back in at 2, so if I haven't heard from Babette at 2 o'clock its now 5 o'clock, so that mean I'm going to hit everybody in the forum to ask them, have you guys spoken from Babette? 'No I haven't talked to her, she hasn't been online,' then my next step is to get to a phone, call her son, call her fiancé, and say 'Look-is Babette okay?' 'Yeah she's fine, she just didn't feel good after lunch.' So I can get back online to the community or forum and say I talked to Babette's family. She okay. She just wasn't feeling well after lunch." (Forum 1) Other variants on the idea of having an alert on the phone to signal an emergency were proposed.
"Is there something that could be on your phone that you haven't checked in or checked your messages so anything within 3-4 h that would set an alert to you or a sister?" (Forum 4) There was general agreement that such an alert could go to a caregiver like a sister, grandmother, or another supportive person who could get to the person with T2DM quickly if necessary. Expanding on this point, another woman suggested "You wouldn't have to say anything. It [could be] just a button to know if someone could come to you, just to let you know he's woke and aware." (Forum 4)

Reinforcing Each Other's Accomplishments through the App
Another theme included using the app as a way to showcase one's accomplishments and motivate others. One man explained: "Like we was saying with the calendar, if your friends or family can see you've been doing good, they can give you encouragement on the calendar. If they see you've been taking your medicine on time lately-congratulations!" (Forum 4) Other providers agreed, adding: "It becomes motivation, when someone tells you my A1c used to be 9.6 and I changed to do something, and do you believe it now? I'm on the 7, or 'I used to take two or three medications but I started 30 min walking-just a simple 30 min walk in the evening and now I'm taking one or two. That can be a sort of motivation that they might listen to." (Provider Interview) Others suggested sharing information about diabetes management, such as glucose values, with other app users, as a way to encourage others.
"Like I said, I was 210. From 210 to 153 . . . and you can put that down-that that's what you accomplished, so everyone in your friends, your circle, know it's an accomplishment." (Forum 1)

Discussion
To our knowledge, this is the first study to investigate which features of an app for T2DM self-management are prioritized within a low-income African American community with a high disease burden. We found that people with T2DM and their friends and family perceived that a social networking app would be valuable for T2DM management. They identified various functions of a potential mHealth app that could facilitate diabetes self-care, and those functions tapped into multiple different forms of social support. Themes were cross-cutting, and no differences were found between persons with pre-diabetes/T2DM and health providers.
Instrumental support emerged as the most common type of support that such an app could facilitate. Examples included finding someone to exercise with, to get reminders from, to help with transportation, to aid in finding equipment for diabetes care, or to check on one another. It is possible that receipt of instrumental support may work in tandem with emotional support (e.g., discussions about T2DM management might be facilitated while getting a ride from someone to a doctor's appointment). That is, dual benefits may result from interacting with others about T2DM. Although less frequent, the app was discussed as a potential means for receiving informational support e.g., through sharing information about neighborhood resources or diabetes-friendly recipes. Ways that the app could enhance appraisal support were uncommon, though in a few forums, it was discussed how a T2DM app could enable feedback from others to enhance self-management.
Data suggested that tapping into social networks was perceived as critical for diabetes self-care. This idea reinforces research that suggests ways in which a patient's social networks can support self-management, including through the sharing of illness narratives [29], direct influence on behaviors [30], help with diet and medication reminders, providing transportation [31], and increasing exercise and foot examinations [32]. Notably, although mHealth interventions have been proliferating, a recent systematic review found that app functionality is often limited to data upload, education, and text message notifications [33].
Some participant suggestions could be considered problematic with regard to translating them into a smartphone app design. One example is the idea of creating an exchange mechanism to share diabetes supplies (e.g., insulin or testing strips). Sharing diabetes-related resources is in conflict with published guidelines of both the Centers for Disease Control and the Food and Drug Administration [34,35], which prohibit sharing of blood glucose monitoring supplies/devices in order to prevent the transmission of infection and blood-borne pathogens [34,36]. Furthermore, types of medications and doses are personal and prescribed by a physician (e.g., using the same dose, but another type of insulin could be dangerous). Another suggestion raising ethical issues was the request for an alert mechanism on the app to notify family/friends of an emergency. Ethical and legal barriers to developing such app functions exist, if the emergency function failed to operate as intended. Therefore, such features should not be recommended for implementation.
One limitation was the inability to distinguish between people with pre-diabetes/T2DM and their friends and family members in our transcripts. A major study strength was the use of information solicited directly from people with pre-diabetes/T2DM and their first order networks. Complementing information from these participants with that of providers allowed us to triangulate findings and look for consistencies and inconsistencies across these groups. Within a high risk demographic area, capturing a sub-population disproportionately affected by T2DM and with limited socio-economic resources, we were able to gather specific information about useful features of a social-network-focused T2DM app, helping to bridge a gap between research and practice [37]. While our findings are relevant to this population, these findings may not be transferable to other socio-economic/cultural groups or rural settings. Although more men than women are affected by T2DM, our participants were predominantly women. It is unknown if our results would apply to men.

Conclusions
In summary, we found that a social network app for T2DM was considered a useful tool for a population of low-income African Americans, a high risk group for disparities in diabetes outcomes and care. Given that few diabetes apps have been evaluated in the peer review literature [21,[38][39][40], future studies are needed to implement these ideas and to test their usage.