Peer Mentoring in the Do-it-Yourself Artificial Pancreas System Community

Abstract

Background:

The do-it-yourself (DIY) artificial pancreas system (APS) community is led by people with diabetes (PWD) and their families. In contrast to standard models of care, in which health care professionals are responsible for the education and training of PWD and their families, within the DIY APS, community users are reliant on their peers for education and training. The provision of this support has not previously been studied. This article focuses on the role of mentors within the DIY APS community.

Method:

Semistructured interviews were conducted with mentors (n = 9) who were identified with assistance from the DIY APS community. Thematic analysis was undertaken. Results were presented to select participants who endorsed the results as an accurate description of mentoring in the DIY APS community.

Results:

Mentors’ primary motivation was altruism. An empathetic, process-focused approach was modeled by prominent mentors. The use of online forums as the primary source of user support was perceived to be an effective method for sharing the workload. Key challenges for mentors included the frequency with which users asked questions that were answered in the existing documentation, dealing with conflict and managing workload. Most participants did not perceive mentoring to be risky, but did take steps to ensure that users were taking responsibility for their own systems.

Conclusion:

This study is the first to examine the role of mentors within the DIY APS community and also within the broader diabetes online community. Future research might seek to quantify the empathetic, process-focused approach to mentoring.

Introduction

The development of do-it-yourself artificial pancreas systems (DIY APS) have disrupted the diabetes technology landscape. ¹ These unregulated technologies, designed by people with diabetes (PWD) and their families, present clinicians with challenges.^1,2 Equally, DIY APS have revealed novel pathways for innovation ² and offer unprecedented opportunities for collaboration between traditional researchers and the DIY APS community.^1,2 Do-it-yourself artificial pancreas systems, like other hybrid closed loop systems, require users to learn how to effectively operate the system and adapt self-care practices to fit with this new treatment paradigm.^3,4 User support and education has been undertaken within the DIY APS community outside of clinical settings. Thus, collaborating with the DIY APS community offers the opportunity to study a model of support for PWD that differs significantly to standard care.

People living with type 1 diabetes (T1D) and their families need to learn how to use a range of diabetes technologies and self-care practices.^5,6 In standard care, patients’ learning is primarily supported through infrequent one to one sessions with clinicians. ⁷ Support for users in DIY APS operates on a peer-to-peer basis, primarily through online platforms such as Facebook groups, Gitter and Zulip (communication platforms for open source software projects), Twitter, Google groups, and Slack channels. Medical literature distinguishes between different types of peer-to-peer support. Mentors are defined by having successfully dealt with a specific challenge or issue. ⁸ Advisors have been given training by health care professionals to deliver specific education. ⁹ Peer support is offered by a person with experience in the same condition as the person or people they are supporting. ¹⁰ This article focuses specifically on mentoring: DIY APS users who have successfully set up DIY APS and who undertake user support on a regular basis.

Do-it-yourself artificial pancreas systems integrate commercially produced insulin pumps and continuous glucose monitors (CGM) with third party hardware and open source, community-developed software to create a hybrid-closed loop automated insulin delivery system. ¹¹ Do-it-yourself artificial pancreas systems have three main platforms: OpenAPS, which runs an algorithm on a linux system ¹² ; AndroidAPS, which implements the OpenAPS algorithm via an android app ¹³ ; and, Loop, a separately developed algorithm housed within an IOS app. ¹⁴

Do-it-yourself artificial pancreas systems share the #WeAreNotWaiting banner with Nightscout, ¹⁵ open-source, cloud-based software for sharing continuous glucose monitor data. Information seeking was common among users of Nightscout’s Facebook group, CGM in the Cloud. ¹⁶ Approximately one-third of users provided technical support to others, while half the users shared information about diabetes and technology. Subsequent research identified that the provision of technical support was not limited to expert users and that the desire to help others, or “pay it forward,” was widespread. ¹⁷ The desire to pay it forward has also been identified among new users of DIY APS. ¹⁸ As yet, however, the provision of user support and, in particular, the mentoring role of capable users within the #WeAreNotWaiting community has not been explored. The aim of this article is to explore mentors’ perspectives on the implementation of mentoring and the challenges of mentoring within the DIY APS community.

Method

This exploratory study was shaped by a collaborative approach with the #WeAreNotWaiting community. ¹⁹ The project was negotiated with key members of the #WeAreNotWaiting community, who contributed to the focus of the study and identification of participants. Ethical approval was granted by the researcher’s institution. Given that mentoring has not yet been examined within the #WeAreNotWaiting movement or other organic diabetes online communities (DOC), an exploratory approach based on semistructured interviews with a representative sample of mentors was used.

The author (HC) holds multiple roles which are relevant to this project. He is a person with diabetes and an active DIY APS community member. HC conducted research with the DIY APS community prior to becoming a DIY APS user. ¹⁸ In keeping with recommended standards for reporting qualitative research, ²⁰ HC includes explanation of the impact of being an insider-researcher as part of the method section. ²¹

In-depth, semistructured interviews were conducted via digital video call with nine mentors (four female and five male). Informed consent was gathered prior to undertaking interviews. A tenth mentor was invited to participate but declined due to work commitments. Five of the mentors had a family member with T1D. Four of the mentors were adults with T1D. Six of the mentors were “superusers,” the community’s terminology for individuals able to solve virtually any DIY APS problem. The three other mentors had skills specific to a particular range of problems. Three of the participants contributed to algorithm development and a fourth was in regular contact with developers. Six mentors had prior expertise in IT or algorithm design. One mentor had previously mentored users in the CGM in the Cloud Facebook group. Two of the mentors had been involved since the first algorithm, OpenAPS, was published in early 2015. Another pair of mentors had been involved since the founding of the Facebook DIY APS support group, Looped, in 2016. The remaining participants had become involved in DIY APS and subsequently mentoring during 2016 and 2017.

HC’s experiential knowledge of the community enhanced trust and rapport with participants and facilitated in-depth discussions with mentors about their roles and experiences. ²¹ Interviews were digitally recorded with permission of participants, then transcribed verbatim. Interviews ranged from 45 minutes to 1 hour 45 minutes. After being checked for accuracy, transcripts were returned to participants for approval. Thematic analysis was undertaken on an ongoing basis by HC, who is an experienced qualitative researcher. The core themes were established within the first six interviews. The remaining three interviews reinforced these themes but no new themes emerged. Throughout the project, HC followed Berger’s recommendations—maintaining a research log, regular reviews of participant data, and peer consultation—for ensuring that his insider perspective did not obscure participants’ views. ²² The rigor of the findings was further enhanced by member-checking themes with a subset of participants. ²³ These participants endorsed the overall results and provided feedback leading to minor clarification of phrasing. Following a collaborative research approach, and in keeping with ethical approval granted by the researcher’s institution, participants could choose to be recognized for their contribution or to retain their anonymity. Those who wished to be recognized are thanked in the acknowledgments section. Participants were assigned codes and findings written in gender neutral language to preserve the anonymity of those who wished to remain anonymous.

Results

Reasons for Mentoring

Altruism was the key reason that participants shared for mentoring. More specifically, mentors wished to share the benefits that they had experienced through DIY APS with others. M7 explained, “everybody should have the ability to get what I have to make their life better.” Additionally, two newer mentors took on mentoring roles in order to reduce the burden carried by the established mentors.

Implementation of Mentoring

Organization of mentoring

The mentors involved in the launch of DIY APS initially offered “very minimal” support (M1). As evidence about the efficacy of the algorithm grew within the community, perceptions of risk reduced and they began to consider how to support other users. These mentors identified two priorities: process-focused mentoring and distribution of user support across the community. M2 explained, “we want you to have a community to go to and understand that this is not just [name of mentor] and [name of mentor]. There’s a whole bunch of people and any one of them can help.” Rather than being formally organized, mentoring operated informally. M1 explained that giving users kudos for helping others with troubleshooting had supported the organic growth of mentoring in the community. As the majority of mentoring took place on public forums, the mentoring process was clearly modeled and newer mentors had learnt through observation. M8 explained that their approach was based on “what [name of mentor] and [name of mentor] normally do.”

Participants noted that many DIY APS users provided user-support. Most problems did not require extensive expertise to solve. Further, M2 noted that superusers could encourage others to offer assistance: “If we step back and don’t engage on certain things so fast, then somebody else will be more likely to jump in and help.” Equally, the three “medium” level mentors explained that part of their self-chosen role was akin to triage. They would answer simple questions allowing superusers to focus on more difficult questions. They also helped users with complex problems to provide adequate background information, allowing superusers to step in and immediately focus on identifying and solving the problem.

Process-focused, empathetic mentoring

Participants emphasized that the aim of mentoring was to build the skills of DIY APS users. M1 explained:

It really became a focus early on, of let’s teach people how to fish: how to read the documentation, how to ask questions, preferably how to ask for troubleshooting [help] in a way that they present the information, that you don’t have to coach them through doing all the trouble shooting.

The problem-solving process would typically involve helping the user to identify the problem, explaining the underlying mechanisms that were leading to the error (usually via links to the existing documentation), and working with the user to improve their system. This process aimed to assist users to solve their current problem while building users’ broader problem-solving skills.

Mentors took an empathetic approach. They recognized that users might feel anxious posting to online support channels. The Looped Facebook group was founded to offer a supportive forum to new users lacking confidence to use unfamiliar platforms, such as Gitter, to ask for help. Looped had been established with the same philosophy as the CGM in the Cloud Facebook group. M4 summarized these values as, “everyone needs to be nice to each other, come from a place of respect and everyone starts somewhere. . .. we all have something to offer and that all we ask is that you pay it forward.” Mentors avoided judging other aspects of users’ self-care. M5 explained “I don’t judge how someone wants to define their comfort zone for A1c.” Mentors noted that they did occasionally remind the community to be empathetic when interacting with others.

Mentoring channels

Most mentoring took place via online forums. However, face-to-face “build parties” were also run. Face-to-face sessions allowed mentors to group new users together according to their needs. Some mentors suggested that face-to-face meetings suited a specific cohort of DIY APS users. Additionally, M5 explained that running face-to-face meetings was “eye opening for me” revealing, “things that I forget that I take for granted.”

There were differences between mentors about conducting mentoring across public and private channels. Superusers were regularly contacted via private channels with requests for assistance. Three superusers suggested that this risked the user becoming reliant on a single person. These mentors preferred all mentoring to take place on public forums, allowing anyone with relevant knowledge to help. Working through solutions in public forums served as a resource for users independently solving problems. Most other mentors engaged in a mixture of public and private mentoring.

Pro-active support

Three of the mentors interviewed were primary creators of documentation for AndroidAPS, Loop, and OpenAPS. For these participants, improving documentation was a pro-active form of mentoring as addressing recurring issues via documentation reduced mentors’ workload. Additionally, three mentors maintained blogs and one produced Youtube videos for users. Although blogging and videos were useful ways to support other users, mentors acknowledged that these were less likely to be updated than formal documentation.

Challenges of Mentoring

Read the documentation

The majority of requests for help related to issues that were answered in the existing documentation. M9 suggested, “probably sixty to seventy percent of questions in various help groups could be answered in five minutes of searching the instructions for OpenAPS or Loop.” M1 estimated spending “probably eighty percent of my time” referring users to specific sections of the documentation. While mentors rationalized the frequency of such questions, most nevertheless found it frustrating. M3, for example, explained their three key pieces of advice for new users “were (1) read the docs, (2) please read the docs, (3) please, for the love of god, read the flippin [sic] docs.”

Dealing with conflict

Three mentors identified dealing with people “who are just willfully ignorant” (M1), as a rare but frustrating challenge. Such people typically offered incorrect answers to other users’ questions, then refused to accept that their answer was incorrect regardless of the evidence supplied by a mentor. These people were usually male. Similarly, some mentors noted antagonism from users who did not understand why DIY APS was DIY and would expect support equivalent to that of a commercial product. Mentors attributed these challenges to the rapid growth of the community.

Managing workload

Mentors’ workload varied according to their specific role. Workload increased when new iterations of software were released or when publicity led to surges in new members. Mentors deployed a range of strategies to retain balance in their life and avoid burnout. These included taking time away from specific roles in the community, turning off notifications from forums, not responding immediately to requests, and devoting time to other interests. M1 noted that multiple volunteers had cycled in and out of active mentoring and development roles since 2015. M6 had experienced a significant increase in workload that was impacting the enjoyment they had previously gained from the role.

Risk

At the time of interviews, only one participant expressed concern at the risk of liability should a person they helped subsequently experience a negative outcome relating to DIY APS. Most mentors were unsure whether they might be legally culpable. Instead, they observed that risk was inherent to diabetes and claimed that DIY APS was the safer than standard care. They suggested that the greatest risks specific to DIY APS were incorrect user-generated settings. Thus, participants argued helping users to understand these settings reduced the risk to users. M9, for example, explained “it’s pretty hard to go wrong with Loop and OpenAPS, except for [sic] through user error of entering something wrong or not understanding how they work properly.” Mentors also suggested that they ensured that the user was responsible for implementing changes. Approximately half of the mentors avoided making treatment recommendations and employed phrases such as “when I experienced this problem, I found this solution worked for me.”

Discussion

Mentors’ altruistic motivation closely aligns with motivation of CGM in the Cloud users.^16,17 Similarly, Litchman et al’s scoping review of the diabetes online community (DOC) suggests that a desire to help others is present in both adolescent and adult users. ²⁴ Actions of mentors in offering empathetic support, creating resources for users, and creating welcoming forums are community building efforts that have parallels to the wider DOC. ²⁵ The policing role that mentors occasionally performed is similar to that of moderators in other DOC forums. ²⁶

Litchman et al note that multiple studies have identified informal mentoring within the DOC. ²⁴ However, the experiences of those taking mentoring roles within DOC have not yet been examined. Mentoring in the DIY APS community follows the approach common within open source software projects. ²⁷ Both the pay it forward ethos of the #WeAreNotWaiting community^17,18 and superusers’ recognition of the need to allow a wide range of users to answer questions, encouraged the burden of user support to be widely shared. Although user support is widely distributed, mentors play a crucial role through modeling a process-focused, empathetic approach, and stepping in when necessary to reinforce this approach or to correct mistaken advice. The distribution of user support may have been an important factor in the growth of the DIY APS community. However, the sustainability of organically formed mentoring and user support should DIY APS continue to grow is unclear.

Mentors’ empathetic, nonjudgmental approach matched the ethos of #WeAreNotWaiting and recognized that there was a steep learning curve, particularly for nontechnical users. This approach, which avoided making judgments about users’ diabetes management decisions, has similarities to increasing recognition of the need for clinicians to recognize that the language used in patient-clinician interactions has a genuine impact on patients.^28,29

Within the existing models of care, peer support has been conceptualized as a supplement to clinical support from health care professionals (HCP).^5,30 In contrast, the contribution of DIY APS mentors arguably surplants aspects of HCP support. The unregulated nature of DIY APS and the lack of training materials for HCP contribute to this. ² Some HCPs may perceive risks to their patients from this loss of oversight. ³¹ Equally, this shift in the balance of care brought about by DIY APS may offer insights to redeveloping models of care. The current one to one model of care is resource intensive and reliant on knowledge transfer from clinicians to PWD during irregular face-to-face appointments. ⁷ In contrast, the DIY APS community distributed learning support broadly, using a one to many model, ² with online channels for both documentation and mentoring. This approach, which provided support to users at the time that they were experiencing difficulties fits with contemporary theories of learning, such as connectivism ³² and communities of practice. ³³

Limitations

This study was based on interviews with a small number of mentors, all of whom were English-speaking. Do-it-yourself artificial pancreas system is a global movement, with documentation and support offered in multiple languages. Mentors working in different languages may have different experiences of mentoring. Users’ experiences of mentoring may not match those of mentors. Do-it-yourself artificial pancreas system users are a self-selected subsection of PWD, the applicability of approaches within this community to the broader community of PWD needs to be carefully tested.

Conclusion

This study offers insights into peer mentoring as a specialized form of peer support for PWD. Mentoring with the DIY APS community was not formally organized, but an empathetic, process-focused approach was modeled. Key challenges for mentors included users being unable to find, or not seeking, answers in documentation prior to requesting support, occasional conflict in the community, and workload. Mentors reframed risk to emphasize the reduction of risk achieved for other users through mentoring. Future research might analyze mentoring by monitoring user interactions on one or more of the DIY APS online forums.