We applied a co-design framework (Figure 1 [41,42]), which combines human-centered design [43] and engineering design [44,45] processes, to create the specifications for a regional, facilitated, and web-based peer support network. The framework includes 4 stages: defining the problem, understanding the context for use, developing and building consensus around functions and specifications that fulfill identified needs, and establishing and pilot-testing design specifications.

The study was approved by the Dartmouth College institutional review board (#2000907).

A waiver of written informed consent was used for the surveys, as the only link to the survey respondent would have been the written informed consent document. No identifiable information was collected and individuals were not paid for participating in human subjects research.

The target population, hereafter referred to as end users, was defined as care partners (ie, informal caregivers or family members) supporting or providing care to adults (aged 18 years or older) with a serious illness and those who have experienced the loss of someone to a serious illness. The term care partner was chosen by the co-design team to reflect their role and relationship in partnering with a person with a serious illness. Serious illness has been defined as one that carries a high risk of mortality and either negatively impacts a person’s daily function or quality of life, or excessively strains their care partners [46]. Care partners include relatives, spouses or partners, friends, neighbors, or others who have a significant personal relationship with, and who provide a broad range of assistance to, a person with a serious illness. We focused on care partners living in New Hampshire and Vermont, the catchment area for Dartmouth-Hitchcock Medical Center.

We formed an 18-member team to co-design the network. The team included 2 active and bereaved care partners of people with serious illness, 4 adults with serious illness, 6 interdisciplinary palliative care clinical team members, and 6 support service staff. Care partners and people with serious illnesses were recruited by our clinical team partners. Clinical team members were affiliated with Dartmouth-Hitchcock Medical Center, a rural tertiary care academic medical center in New Hampshire. Facilitation of the co-design process was led by researchers with expertise in co-design, evaluation, and quality improvement (EAO and ADVC). The design process was informed by consultation with an expert in human-centered design (EK) and a systems engineer (ISK). To ensure our design aligned with best practices, we consulted with external advisors with expertise in facilitated support networks (DG and CY), met with regional health care and support partners, and obtained input from an external advisory committee with expertise in scaling innovations, business, and serious illness.

The co-design team met twice a month for 8 months (April to November 2019) to identify and prioritize the functions and specifications of the network and met monthly for 6 months (December 2019 to May 2020) to test prototypes.

We conducted human-centered design exercises [43] to elicit community needs and assets, define the problem, and understand the context for use. We drew upon stories of serious illnesses shared by care partners to identify needs arising from lived experiences. Design facilitators shadowed [43] outpatient palliative care visits and attended interdisciplinary palliative care team meetings to further understand the context of use; services provided by the care team; and the daily lives of people with serious illness, their care partners, and clinicians [47]. We developed empathy maps [43] to reflect and articulate what end users hear, see, say, do, think, and feel, and to identify points of pain and gain. We used a visual thinking exercise [48] to sketch ideas for an ideal support network (example, Figure 2). Design exercises were reviewed during design sessions to discuss critical functions and important features that fulfill and support these functions.

We supplemented design activities with surveys of potential end users. Between December 2018 and April 2019, we collected 28 surveys from a convenience sample of active care partners presenting at the Dartmouth-Hitchcock Medical Center outpatient palliative care clinic and 21 surveys from a convenience sample of bereaved care partners affiliated with the clinic or regional health organizations (eg, hospice) to elicit information on the challenges, needs, and desire for peer connection among active and bereaved care partners (surveys are provided in Multimedia Appendix 1). We used descriptive statistics to summarize categorical data, and thematic analyses to identify themes from open-ended questions.

We collated, organized, and systematically described the identified needs into a hierarchical list. We converted the highest-level needs into functions for the network to achieve. At the highest level, for example, “care partner need for information” was converted to “connecting care partners to information” and “care partner need for emotional support” was converted to “facilitating emotional support.” We identified potential specifications associated with each function of the network.

Co-design team members independently rated the importance of each specification using a five-item prioritization scale: (5) must have, (4) high importance (feasible without), (3) should have (very important), (2) could have (consideration), and (1) desirable (will not have at this time) [49]. Team members were provided an opportunity to describe their understanding and thought process for preference identification as a group. Scores were weighted and averaged by respondent type to ensure that patients and care partners, clinicians, and support service staff had equal participation weight (eg, scores from 6 clinicians were averaged to create 1 clinician average).

We tested a series of prototypes to identify the importance of different network components. These included storytelling exercises, face-to-face group conversations, one-to-one matching of care partners, group videoconferencing, an online discussion forum, and a “Caregiver Day” event at the health center. Summaries are provided in Multimedia Appendix 2.

After finalizing design functions and specifications, we identified potential vendors to host the network by conducting an environmental scan of web-based networks focused on people with serious illnesses and care partners and mapping desired design specifications to vendor capabilities. Four vendors provided demonstrations of their platforms between April and May 2020. Vendor selection was driven by the vendor’s ability to provide the prioritized functions and specifications, the cost to build and maintain the network, and being a US-based company. Following vendor selection, the web-based network platform was customized by the design team to deliver upon design functions and specifications.

User acceptance testing was conducted between April and July 2021. Four care partners and 2 people who previously had a serious illness were invited to register as founding members of the network in April. These founding members were encouraged to invite care partners they knew into the web-based community to test the feasibility of enrolling members. Three clinical champions (physicians and social workers), a chaplain, and a staff member who manages complementary care programs referred care partners to the network. The research team met weekly with the vendor during the user acceptance testing period to resolve issues. The web-based platform remained available for registered members to use while issues were addressed. The network moderator met monthly with the design team to plan future improvements using a quality improvement framework [50].

The co-design process led to clarity around the objective of the network: to help care partners cope with the surprises that arise during serious illness and bereavement. The network, named “ConnectShareCare,” was intended to supplement existing services, to be provided outside of clinical encounters with the health care system or regional professional support and service organizations, and to tap into the wisdom of those with lived experiences.

If successful, the design team anticipated that the network would benefit 4 groups, as outlined in Table 1.

Our co-design process identified that active and bereaved care partners have different needs but have common interests in sharing information and providing or receiving support. Active care partners who completed an assessment survey were most challenged by emotional difficulties (eg, worry, uncertainty, or lack of control; 12/28, 43%), providing care and emotional support (7/28, 25%), and practical matters (6/28, 21%). Bereaved care partners were most challenged by loneliness (10/21, 48%), managing grief and emotional difficulties (6/21, 29%), and managing practical matters (5/21, 24%). Active care partners were most helped by support from friends, family, or other social connections (12/28, 43%), as well as by medical professionals (9/28, 32%), while bereaved care partners were most helped by support from friends, family, or other social connections (16/21, 76%) and by developing self-care strategies that led to personal resilience and growth (12/21, 57%).

Most active (18/27, 67%) and bereaved (18/21, 86%) care partners were interested in 1 or more forms of connecting with other people who have shared a similar care experience. Both active and bereaved care partners anticipated that a network could provide support, knowledge, and resources but anticipated challenges associated with time to participate and with forming personal connections.

The series of human-centered design exercises and interviews led to additional insights. First, active and bereaved care partners may benefit from connecting and sharing information with each other. Peer-generated information from care partners who have shared a similar experience feels more authentic, detailed, and actionable. Second, care partners wished to belong to a local support community that was connected through geographic proximity and could provide recommendations for local resources. Third, a web-based network enables care partners to access information at any place or time, allows anonymity, improves access for people who are home-bound or grieving, and may reach an increased number of care partners. Fourth, trained staff who can moderate, promote, and manage the web-based community and volunteers who can recruit and engage users are important. Paid or volunteer moderators can play an important role in listening, making connections, and highlighting information, services, and programs. Fifth, a web-based network would benefit from supplemental opportunities for the community to meet face-to-face or through digital programming.

The design process also identified several potential risks and possible mitigation strategies. First, there was a risk of causing harm to vulnerable end users if the design failed to provide a safe and supportive environment, protect the privacy of sensitive information, or enact acceptable data ownership guidelines. Second, there was a risk associated with the usability of the network among end users who were less facile with web-based services. Third, there was a risk associated with the inability to form a personal connection with peers through a web-based network. Other potential risks included those associated with competition from other networks, inaccurate content or information, and care partners having minimal time to participate in a web-based support network due to other responsibilities.

Two primary functions emerged from the design activities: to support care partners in (1) providing each other with emotional support and (2) exchanging helpful information and resources. We developed specifications related to these functions (Table 2), as well as the form of the network, including the user interface, data and security, and other considerations (Table 3).

The most highly prioritized specifications to support each function (Table 2) included providing a supportive and respectful space; providing connections to trusted and curated local, national, and international resources; reducing the difficulty of asking for help by normalizing needing help; and providing curated resources to address the most common concerns (eg, easy access to frequently asked questions and answers).

The user interface (Table 3) must be simple and intuitive, provide technical support for users, and be easy to access. It was highly important for the user interface to be aesthetically refined, include appropriate use of interactive elements, and not allow external advertising. The platform must be secure and protect user privacy. It must be available as a public forum for information but also allow participants to communicate via discussions not visible to others outside of the network. Other highly important considerations included hosting by a vendor with experience in providing health-focused networks and providing features that support sustainability and scalability (such as member incentives and the ability to customize or add functionality over time).

Following the vendor selection process, we worked with CareHubs, an online health network vendor, to build the ConnectShareCare network. The network included (1) a single support community for active and bereaved care partners; (2) a short list of curated resources based on needs identified during the design (planning ahead, practical issues, emotional issues, communication issues, and family resources); (3) a calendar of online and in-person regional support programming; (4) a story from an end user about their experience as a care partner; (5) a roster with member profiles; (6) a help center; and (7) community guidelines. Screenshots are included in Multimedia Appendix 3.

The initial feasibility of the network was demonstrated through the active participation of founding members, beginning in April 2021, with expansion to include 12 new care partner members in May 2021 and 15 new members in June 2021. During this period, the network had an average of 135 posts per month. A total of 16 members posted a message to the network, with an average of 11 members posting per month.

A co-design process generated a useful and feasible regional, facilitated, and web-based peer support network for care partners of people with serious illnesses. The co-design process ensured that all voices were heard, especially among people who typically may not work together. Design decisions were made collectively and systemically, which allowed network functions and specifications to be identified and prioritized. By doing so, the co-design process ensured that the most critical decisions were responsive to regional needs and preferences. The resulting network connects active and bereaved care partners with peers to facilitate emotional support and exchange information related to caregiving for people with serious illness.

Similar to other networks [15,33,51-62], ConnectShareCare has a clear purpose, includes mechanisms to foster a strong sense of community and support among regional care partners, and provides value to a variety of groups. The network builds on the resources, wisdom, and experiences of care partners. The inclusion of a moderator helps ensure a safe environment that is protected from misinformation, trolling, or cyberbullying. The moderator sets the tone and etiquette with members, modeling behavior and other preventative measures, as well as moderating posts, facilitating connections, and providing feedback to adjust member behaviors [20,23].

Our design process had several strengths. First, our process engaged people who would be end users of the network in making critical design decisions [40]. In contrast to asking end users about single decisions, our process allowed end users to make decisions in the context of all other design requirements and options. This led to a more systemic approach to engagement that ensures that decisions are optimized to fit together. Second, our process brought together people with different expertise (in being a person with serious illness or a care partner, in medicine, in health network design, and in community management and moderation) that may typically not work together to create services, and each had different needs to maintain the value of the network. The process also engaged people with engineering design [44,45] and user-centered design [43] expertise to ensure that the process was rigorous enough to produce a network tailored to the needs of end users. Third, our design was responsive to regional needs by addressing gaps in available services and drawing upon local assets. We intentionally worked with multiple health care and support organizations in the region to broaden our network and reach care partners most in need of support, regardless of where formal health care services were received. Care partners are a vulnerable population who often receive minimal structured support from the health system, yet they have significant knowledge to contribute on how to navigate health care systems, health and social resources, and losses at every stage (eg, relationships, identity, and freedom). This knowledge is often actionable by peers. Finally, our process may have supported growth in network participation due to health care and regional support partners feeling heard and included in the design process.

This project has certain limitations that should be considered by those wishing to adapt the methods and findings to different contexts. First, the project was built around a recognition that people who live in north-east rural areas can be particularly isolated and lack access to sources of support. It is unclear whether the network will meet the needs of people in other regions of the country. Second, similar to local demographic characteristics, our design team had limited racial or ethnic diversity. We do not know how greater racial, ethnic, or sociodemographic diversity would enhance or create barriers to its success. Third, the network is built around an asynchronous model. This can be very important because it respects the different schedules that people are on; however, it limits the opportunities for people to hear and see each other in real time. It is unknown whether a network that combines synchronous and asynchronous components would be useful in our context. Finally, while the design process requires extensive back and forth among participants and may not be feasible in other situations, it also represents a strength in creating a network that more closely reflects community needs. In our situation, the decision to create a new network was a result of the recognition that other solutions are not likely to fulfill the needs of our end users.

Care partners of people with serious illness often lack support and are likely to experience significant challenges and unmet needs. We followed a structured co-design process to collaboratively identify and prioritize the functions and specifications of a regional web-based facilitated peer support network to help care partners cope with the surprises that arrive during serious illness and bereavement. The network was designed to provide emotional support and exchange information related to serious illness caregiving. The coproduction of accessible peer-led information, resources, and support may extend the scope of services offered by a health system to support lay care partners—becoming part of a sustainable, person-centered value-creation system [63,64]. Opportunities exist to evaluate the feasibility of actively engaging community members and moderators in the network [65,66] and will be reported on in a publication under development. Moreover, there is a need to understand effective mechanisms to recruit and retain participants and provide a safe environment to people who are in vulnerable situations; to monitor the network life cycle through metrics related to activity and growth [20,22]; and to consider the creation of network subgroups to support care partners of people with particular illnesses (eg, cancer, dementia, and Parkinson disease) or people from specific minority populations. Finally, there is an opportunity to understand the impact of a regional support network on care partner quality of life, self-confidence, loneliness, and isolation [67,68]; and on health system reputation, use, and visibility.