Development of a Virtual Transition Program for Adolescents and Young Adults With Type 1 Diabetes: Three-Stage User-Centered Design Study

Introduction

Type 1 diabetes (T1D) is a serious medical condition that affects more than 300,000 children and adolescents in the United States [1]. T1D poses a significant burden on patients and families in terms of health-related quality of life, missed school days, missed work days, and health care costs [2-4]. Previous research suggests that increased independence and a desire for autonomy during adolescence and young adulthood may lead to reduced adherence to medications and other self-care strategies, as well as lapses in communication with health care providers [5-9]. The transition from adolescence to young adulthood has been identified as a particularly vulnerable stage for lapses in care; up to one-third of patients with T1D fail to complete an initial visit with an adult endocrinology provider within 12 months after their final pediatric endocrinology visit [10,11]. For patients with T1D, the transition phase is associated with high rates of hospitalization due to diabetic ketoacidosis (DKA), a life-threatening complication that can be prevented with appropriate management [9,12]. Patients from low-income families are the most vulnerable to these lapses in care and subsequent hospitalizations; intervening in the transition period is therefore essential to prevent the widening of disparities into adulthood [12,13].

Current guidelines from the American Diabetes Association suggest that the transition process from pediatric to adult specialty diabetes care should begin at least 1 to 2 years prior to the actual transfer of care to allow for a gradual increase in adolescents’ self-management responsibilities and continuous education on navigating the logistical aspects of adult health care [14]. Previous T1D transition programs have demonstrated success in improving continuity of care and reducing hospitalizations using full-time transition coordinators, case management, designated young adult clinics, delaying transition until age 25 or 26 years, and shared medical appointments for adolescents and young adults (AYAs) [15-19]. However, more recent studies highlight ongoing gaps in transition preparation, including unmet psychosocial needs and variability in both transition readiness and the effectiveness of existing interventions [20-22]. In addition, transition interventions have not been systematically implemented, even at large academic medical centers, due to limited resources. We sought to design a low-resource, sustainable T1D transition intervention that could be easily implemented and scaled with existing resources within pediatric diabetes clinics. Here, we describe the human-centered design (HCD) process based on the “3I” framework of IDEO used to develop and test the transition intervention, as well as the key components we identified as essential to transition programs for AYAs with T1D.

Methods

HCD and Intervention Development Process

We used an HCD approach based on the “3I” model of IDEO, which includes 3 phases: inspiration, ideation, and implementation [23,24]. Across the 3 phases, we engaged distinct participant groups and data collection approaches, including provider and staff interviews (phase 1), stakeholder customer discovery interviews (phase 2), and intervention delivery with surveys and qualitative interviews among AYAs and mentors (phase 3). We incorporated the principles of HCD throughout, including engaging stakeholders early and throughout the design process, understanding the needs of patients, clinicians, and staff involved in diabetes care, and taking a systems approach to design [25,26]. We enhanced standard HCD steps with rigorous qualitative and customer discovery methods [27], adapted for implementation science [28], to aid in a rigorous, adopter-centered assessment of the “jobs, pains, and gains” that would be addressed by the transition intervention. Collectively, this iterative process followed best practices for designing for dissemination and sustainability and ensuring strong fit-to-context [29].

Phase 1: Inspiration

In the first phase of the project, we gathered perceptions and ideas from providers and staff involved in delivering specialty diabetes care across 4 University of California medical centers. An open-ended question about transitions from pediatric to adult care was added to a semistructured interview guide for an existing qualitative study examining telehealth use for specialty diabetes care. This study used purposive and snowball sampling to identify adult and pediatric endocrinologists, nurses, and staff involved in specialty diabetes care. Interviews lasted 1 hour, were conducted on Zoom, and were recorded and transcribed. We conducted thematic analysis using line-by-line coding in Dedoose (Sociocultural Research Consultants, LLC) [30], as fully described elsewhere [31]. Responses to the transition question were coded and analyzed separately using qualitative content analysis. The research team organized and discussed the main findings for this question.

Phase 2: Ideation

The second design phase was customer discovery, a process widely used across industries to identify intervention adopters’ and influencers’ “jobs, pains, and gains” and to develop value propositions tailored to different types of stakeholders (ie, adopters and influencers) involved in the clinical transition workflow [32-34]. Our process, adapted for implementation science, focused on understanding (1) adopters’ and influencers’ perspectives on the most important jobs to be done regarding transitions for patients with T1D, (2) the “pains” (ie, current challenges in supporting transition) and “gains” (ie, opportunities for improved experience or impact) for each group, and (3) how each group defines success [27,28,35]. We first developed a value proposition hypothesis, then conducted customer discovery to validate this hypothesis, and finally established a final value proposition statement. Value proposition statements articulate how the proposed intervention will relieve pains and create gains while carrying out the jobs to be done. Four groups of adopters and influencers were included in customer discovery: (1) people with T1D who had experienced the transition from pediatric to adult care in the past 10 years, (2) pediatric endocrinologists, (3) adult endocrinologists, and (4) diabetes educators. We recruited participants through a snowball sampling strategy, beginning with the participants from the phase 1 qualitative study. Customer discovery interviews were conducted over the telephone and lasted between 15 and 30 minutes; the interview guide was informed by standard customer discovery questions [27]. Questions for people with diabetes included the following: (1) Can you tell me about your experience transitioning from pediatric to adult care? (2) What types of things were you most concerned about? (3) What were the main challenges and how did you deal with them? (4) Is there anything that could have made the transition easier? (5) What would an ideal transition intervention look like? (6) How would you define a successful transition? Questions for providers and staff included the following: (1) What is your role in transitions of care for patients with T1D? (2) Can you walk me through a typical transition? (3) How do you feel your health system/department/division/clinic is doing at transitions? (4) What resources are offered? (5) How do you think transitions could be improved? (6) How important do you think transitions are (for you, your colleagues, your patients, and leadership)? (7) What motivates you to want to improve transitions? Responses were captured in notes during interviews and were summarized immediately afterward.

Phase 3: Implementation

Participants

To refine the intervention components identified in phases 1 and 2, we tested our intervention with 3 cohorts of 6 to 10 AYAs. Participants were eligible for the study if they were (1) currently receiving pediatric endocrinology care at UC Davis, (2) between the ages of 17 and 21 at the time of enrollment, (3) planning to transition to adult endocrinology care at UC Davis, (4) insured under Medicaid or California Children’s Services (CCS) insurance, which we used as a proxy for low-income status, and (5) able to participate in English-language video sessions. Participants were recruited by a trained clinical research coordinator both during in-person visits and through messages sent via the electronic health record (EHR). The clinical research coordinator confirmed the eligibility criteria and obtained informed consent as described below.

Intervention and Adaptations

The first of our 3 cohorts received three 75-minute group virtual sessions led by a diabetes nurse educator and supported by 2 “T1D mentors,” adults aged 23 to 35 with T1D who had previously transitioned to adult endocrinology care. Initial content included modules on the following: (1) navigating the health care system, (2) “adulting” with diabetes (eg, alcohol, pregnancy, social interactions, and explaining diabetes to others), and (3) developing a transition plan. Sessions were conducted synchronously via Zoom and included a combination of facilitator-led discussion, mentor storytelling, and interactive activities. At the end of each session, we asked for feedback from participants on the intervention components, discussion topics, and meeting format. Feedback was tracked and discussed with the research team to make subsequent adaptations to sessions. Two subsequent cohorts of AYAs received the adapted intervention.

Qualitative Interviews

Following the completion of each cohort, we invited all participants and T1D mentors to participate in a qualitative interview about their experiences with the intervention. Interviews were completed via Zoom or phone, lasted approximately 30 minutes, followed a semistructured interview guide, and were audio-recorded and transcribed. Deidentified transcripts were coded line-by-line by 2 independent coders. Coders met regularly to review codes and resolve discrepancies through discussion. Final themes were discussed, categorized, and refined by the research team. Findings from interviews from the first 2 cohorts were also incorporated into intervention adaptations.

Survey Measures

All participants completed a baseline survey on demographics, employment, educational status, and time since T1D diagnosis. After completing the intervention, facilitators and T1D mentors completed 2 short validated surveys that measure perceived intervention appropriateness and feasibility on a 5-point Likert scale: the intervention appropriateness measure and the feasibility of intervention measure [36]. Participants were also asked to complete a third validated measure, the acceptability of intervention measure. Surveys were analyzed using means and SDs.

Ethical Considerations

This study was reviewed and approved by the UC Davis Institutional Review Board (approval number 1982341). Interview participants for phases 1 and 2 provided verbal consent. AYA participants in phase 3 provided written informed consent if aged 18 years or older and written assent along with written informed consent from the parent or guardian if younger. Participants who turned 18 during the study were reconsented.

Results

Phase 1: Inspiration

Qualitative study participants included 5 pediatric endocrinologists, 7 adult endocrinologists, 4 diabetes educators, 2 diabetes pharmacists, and 8 staff members involved in specialty diabetes care. We identified 3 main findings related to transitions from pediatric to adult care for persons with T1D: (1) There is a perception among all types of clinicians and staff that transitions to adult care need improvement; (2) Levels of support for transitions vary widely depending on the patient’s distance from the health system; and (3) Components that may be helpful for a transition program include navigating the health care system independently (eg, insurance, appointments, pharmacy, and device companies) and a transition plan available to all members of the care team (pediatric and adult) in the EHR.

Phase 2: Ideation

We completed customer discovery interviews with 36 participants, including 16 people with diabetes, 8 pediatric endocrinologists, 6 adult endocrinologists, and 6 diabetes educators. Adults with diabetes who had previously transitioned emphasized that, during that period of their lives, their primary concern was not health outcomes per se but succeeding in broader developmental milestones, such as starting college, entering the workforce, or living independently. They described the transition as most effective when it supported their ability to manage diabetes in a way that enabled these life goals, rather than focusing narrowly on avoiding adverse health events like DKA or hospitalization. When reflecting on what would have helped them most during the transition, participants consistently identified 3 essential components: access to peer support, a space to ask questions without judgment, and the development of strong self-advocacy skills. Table 1 summarizes the motivating gains, pains, and value proposition statements for each adopter or influencer group.

Across interviews with pediatric endocrinologists, adult endocrinologists, and diabetes educators, several key themes emerged about the transition from pediatric to adult diabetes care. Pediatric endocrinologists saw their primary role as initiating transition conversations and referring patients to adult care, often with support from diabetes educators. Many described significant emotional distress when patients failed to transition successfully, especially when long-term patients experienced poor outcomes. They also expressed discomfort about continuing to refill insulin prescriptions for patients who should have moved on, and they described time and resource limitations as ongoing challenges. Adult endocrinologists focused more on ensuring that young adults received the care and tools they needed to manage diabetes in adulthood. While they reported less emotional strain, given they typically meet patients after the transition, they highlighted high no-show rates and limited background information as key pain points. Diabetes educators emphasized their role in providing resources and answering questions about what to expect in adult care but noted that strict oversight of how they use their time limited their ability to support transitions more fully. They were enthusiastic about anything that could reduce the burden of 1:1 education and help more patients. Customer discovery significantly influenced the intervention design, including prompting the addition of “T1D mentors,” adults who have transitioned in the past 10 years, to share their experiences and strategies, and having a specific focus on advocating for oneself (eg, explaining diabetes to others, asking for accommodations, etc). These features created the “gains” sought by participants in terms of connecting with other people with T1D and improving confidence, and addressed the “pains” of anxiety around self-advocacy.

Phase 3: Implementation

We identified 73 eligible patients at the beginning of the study. Of these, 25 were enrolled through in-clinic recruitment and EHR messaging. Seven eligible participants declined, citing that they were either not interested in research (n=4), had a work conflict (n=1), were concerned about the time commitment (n=1), or did not feel comfortable participating in English (n=1). The remaining 41 did not respond to StudyPages (Yuzu Labs PBC) messages and were not seen in the clinic during the recruitment window. Table 2 shows the demographic and health characteristics of participants for phase 3 testing. The group was diverse, with 48% (12/25) female participants, 40% (10/25) of Hispanic or Latino ethnicity, and a variety of self-identified races. The median age of participants was 18 (IQR 18-19), and median number of years living with diabetes was 6 (IQR 4-12.5). Self-reported employment status varied, with 44% (11/25) enrolled as students (including 3/25, 12% with part-time jobs), 12% (3/25) working part-time, 4% (1/25) working full-time, and 20% (5/25) unemployed and not in school.

The intervention for the first cohort consisted of three 75-minute sessions scheduled monthly. In the first cohort, 7 out of 9 participants enrolled attended the first session, but only 3 participants attended the subsequent 2 sessions, and only 2 participants completed all 3 sessions. We subsequently adapted the intervention to 2 sessions delivered over consecutive weeks in response to feedback from cohort 1 participants. In the second cohort, 5 out of 8 participants enrolled attended session 1, and 4 also attended session 2. In the third cohort, only 2 out of 8 participants enrolled attended the first session, and only 1 attended both sessions. Attendance appeared to be influenced in part by the timing of the sessions relative to enrollment, with lower participation observed when there was a longer delay between enrollment and the session start.

After the completion of cohort 1, we also increased the use of the chat function in subsequent sessions, as many participants were uncomfortable speaking on video but actively participated in the chat when asked to do so. We also adapted the transition plan component of the intervention; rather than asking participants to fill out and share their transition plans, participants preferred to watch one of the T1D mentors fill out the transition plan template and discuss the rationale behind their choices. Table 3 shows the resulting core components and adaptive forms of the intervention.

We completed a total of 15 qualitative interviews (9 participants, 5 T1D mentors, and 1 facilitator) during phase 3. Table 4 summarizes qualitative findings and provides illustrative quotes. Participants described the intervention as helpful, prompting them to think about aspects of managing their care they had not previously considered, and improving their confidence that they would transition successfully to adult care. The 2 components highlighted by participants as being most valuable were the T1D mentors and peer support. Many participants had not previously spoken with other people with T1D and appreciated the open discussions about challenges others had encountered and how they overcame them. These findings suggest that the core needs identified during customer discovery, particularly peer support, judgment-free discussion, and self-advocacy, were reflected in participants’ experiences of the intervention. T1D mentors and facilitators also perceived the intervention to be valuable. Mentors and participants had different perspectives on session participation. While mentors were concerned that some participants were not engaged in the sessions, participants themselves noted that, although they may have had their cameras turned off or not spoken often, they were nonetheless engaged and learned from the discussions between the peer mentors and other participants. Qualitative interviews also provided insights into the intervention timing, ideal number of participants, and perspectives on topics and exercises.

Seven facilitators and mentors (100% response rate) completed the intervention appropriateness measure and feasibility of intervention measure surveys. They rated the intervention as having good feasibility (mean 4.25, SD 0.52) and appropriateness (mean 4.43, SD 0.53). Of the 14 AYAs who participated in the sessions, only 3 completed the acceptability of intervention measure survey. These 3 participants rated the intervention as having high acceptability (mean 4.67, SD 0.29).

Discussion

This study describes the development and iterative refinement of a human-centered virtual transition intervention designed to support AYAs with T1D during the transition from pediatric to adult care. The intervention was intentionally designed to align with existing clinic workflows and resources while addressing priorities identified by AYAs, peer mentors, and clinicians. Stakeholder input was gathered at every stage, including from patients, peer mentors, pediatric and adult providers, and educators. Across phases, we identified peer support, judgment-free education, and the development of self-advocacy skills as essential components of an effective transition program. Through iterative fit-to-context validation and refinement, we arrived at a virtual, mentor-supported intervention focused on practical concerns such as navigating insurance, managing diabetes in educational or work settings, and advocating for accommodations. While these components were consistently identified as valuable by participants, this did not necessarily translate into sustained participation, underscoring the distinction between intervention content and participant engagement.

Consistent with previous studies, participants described the transition as a time of vulnerability, not only because of elevated risk for adverse health events but also due to major life changes such as moving out, starting college, or entering the workforce [37-39]. While DKA and hospitalizations are often used as clinical indicators of a failed transition, participants in this study defined success differently. For them, successful transition meant maintaining diabetes care in ways that supported broader developmental goals. These findings highlight the importance of aligning intervention content with the real-world priorities of young adults rather than focusing solely on clinical end points [40].

Our intervention shared several elements with prior programs, including peer mentorship and a focus on building self-management skills [16,17,41-43]. However, unlike many existing models that depend on intensive resources, such as full-time coordinators or dedicated young adult clinics [44], this intervention was designed for use in settings with limited capacity. The structure of 2 virtual sessions, led by a diabetes educator and supported by trained mentors, may offer a practical approach for clinics with limited resources to support transition care. This is especially relevant for clinics serving large numbers of publicly insured patients or those with limited infrastructure to support transition care. However, efficiency is dependent on participant engagement, and low attendance may limit cost-effectiveness and scalability in practice. Prior educational and transition-focused interventions for AYAs with T1D have similarly reported mixed findings regarding engagement and retention, with some programs demonstrating strong participation and others encountering substantial challenges with sustained involvement [16,45-47]. Difficulty engaging AYAs and families in structured diabetes education has also been identified as an important implementation barrier, even in larger, well-resourced programs [48].

We also identified several important findings related to implementation. We observed lower attendance in the third cohort, which may be partially explained by the longer delay between enrollment and the start of the sessions. More broadly, engagement and retention were variable across cohorts, reflecting common challenges in delivering interventions during the transition stage. Future programs may benefit from enrolling participants closer to the intervention start date or developing strategies to maintain engagement during waiting periods. In addition, small adaptations, such as increased use of the chat function and a shift toward observational learning, were well received by participants. These changes reflect the value of responsiveness to participant preferences and highlight how flexible delivery can enhance engagement. Future adaptations may explore shorter or more flexible session formats to better align with participant preferences and competing demands.

A key strength of this study is its focus on publicly insured AYAs, a group often underrepresented in transition research and at increased risk for poor transition outcomes. AYAs from low-income households may face additional barriers, such as limited access to technology, competing work responsibilities, or less involvement from parents or caregivers. These factors can affect both participation in and the success of structured transition programs. Future research should explore how to further reduce these barriers. Potential strategies include integrating content into clinical workflows, using mobile-first platforms, or offering incentives for completion. This study has several limitations. First, the phase 3 testing was conducted at a single academic medical center in Northern California, which may limit generalizability. While some features of the intervention may support adaptation in other underserved settings (eg, the virtual format, limited number of sessions, and use of existing clinic personnel), implementation and engagement may differ in settings with fewer institutional resources, different staffing models, rural populations, or more limited access to reliable technology. Second, the phase 3 testing experienced notable attrition across sessions, with declining attendance over time. Interpretation of feasibility, appropriateness, and acceptability findings is therefore limited by low participant survey response rates and variable attendance. In addition, a substantial proportion of eligible participants were not reached or did not respond to recruitment, underscoring the challenge of engaging AYAs during the transition stage. These individuals may represent those at highest risk for care gaps, suggesting a need for alternative outreach and engagement strategies. Finally, the study focused on user testing and did not assess clinical outcomes, such as appointment attendance or hospitalization rates. Future work should evaluate the impact of this intervention on health outcomes and explore adaptation for other populations, including those in rural areas or non–English-speaking groups.