This qualitative study used semistructured interviews to explore the dynamic nature of user engagement in OHCs. A qualitative approach was chosen because it offers the flexibility to explore the depth and complexities of users’ attitudes, behaviors, and lived experiences, aspects that are difficult to capture through quantitative methods alone [27-29]. This approach enabled us to examine the engagement patterns that emerged, understand why users engaged in particular ways, and identify their evolving needs across different points in their health information-seeking and self-management journeys.

The study was conducted within the AHA SNC, a US-based OHC that provides social support and health education to individuals and their caregivers managing various forms of cardiovascular disease (CVD). The platform offers several key features typical of established OHCs, including waiting room support for new members, small interaction groups organized by predominant CVD types, discussion boards where members can ask questions and share knowledge, and moderated educational content and resources such as news and academic articles curated by community managers. At the time this study was conducted, the program managers reported that the AHA OHC had about 12,000 registered users.

Throughout the study, we used a participatory research approach, collaborating closely with community managers who serve as important stakeholders in how OHCs function. This collaboration ensured that our research instruments were contextually appropriate and that findings would be actionable for community improvement.

Participants were recruited through the AHA SNC. The research team developed a recruitment survey containing eligibility questions and baseline sociodemographic items. This survey was iteratively reviewed and revised with community managers to ensure relevance and appropriateness for the community context. Community managers then distributed the survey across the platform via community announcements and direct communication channels, inviting interested members to participate.

Survey completion was voluntary, and only individuals who completed the survey, indicated interest in participating, and met eligibility criteria were invited to the interview. Eligibility criteria were intentionally broad to capture diverse engagement experiences: (1) participants must be at least 18 years of age, and (2) participants must be current members of the AHA SNC, regardless of membership duration, engagement level, or specific cardiovascular condition.

Interviews were conducted in batches with concurrent data analysis, allowing emerging insights to inform subsequent interviews while maintaining consistency in core questions. An initial round of recruitment revealed demographic imbalances, particularly an overrepresentation of White participants, prompting targeted recruitment of underrepresented participants. Recruitment continued until data saturation was achieved [30], defined as the point at which new interviews provided existing insights and themes became well-established across the dataset. A total of 19 participants were interviewed for this study.

We collected data through one-on-one semistructured interviews via Zoom (Zoom Video Communications) audio over 2 months, from mid-June to mid-August 2023. Each interview lasted approximately 1 hour. The lead author (M Jayeoba) conducted all interviews, using a semistructured protocol guide that allowed for flexible exploration of participant experiences while ensuring systematic coverage of key topics across all interviews [31]. The interview protocol explored initial engagement and engagement patterns, information exchange behaviors, role evolution, platform usability and navigation, unmet needs and barriers, and design preferences and suggestions (Multimedia Appendix 1).

Before each interview, participants were reminded of their rights, and the interviewer sought oral confirmation of consent in addition to the written consent previously obtained. Participants were informed that audio would be recorded for analysis purposes. All sessions were recorded and saved to Zoom Cloud Service, then professionally transcribed by Rev human transcription service to ensure accuracy.

Throughout data collection, M Jayeoba maintained detailed reflexive memos documenting observations, emerging patterns, methodological considerations, and questions for follow-up. These memos were discussed regularly with the research team, who provided feedback and suggestions for adapting the interview protocol as patterns emerged. This iterative approach allowed us to probe more deeply into unexpected themes while maintaining consistency in data collection. The study did not offer monetary compensation; however, the interviewer and community managers thanked the participants for their time and knowledge contributions.

The interview transcripts and memos were imported into Microsoft Excel and MAXQDA (VERBI Software) for data analysis and management. Upon reviewing the interview transcripts, we identified discrepancies in role designation for 2 of 3 participants who identified as family or friend and formal caregiver in the recruitment survey. The 2 participants’ interview narratives described their own recent diagnoses and personal health journeys. One of the 2 participants explicitly stated joining the community after being diagnosed and described personal experiences consistent with those of patients. The second one mentioned that they initially joined the community as a caregiver but later received a personal diagnosis that prompted them to seek support for their own health. Since participants received no financial incentive, we believe these discrepancies likely stemmed from unintentional survey errors rather than misrepresentation. To preserve analytic coherence, these participants were reclassified as patients. Additionally, we excluded the data from the third participant who identified as a formal caregiver since we no longer had sufficient representation from this user group. Our final analytic sample conclusively includes 19 individuals managing cardiovascular conditions.

To analyze our data, we used thematic narrative analysis, a qualitative method that combines thematic analysis with narrative analysis [32]. Thematic analysis guided our coding and theme construction process [33,34], while narrative analysis was used to organize participants’ retrospective experiences with the AHA SNC into stages [32,35]. This blending reflects a methodological stance that meaning in chronic illness contexts is often best understood through patients’ storytelling [36], and applying multiple methods of systematic evaluation, such as thematic and narrative techniques, can increase interpretive validity and lead to richer, more accurate representations of lived experience [36,37]. Meanings were created inductively at a semantic level to capture community members’ stories and reflections. Despite this methodological flexibility, our thematic narrative analysis processes followed a systematic approach that was structured into 6 stages [33,34]. We initially performed data immersion, reflecting deeply on the data for familiarization, meticulously reviewing the transcripts, and verifying their accuracy and completeness against the audio recordings. Following this, we first developed preliminary codes from open coding of 5 transcripts using Microsoft Excel. These codes were refined through team discussions to resolve discrepancies and ensure coherence, and Microsoft Excel provided the flexibility for that. The final code structure was then transferred into MAXQDA for structured analysis of the remaining 14 transcripts, allowing for organized data management and code frequency review.

The participant data were not analyzed solely based on their current engagement stage, as their narratives spanned the breadth of their experiences since first discovering and joining the OHC. Instead, we clustered these diverse experiences and narrated them into progressive yet nonlinear stages of engagement. This approach acknowledges our reflection on the data, that participants do not always fit neatly into a single stage at a given moment, but rather exhibit behaviors and motivations that traverse multiple stages, offering a holistic understanding of their long-term engagement journey. Coding decisions evolved through continuous refinement until conceptual saturation was determined when no new themes emerged across successive interviews [30,38]. After the 19 transcripts were coded, the lead researcher (M Jayeoba) grouped the codes into potential themes, and all the authors discussed the themes to mitigate bias and establish congruence. The discussion led to collaborative definitions and labels of the themes that comprise the thematic map of the analysis.

After data analysis, we compiled a high-level report summarizing each theme, including representative participant quotes, and presented it to the AHA community managers in a virtual feedback session. This session was conducted to ensure that those responsible for managing the platform were informed by participants’ perspectives and could use the insights to guide decision-making and improve community engagement strategies. During the meeting, we invited open-ended reflections on the findings, where community managers noted that many of the findings aligned with their own observations, but we did not systematically analyze this feedback. While the community managers’ input did not influence our coding or themes, it provided an informal validation and practical relevance of our findings from a stakeholder perspective.

Finally, to enhance research quality, we used several strategies recommended for qualitative inquiry [39] (Checklist 1). We documented all analytical decisions and theme development processes. Multiple researchers were involved in data interpretation to provide diverse perspectives and challenge assumptions. Member checking occurred informally through our ongoing collaboration with community managers, who reviewed emerging themes and confirmed alignment with their observations.

The protocol for this study was jointly developed with the community and program managers of the AHA SNC to ensure compliance with community values and ethical standards. The institutional review board at Northwestern University reviewed the protocols and granted an exemption (STU00216828). Before data collection, all participants provided written informed consent via REDCap (Vanderbilt University Medical Center). Participants had opportunities to clarify ethical concerns before engaging in the study and were reminded of their rights at the start of each interview session. All data were anonymized with names and identifying details removed to ensure privacy compliance. Participants did not receive monetary compensation for their time, as the project was not funded. However, the researchers and program managers sincerely thanked the participants for their time and contributions.

This section presents our analysis of user engagement patterns in an online cardiovascular health community through a stage-based model. Our study included 19 members of the AHA SNC who participated in the semistructured interviews. The sample included 10 female and 9 male participants, with participants identifying as Asian (n=1), Black or African American (n=1), Hispanic or Latino (n=1), White (n=15), and other (n=1) backgrounds. Most participants were highly educated, with 10 participants holding advanced degrees (MA, MS, MBA, PhD, MD, and JD), 5 having completed college degrees (BA or BS), and 4 having some college, technical degrees, or associate degrees. The sample was primarily older adults, with the age distribution showing concentration in the 65‐ to 74-year range, reflecting a mature cohort navigating the complexities of long-term health management. Participants reported diverse cardiovascular conditions, with many managing multiple diagnoses simultaneously. It is worth noting that all the study participants used the same OHC platform and information architecture, or set of features, within the platform, thereby contributing to the validity of our method and results. See Table 1 for detailed participants’ sociodemographic and health characteristics (N=19).

Our analysis reveals how members progress through 4 distinct stages of engagement in the AHA SNC: discovery, exploration, commitment, and integration. The stages represent phenomenological or retrospective journey reconstructions rather than true time-series trajectories. At the discovery stage, members exhibit initial awareness with limited engagement, primarily focusing on orientation needs. During exploration, members begin trying out features and engaging in some interaction, and their need for connection becomes paramount. As members advance to the commitment stage, they demonstrate consistent participation and relationship building, often taking on supportive roles and seeking recognition for their contributions. Finally, in the integration stage, members achieve sustained but flexible participation, requiring support to maintain a healthy balance in their community engagement. We organized our findings into 2 broad categories: first, we identified stage-specific engagement patterns and the challenges affecting progression through each stage; second, we uncovered emerging opportunities to support members’ needs at each stage of their community involvement. See Textbox 1 for the categories and themes that showcase the perceived stage-specific engagement patterns and the design opportunities to facilitate long-term engagement across stages.

This study explored perceived engagement patterns, including user behaviors, motivation, and needs in a large OHC through a stage-based framework. The study highlighted 4 distinct, nonlinear yet progressive engagement stages: discovery (crisis-driven initial engagement), exploration (navigation and orientation), commitment (active engagement and information management), and integration (sustained engagement and mentorship). Key barriers included information architecture complexity, concerns about misinformation, limited support for role transitions, and decreased participation as health management improved. Participants identified opportunities through which OHCs could increase stage-based engagement, including adaptive recommendation systems, health information literacy programs, structured role transition support, and alternative engagement modalities, such as synchronous interactions and health tracking tools. We summarized this stage-based model through a visual representation in Figure 1.

Previous research on OHC engagement has typically classified users into discrete categories such as lurkers, contributors, and leaders [40-42]. While these classifications provide useful snapshots of participation patterns, they obscure the dynamic nature of engagement over time. Our stage-based model extends this literature by revealing how and why users transition between different forms of participation. Rather than representing fixed user types, these patterns reflect different phases in members’ engagement journeys, with progression depending on whether the platform adequately supports their evolving needs.

Our findings align with legitimate peripheral participation theory [43], which posits that newcomers gradually move from peripheral observation to full community membership through increasing engagement and identity development. However, we found that this progression is neither automatic nor unidirectional. The discovery stage represents a critical intervention window where newly diagnosed patients seek information and emotional support during moments of crisis. Yet, many encounter barriers that prevent progression to deeper engagement. Consistent with existing studies [26,44,45], participants highlighted the overwhelming nature of navigating vast amounts of information during this stage. These barriers include information overload, health literacy gaps, and technical challenges that compound the emotional distress of a recent diagnosis. OHC architectures should account for the reduced cognitive capacity and heightened vulnerability of those in crisis-driven entry.

During the exploration stage, our findings challenge assumptions about how condition-specific communities should be structured. Participants sought connections not just with others sharing the same diagnosis, but with those sharing similar circumstances, treatment approaches, and demographic characteristics. This aligns with social identity theory [46], which emphasizes the importance of identity-based group formation for fostering belonging and sustained participation. The evolution from transactional information exchange to relational connection demonstrates how appropriate relationship formation enables shifts in engagement priorities and facilitates progression toward more active roles.

The commitment stage introduces a paradox not adequately addressed in prior OHC literature. As members become more knowledgeable and engaged, they develop sophisticated strategies for managing misinformation risks through disclaimer practices, systematic verification, and topic-based participation patterns. While these behaviors reflect responsible community norms, they also create barriers to the detailed, experience-based knowledge exchange that initially attracted members to the community. This tension between information seekers desiring specific, actionable details and information sharers providing cautious, generalized advice creates dissatisfaction on both sides and may contribute to reduced engagement over time.

A critical finding that extends existing engagement models is the fluid nature of user roles. Consistent with recent literature emphasizing the temporal dimensions of online participation [17,40,41], we found that users oscillate between information-seeking and information-sharing based on health status changes, knowledge saturation, and personal circumstances. What may appear as disengagement often represents appropriate adaptation to changing needs, such as members whose health has stabilized, maintaining peripheral connection through email alerts while reducing active participation within the community platform. This challenges linear models of engagement progression and suggests the need for platforms that accommodate fluid role transitions.

These findings have important theoretical and practical implications for understanding and supporting OHC engagement. Theoretically, our stage-based model advances understanding of how engagement evolves by specifying the mechanisms that drive progression and the barriers that impede it at each stage. This moves beyond static typologies to reveal engagement as a dynamic process shaped by the interaction between individual needs and platform affordances.

Several critical barriers emerged that prevent members from progressing through engagement stages. Information architecture complexity prevents users from locating relevant content during exploration, despite high initial motivation. Current platforms use broad categorization schemes that may not account for heterogeneity within diagnostic categories. Limited structural support for role transitions represents another critical gap. While participants expressed clear interest in providing mentorship as they gained experience, many lacked pathways to assume supportive roles. This reveals that role transition support requires more than opportunity; it requires structured onboarding, training, recognition mechanisms, and ongoing support from community managers.

From a practical standpoint, our findings suggest 4 design opportunities to address these barriers and support stage-based engagement. Adaptive personalization systems represent a promising approach to reducing overwhelming information volumes during discovery, facilitating identity-based connections during exploration, and identifying role transition opportunities during commitment and integration stages. Implementation should consider multiple dimensions, including diagnosis and cause, treatment approaches, demographic factors, health journey stage, and engagement preferences, leveraging both explicit user profiles and implicit behavioral signals.

Patient education and digital literacy support offer another opportunity for improvement. Educational interventions should address the perception that experiential knowledge is less valuable than medical expertise by clarifying appropriate boundaries between experiential sharing and medical advice, teaching effective verification strategies, normalizing disclaimer practices, and building confidence in peer support value. Such education should be stage-appropriate, from platform orientation during discovery to mentorship training during integration.

Addressing the challenge of role transitions requires structured support mechanisms that provide clear pathways from information-seeking to mentorship roles with transparent descriptions of responsibilities and available support. This includes formal onboarding processes, ongoing training opportunities, recognition systems, and community manager support. Given the fluid nature of user roles, platforms should support flexible transitions that accommodate members’ evolving needs while providing structure for those seeking to contribute.

Finally, diversifying engagement modalities through synchronous activities and health tracking tools can address unmet needs beyond asynchronous forums. Synchronous formats offer immediacy, social presence, and relationship-building opportunities particularly valuable during exploration and commitment stages, while hybrid approaches may best serve diverse member needs by combining asynchronous core functions with optional synchronous activities.

Several limitations should be acknowledged. First, our sample was drawn exclusively from the AHA SNC, which had over 12,000 users at the time of this study and focuses on CVD management. The AHA support network is a moderated forum hosted by a major health organization, and so, features like content moderation, the presence of expert material, or the specific community culture could influence engagement patterns in ways that might differ from other health foci (eg, unmoderated peer groups or those for other health conditions). While the availability of these features provided research depth, it may limit generalizability to other health conditions or OHC platforms with different structures, moderation approaches, or demographic compositions. Different disease communities may exhibit unique engagement patterns based on disease characteristics, prognosis, and treatment trajectories. As such, the proposed stage-based model should be viewed as a conceptual framework that may require adaptation across different community types. Second, our sample was predominantly White, highly educated, and older adults, which may not reflect the experiences of more diverse populations. Younger users, individuals from different racial and ethnic backgrounds, or those with lower educational attainment may face different barriers and have different engagement preferences. The older adult demographic may particularly influence our findings regarding digital literacy and platform navigation challenges. Moreover, our sample size (N=19), although appropriate for qualitative inquiry, may not fully capture the diversity of experiences within each engagement stage. Some stages may be underpinned by fewer cases, and additional research is needed to confirm the stability and completeness of these stages in broader or longitudinal studies. Third, our retrospective interview approach relied on participants’ recall of their engagement journey, which may be subject to memory biases or retrospective sense-making that does not accurately reflect their experiences at each stage. Additionally, the use of narrative analysis to construct stages of engagement may potentially limit our findings, as the data are based on retrospective self-reports that are vulnerable to post hoc rationalization and social desirability bias. Participants’ narratives may reflect how they currently make sense of past engagement rather than how engagement actually unfolded over time. As such, the staged model risks overinterpreting sense-making narratives as actual engagement trajectories. To ensure the validity of our claim, we present our findings as perceived engagement patterns. In the future, longitudinal studies following users from initial discovery through integration should be conducted to provide more accurate stage-specific data. Fourth, we did not collect systematic use data (eg, posting frequency, login patterns, and types of content engaged with) to triangulate with interview findings, which could have provided additional objective measures of engagement patterns. Finally, participants who agreed to be interviewed may represent more engaged or satisfied users, potentially missing perspectives of those who discontinued use or remained peripheral lurkers. This may bias the findings toward more affirming engagement trajectories and obscure potential barriers faced by disengaged or underserved users. Future work should aim to capture these critical viewpoints to enrich and challenge the current model.

Through narrative thematic analysis, this study identifies perceived stage-based patterns in how participants retrospectively described their engagement with an OHC, suggesting that engagement is experienced as dynamic and nonlinear in relation to perceived changes in health status, knowledge, and social needs. By applying a stage-based framework—discovery, exploration, commitment, and integration—this study highlights how mechanisms such as self-efficacy building, trust formation, and role transition shape participation over time. While information needs drive initial engagement, sustained participation requires facilitators such as opportunities for meaningful contribution, relationship-building, and alternative forms of engagement beyond traditional question-and-answer interactions. Importantly, this study identified that concerns about misinformation not only affect information-seeking but also members’ willingness to share lived experiences, creating tension between safety and openness.

Sustaining engagement requires adaptive strategies that anticipate these shifts: personalized content delivery through machine learning, health information literacy programs to empower safe participation, and structured pathways for mentorship and leadership roles. Diversifying engagement modes, including synchronous interactions and self-management health tools, can further maintain the relevance and value of OHCs for experienced users. Designing for the full engagement cycle, from crisis-driven entry to integrated community leadership, offers a roadmap for platforms to foster resilience, reciprocity, and long-term value. Future research should validate these interventions across diverse communities and examine their impact on health outcomes. Ultimately, supporting continuity in engagement is essential for realizing the promise of OHCs as enduring sources of peer support and collective knowledge.