In order to identify the factors at multiple levels of analysis contributing to interorganizational knowledge sharing, we adopted a longitudinal qualitative approach in the interpretive tradition [22], drawing on multiple case studies and conducting interviews with experts in the field within and beyond organizational settings.

We studied a national digital transformation initiative in the English health care sector from 2017 to 2023, called the GDE Program. The GDE Program was a national investment in the transformation of a variety of care settings, including acute care, specialist care, mental health, and ambulance services of the National Health Service (NHS) in England, with a focus on digitally advanced sites to become international exemplars of excellence in order to provide learning for later organizations planning to go through digital transformation. In this program, provider organizations, henceforth referred to as GDEs, were paired with somewhat less mature fast follower (FF) provider organizations to promote learning across settings. GDEs and FFs captured the best practice models and lessons, including technologies, processes, and structures, in “blueprints,” which were aimed to be disseminated within and beyond the program to accelerate the spread of transformation knowledge. Shared learning in the program was also planned to be facilitated by learning networks launched in the first half of the program, bringing together GDE and FF team members with an interest in specific topics.

Our data analysis integrated well-established approaches, including longitudinal case analysis, grounded theory development, and content analysis. The process was iterative, requiring continuous movement between the data and the evolving theoretical framework [23]. However, for clarity, we present our analysis as a structured sequence of 5 main stages. In stage 1, the lead researcher, HM, utilized NVivo 14 (Lumivero) to organize and integrate all data sources and conducted initial open coding to capture the breadth of stakeholder views and experiences. This stage generated a large pool of provisional codes across interviews, observations, and documents.

In stage 2, we adopted a focused coding strategy to move from initial descriptive codes to analytically meaningful first-order codes. Guided by the principles of parsimony, conceptual salience, and recurrence across data types, we treated sentences and paragraphs as coding units [24] and read the full dataset iteratively, including catalogs and supporting materials. We assigned labels using in vivo codes where appropriate or concise descriptive terms and then grouped related segments into coherent conceptual categories [25]. Importantly, our intention at this stage was not to catalog every detail within such a large dataset but to distil patterns that were recurrent, theoretically relevant, and directly related to our research questions. This analytic focus explains why the final set of first-order codes is relatively concise (n=33). To ensure robustness, coding decisions were discussed within the research team at multiple points, allowing for refinement of categories, challenge of interpretations, and agreement on the conceptual boundaries of each code. Throughout this process, we examined perspectives at multiple analytic levels, including individual, adopter organization, government, and technology supplier, as well as across different stages of digital transformation (procurement, adoption, and use). This enabled us to generate a focused yet comprehensive set of first-order codes that meaningfully represented mechanisms shaping knowledge sharing across organizations.

In stage 3, we identified links among first-order codes to form the emerging second-order themes [24]. This involved discussing disconfirming evidence within the extended research team to minimize the risk of researcher bias.

Stage 4 involved linking second-order themes to form aggregate dimensions [24,26], associated with our research question. These 3 stages (2, 3, and 4) formed our data structure, as presented in Table 1. The Results section begins with explaining this data structure in detail.

In stage 5, after identifying the aggregate dimensions in stage 4, we noted strong parallels between our grounded analysis and context-mechanism-outcome (CMO) [27] configurations described in realist literature, that is, the taxonomy of multilevel factors incorporated context and mechanisms, and shaped the spectrum of interorganizational relationships and partnerships (outcomes). Building on the similarities, we propose a framework that shows the relationship between the aggregate dimensions. We structured the reporting of this study in accordance with the COREQ (Consolidated Criteria for Reporting Qualitative Research) guidelines to enhance rigor, transparency, and completeness [28].

We obtained institutional review board approval from the University of Edinburgh Research Ethics Committee (27.11.2017). Informed consent was obtained from all participants prior to their involvement in the study, and all data collected were anonymized to ensure confidentiality and protect participant and organizational privacy. Participants were not compensated for their time.

Overall, our dataset consisted of 228 interviews; observations of 113 national strategic meetings, user groups, and conferences; analysis of 277 documents; and desk research into online platforms of knowledge-sharing settings such as forums and webinars. Table 2 presents the interviewee characteristics. We present the findings under 2 key headings: multilevel factors contributing to the sharing process and the spectrum of interorganizational relationships and partnerships. Subsequently, informed by the CMO perspective, we present an integrative view of how the factors across 4 levels—macroenvironmental, interorganizational, organizational, and individual—influence knowledge-sharing modes. The factors identified in each category are not exhaustive; however, they highlight some of the most recurrent factors identified in our data within each category.

The factors identified in our study do not operate in isolation; rather, elements across macroenvironmental, interorganizational, organizational, and individual levels are highly interconnected and often mutually reinforcing in shaping knowledge-sharing relationships. For example, national policy developments, such as the Wachter Review, not only stimulate the creation of new digital leadership roles (eg, Chief Clinical Information Officer and Chief Nursing Information Officer) but also contribute to the emergence of a digitally skilled cadre through initiatives like the NHS Digital Academy, thereby improving the absorptive capacity of organizations and their ability to engage meaningfully in communities of practice. Similarly, program-level structures, such as mandated GDE/FF pairings and blueprinting requirements, interact with relational dynamics, including the formation of collective identities (eg, GDE status) and the development of trust-based recognition between partner sites, which in turn facilitate voluntary knowledge exchange beyond formal contractual obligations. Technology supplier strategies also intersect with organizational search behaviors: suppliers’ diversification of the market can drive adopter organizations to seek external knowledge (eg, identifying who has implemented specific technologies), often aided by intermediary actors or technology-specific user groups that bridge otherwise disconnected sites. Geographical proximity further amplifies these interactions, as staff from nearby organizations are more likely to visit one another, participate in shared communities of interest, or move across settings, thereby strengthening informal knowledge flows that complement formal governance mechanisms. Finally, individual-level motivations, such as professional reputation, intrinsic interest, and past experience in handling knowledge requests, shape how actors engage with these structures, while personal inhibitors can constrain knowledge sharing even when enabling conditions are otherwise present. Together, these interconnected elements illustrate how multilevel factors combine to activate and shape the modes underpinning knowledge sharing across the health system.

Together, the above interconnected conditions contributed to a spectrum of interorganizational relationships and partnerships—what we refer to as modes of knowledge sharing—that emerged from multiple origins and were shaped by varying degrees of structure and formality. These mechanisms ranged from highly orchestrated arrangements to more flexible, organically developed connections. Across this spectrum, organizations used diverse tools and practices for knowledge sharing as described below.

The origins of relationships ranged from internal (individual or communal within and across transforming organizations) to external (beyond the transforming organizations). In the case of internal origins, sometimes provider organizations collaborated on a local needs basis (eg, site visits or phone calls between individuals) or more strategically entered into intentional cultivation of networks and communities in response to collective needs or to achieve collective strategic benefits. For instance, when multiple organizations needed particular functionalities from the same vendor, organic and spontaneous collaborations between 2 or more user organizations helped to create a collective force to exert leverage over technology vendors or to collaborate and form a technological solution that benefited multiple organizations.

In the case of external entities initiating relationships, in some instances, interorganizational relationships were initiated by formal mandates from the government. For instance, in our study, 19 strategic pairings were formed between GDEs and FFs. These partnerships were based on a range of factors such as common core technologies and local strategic groupings, for instance, between organizations that had a particular electronic patient record system or between nationally coordinated collaborations of health care organizations and local authorities.

Technology suppliers were also initiators of particular types of relationships around specific technologies, both nationally and internationally.

The wide spectrum of interorganizational relationships had different governing structures and durations. For instance, government-initiated partnerships could range from national conferences or conference series (short-term, flexible periodic engagements) to fixed-term/long-term contracts between different provider organizations (eg, GDE/FF partnerships). These contracts mandated a number of aspects, such as the type of sharing (documentation), the interval for sharing, and what is to be shared. Despite investments into these formal means of knowledge sharing, many of the initiatives did not reach their anticipated longer-term benefits, despite being easy to measure, and their purposes changed over time. One example is blueprints, which were mandated by the GDE Program and were designed as tools for sharing the knowledge needed to implement specific models of change to be adopted by the wider ecosystem. While they offered codified knowledge about new process designs, they did not provide the means to implement certain technologies in the new context, as the implementation process involved a range of tacit knowledge that was not included in the blueprints. Many provider organizations used blueprints as a tool to identify which organizations and which actors were sources of particular knowledge in order to establish informal contact and to allow for the sharing of knowledge beyond codified documents.

We also noted that not all formal initiatives proved to be self-sustaining over time. For instance, GDE Program managers launched a number of learning networks, many of which were terminated after the end of national funding. Only some networks driven by professional agendas and interests (eg, the Hospital Electronic Prescribing Medicines Administration Network) sustained and transformed into national communities of practice after the completion of the initial funding period. Collaboration and sharing of knowledge in these communities were not based on policy-mandated contracts, and instead, knowledge sharing was orchestrated through collaboration of provider organizations as they captured local and collective benefits from contributing to the community.

Building on these findings, we developed an integrative model (Figure 1) that brings together the multilevel factors (context and mechanisms on the left-hand side) identified in our analysis with the resulting spectrum of interorganizational relationships and partnerships (outcomes on the right-hand side). The model illustrates how different contextual conditions and mechanisms combine to produce a diverse array of knowledge-sharing relationships, which vary not only in formality but also in their roots and modes of initiation. To capture this variation, we conceptualized a 2D spectrum: the x-axis represents the degree to which a relationship is formal or informal, while the y-axis indicates whether its origin is internal (arising from organizational strategies, leadership roles, or intrasystem initiatives) or external (emerging from policy directives, supplier strategies, intermediary actions, or professional networks). Importantly, the left and right sides of this spectrum are connected through a range of influences, such as triggering, mandating, controlling, and enabling, which shape, prompt, or support interorganizational knowledge sharing across different settings. By positioning relationship types within this matrix, the model demonstrates how combinations of enabling factors give rise to different patterns of collaboration and partnership across the ecosystem, offering a coherent way to understand the diversity and dynamics of interorganizational knowledge-sharing arrangements.

Our findings show that effective interorganizational knowledge sharing is shaped by different configurations of factors linking across the levels of a system. At the macro level, policy stability, national investment, and the presence of formal digital transformation programs created perceived legitimacy, shared purpose, and a sense of mandate among organizations, resulting in higher engagement in knowledge-sharing activities and more formalized collaborations. At the interorganizational level, networks characterized by trust, prior relationships, and aligned implementation timelines enabled modes of reciprocal exchange, openness, and problem-solving, leading to sustained cross-site learning, joint solution development, and more mature partnerships. In contrast, competitive procurement pressures or misaligned incentives triggered protective behaviors, producing fragmented or transactional exchanges. These examples of different configurations illustrate how and why knowledge-sharing modes vary in effectiveness and evolve over time.

In this section, we compare our empirical findings with the broader literature on knowledge sharing across organizations and present how our findings contribute to the development of the concept of evolving learning ecosystems.

Combining our findings with the extant literature that addresses macroenvironmental factors on interorganizational knowledge sharing, we identified a wide range of macroenvironmental factors, including policy developments, intellectual property regimes [29], technological sophistication levels [30], technology suppliers and their strategies [31], diverse markets [32], and other external agencies (eg, universities, civil society groups, third-party firms that have close connections with the health care system, etc), which can trigger, mandate, control, or enable interorganizational knowledge sharing. Policy-related factors can act as “windows of opportunity” [29] that enable or mandate the formation of relationships for sharing specific types of knowledge [33]. However, the outcome of using the mechanisms mandated or enabled by policies is dependent on how knowledge-sharing stakeholders exploit these opportunities [18]. For instance, when the government creates formal partnerships and mandates the sharing of documents in a repository accessible to a wide range of organizations, the scale of sharing may be large, but the scope and depth of sharing may be limited. Other external entities, such as suppliers (through their strategies) and intermediaries, also shape the types of relationships being formed and the collaborations that may take place between different organizations [30]. Typically, relationships formed by suppliers are less formal, sometimes ongoing, and sometimes on an ad-hoc basis. A common finding in the GDE Program was supplier-coordinated networking activities such as user groups, pilot site visits, and connections of key individuals across organizations. Intermediaries may offer formal partnerships by providing specialized professional services to organizations and facilitating sharing between them [34,35] or providing technology brokerage activities [36]. Intermediaries may also enable the formation of communities of interest, where the relationship is less formal. In general, the macroenvironmental factors impact interorganizational knowledge sharing both directly and indirectly. In direct terms, they can mandate (eg, blueprinting activities in the GDE Program [18]) and structure relationships (eg, GDE and FF relationships formed between more digitally advanced organizations and less digitally advanced organizations to share experiences [37]). In indirect terms, they can influence the ability and motivation of organizations to form relationships, for example, by identifying who holds what knowledge about a particular type of technology.

Interorganizational factors, including relational factors [38], structural governance mechanisms, and interorganizational proximity [38], have a profound impact on shaping the early stages of relationship/partnership building. Our research shows that by building a collective identity that involves the development of a common objective or a mutual win and involves shared rules and regulations in terms of both formal [39] and informal [40] relationships, knowledge sharing can be encouraged. These can be achieved through building interorganizational teams or communities of interest that invest in modes of knowledge exchange and integration in order to solve a problem or overcome a common challenge [41-43]. Additionally, structural governance mechanisms (eg, whether they are time-bound or not, or whether they are periodic or run on an ad-hoc basis) also affect the type of relationship and the direction of knowledge flow. Furthermore, organizational proximity plays an important role in the formation of both formal and informal sharing. The extant literature highlights various aspects of organizational proximity, including cultural closeness [44-46], similarity in technological artifacts in use [47], alignment in strategic direction [48], and geographical proximity, and these facilitate smoother mobility of personnel, socialization, in-person exchange, and informal ties [36,49].

In terms of organizational factors, our study highlights the importance of knowledge search strategies by organizations, absorptive capacity, and past knowledge-sharing experiences in triggering and enabling sharing, as well as a wide spectrum of interorganizational knowledge-sharing modes. The extant literature has focused on the importance of absorptive capacity, which refers to the ability to recognize the value of knowledge being shared by both the donor and receiver organizations and the ability to utilize it [50]. The absorptive capacity is influenced by an organization’s past experiences of identifying and building relationships [51], ability to integrate knowledge [52], and capability to retain knowledge [53]. We also found that as organizations gain experience in sharing and managing knowledge demands, they become more capable of managing and coordinating relationships as well as organizing the learning process. Organizations may use tools, such as IT-enabled systems, to eliminate barriers of communication and thereby heighten the level of knowledge sharing.

In relation to individual and interpersonal factors, we highlight the role of intrinsic and extrinsic motivations in encouraging knowledge sharing and the personal inhibitors that hinder the sharing process. Research has revealed a range of explanations for why individuals contribute to knowledge sharing in formal and informal settings. Many explanations point to intrinsic (self-interest) and extrinsic (altruistic) motivations [54]. Such motivations include self-benefits received from participation in knowledge-sharing communities and settings, such as access to expert advice [55], insights into others’ beliefs and opinions, enhanced reputation and professional status [56,57], networking [58], and improved self-image and confidence [59]. Benefits to others include a desire for building or sustaining communities of interest [59,60], feelings of companionship [57], and reciprocity [61]. These motivations may vary in terms of nature and depth in formal and informal knowledge-sharing relationships. Bateman et al [62] highlighted that an individual’s commitment to the organization they belong to, as well as the community, has a significant effect on participating in knowledge sharing. In this way, individuals may believe that they benefit from knowledge-sharing relationships [63] in both formal and informal settings, or they may feel obliged to pay back to existing relationships [56,62]. While the literature highlights the role of motivations in facilitating knowledge sharing, it also identifies factors that may hinder the transfer process. These include the time required for sharing activities [59] and the feeling of a strong psychological ownership over knowledge [63]. Emotional attachment to the organization may also discourage individuals from sharing certain knowledge, as it leads them to prioritize the organization’s interests [64], influencing their decisions on what knowledge to protect and how to control its release [65].

Our study contributes to the literature on digital transformation and interorganizational learning by advancing the understanding of how learning ecosystems emerge, evolve, and take shape across multiple levels of the health care system. Rather than treating macroenvironmental, organizational, interorganizational, and individual factors as discrete categories, we demonstrate that they operate as dynamically interconnected influences that jointly shape the evolution of knowledge-sharing relationships. Through our multilevel analysis, we show how contextual conditions, such as national policy directives, program mandates, supplier strategies, and organizational digital maturity, activate mechanisms, including identity formation, absorptive capacity, motivations, and relational recognition. These mechanisms, in turn, produce a spectrum of relationship types that vary in formality and origin, ranging from informal peer interactions to mandated partnerships.

By integrating these insights, our work conceptualizes the learning ecosystem as an evolving sociotechnical system in which relationships and knowledge flows continually adapt in response to shifting contextual pressures and emerging opportunities. We demonstrate that learning ecosystems do not develop through linear or centrally orchestrated processes; instead, they emerge through recursive interactions across levels, where changes in one domain, such as policy initiatives or new digital leadership roles, can reconfigure organizational strategies, reshape interorganizational networks, and influence individual patterns of knowledge sharing. The integrative model we propose brings these dynamics together by illustrating how multilevel factors (contexts and mechanisms) give rise to diverse modes of interorganizational collaboration (outcomes), and how triggers, mandates, controls, and enablers collectively shape the system’s trajectory over time. This contribution extends existing frameworks on interorganizational knowledge sharing by offering a holistic explanation of how learning ecosystems are formed, how they function, and why they evolve differently across settings. In doing so, it provides theoretical clarity for understanding and supporting the development of interorganizational learning ecosystems capable of enabling digital transformation at scale.

This study offers important implications and insights that can be used in enabling and supporting the digital transformation process in the health care sector. Drawing on our findings, we have developed a structured set of questions designed to help policymakers and practitioners diagnose the conditions shaping knowledge sharing and monitor how these conditions evolve over time. These questions focus on the macroenvironmental, interorganizational, organizational, and individual factors that influence the emergence and functioning of learning ecosystems. We present these questions in Table 3. These questions serve as both a diagnostic aid (helping stakeholders assess what they are trying to achieve and the contextual factors shaping these efforts) and a monitoring aid (helping them evaluate how learning conditions and relationships develop over time).

To address macroenvironmental questions, policymakers and practitioners must examine public policies that facilitate or hinder the formation of formal and informal partnerships, as well as the role of external agencies in enabling knowledge sharing. Additionally, it is important to identify external factors, such as competition and cooperation. This requires assessing whether policymakers or other external entities demand the sharing of specific types of knowledge and whether industry characteristics, such as the levels of technological complexity and market maturity, affect partnership and community-building. To answer interorganizational questions, policymakers and practitioners should explore structural governance mechanisms such as community-based orchestration, contracts, defined processes, and routines. They must also examine relational and cognitive governance factors, including trust and power dynamics. Exploring organizational questions requires assessing the capabilities and experiences of donor and recipient organizations in forming and managing partnerships or participating in communities. This includes their ability to manage alliances, learn from and apply external knowledge, and facilitate knowledge sharing. Additionally, behavioral aspects, such as organizational motivations, incentives for knowledge sharing, and resistance to new knowledge, must be considered, along with organizational characteristics like the credibility of the donor and the receiver’s flexibility to adapt to change. Finally, addressing individual-level questions requires examining both intrinsic and extrinsic factors influencing knowledge-sharing behavior.

Building on this diagnostic foundation, we offer a second set of questions that guide stakeholders in identifying and selecting appropriate modes and mechanisms of knowledge sharing for digital transformation initiatives. Rather than prescribing a fixed sequence of steps, these questions encourage reflection on the purpose of different knowledge-sharing modes, their temporality, the coordination techniques they require, and the tools that support them. These questions are presented in Table 4. Stakeholders can populate them with context-specific examples and options relevant to their initiatives. Table 4 includes 3 illustrative columns to demonstrate how different knowledge-sharing modes may be examined and compared.

Consistent with existing literature emphasizing the importance of relational maturity and temporality in knowledge-sharing processes [45,52,66,67], we propose that the questions in both tables should be revisited and iteratively assessed over time. Together, Tables 3 and 4 provide a structured yet flexible resource to support the design, implementation, and continual refinement of interorganizational learning within digital transformation programs.

By addressing these 2 sets of questions, this study provides policy and practice tools and guidelines to enhance and accelerate the digital transformation of the health care sector. First, it identifies where and how governments and other intermediaries can support interorganizational knowledge sharing. Second, it highlights the formal and informal modes and internal organizational and individual factors that organizations can leverage as they navigate the transformation journey.

While this study advances the understanding of how learning ecosystems emerge and evolve across multiple levels, several limitations create opportunities for future research. First, our analysis focused predominantly on macroenvironmental and interorganizational factors, reflecting the design of the national digital transformation program and the nature of the available data. As a result, less attention was given to micro-level organizational routines and individual behaviors that may shape, or are shaped by, the learning ecosystem. Second, although our multicase design enabled rich comparative analysis across organizations, the scope of the study was limited primarily to hospital settings within a national health care system. Learning ecosystems may function differently across other care settings, sectors, or international contexts with different governance, market, or cultural dynamics. These limitations offer a valuable foundation for theoretical propositions that can guide future research aimed at advancing, refining, and testing the model developed in this study.

To guide such research, we propose a set of theoretically informed propositions. These propositions posit that cross-level contextual alignment is central to effective ecosystem formation; mechanisms, such as identity and relational recognition, significantly influence knowledge flows; the balance of formal and informal relationships affects ecosystem adaptability; and intermediaries are essential brokers in sustaining learning systems. Moreover, we propose that external influences shape early ecosystem trajectories, internal factors become more influential as ecosystems mature, learning outcomes vary with specific factors, and organizational- and individual-level factors warrant a deeper empirical investigation. Together, these propositions establish a clear agenda for advancing the theory on the emergence, evolution, and functioning of interorganizational learning ecosystems in digital transformation contexts.

Digital transformation has presented a significant challenge in the health care sector, which requires knowledge, experience, and expertise for organizational gains and new value creation. Interorganizational knowledge sharing can be seen as an enabler, supporter, and accelerator in assisting digital transformation at scale. Policymakers and organizations must proactively address the internal and external factors that impact knowledge sharing to harness its full potential in driving digital transformation. Understanding and evaluating approaches to creating a learning ecosystem is far from a trivial task and requires multilevel inquiries that follow actions within and beyond the organization in understanding this complex phenomenon.

Our framework illustrates how multilevel influencing factors shape a spectrum of interorganizational relationships and partnerships. The framework explores the dynamic nature of factors over time and therefore goes beyond shedding light on snapshots of the knowledge-sharing process by offering an evolutionary view of the process and its outcomes. By proposing this framework, we do not suggest that the domain of interorganizational knowledge sharing can be simply reduced to the aforementioned dimensions. Moreover, we do not propose this as a simple mechanistic account that could decouple different variables for the testing of modes of sharing and factors. Instead, we present a set of dimensions that are essential to explore in order to theorize a conceptual account of the learning ecosystem. We introduce this not just as an abstract framework but as a practical tool to support reflection and guide action for the various actors involved in shaping and sustaining it.