How lean routines shape a continuous improvement dynamic capability in hospitals Open Access

Hospitals worldwide are encountering major challenges as they strive to deliver efficient and high-quality care in an environment of escalating costs, resource scarcity, and increasing patient demands. In response, many hospitals have turned to adopting Lean practices, commonly used in a wide variety of industries, to continuously improve their operations and to enhance customer/patient outcomes (D'Andreamatteo et al., 2015). Despite Lean's positive potential, its implementation in hospitals can be challenging due to the complex, dynamic nature of hospital environments (Fillingham, 2007; Fournier et al., 2023; Thirumalai and Devaraj, 2023; Waring and Bishop, 2010). Hospitals are characterized by diverse professional disciplines, high patient variability, and rapidly evolving medical technologies. For example, hospitals must adapt continuously to fluctuating patient needs and advances in treatments, whilst also coordinating across multiple highly specialized departments such as emergency care, surgery, and outpatient services. This complexity is amplified further by the critical nature of decisions, requiring real-time collaboration between clinicians, nurses, and support staff. Understanding how Lean can evolve from isolated practices into a system-wide approach to continuous improvement (CI) is – both practically and scientifically – crucial (Burgess et al., 2025). In hospitals, this “processual shift” represents a pathway to sustainable operational excellence, where efficiency improvement and enhanced patient care become deeply embedded into their organizational routines (Sunder and Linderman, 2024). Academic research on those necessary improvements should start unpacking how such a shift occurs in practice. Exploring this process may offer valuable contributions to the dynamic capabilities (DCs) theory: by bridging the gap between micro-level Lean practices, organizational routines and their evolution into macro-level capabilities (Chen et al., 2023; Gutierrez et al., 2022).

We view Lean as a complex hospital-wide system aimed at improving the quality of care by understanding what is valuable for the patient, and involving the staff in a CI process (Leite et al., 2022; Secchi and Camuffo, 2016). Implementing Lean as a system goes beyond merely implementing a set of Lean practices (Camuffo and Poletto, 2023; Saabye et al., 2023). It is typically conceived as a process through which organizations seek, create, and store Lean knowledge to foster continuous learning (Secchi and Camuffo, 2016). This ongoing learning process aims to develop DCs that drive CI and enhance organizational adaptability (Anand et al., 2009; Furnival et al., 2019; Sunder and Linderman, 2024). Ultimately, DCs can enable widespread, continuous problem identification and resolution, allowing organizations to better serve customers and patients (Shah and Ward, 2007). More generally, a DC denotes an organization's strategic ability to renew its resources in its changing environment, to gain and maintain competitive advantages (Helfat and Peteraf, 2015). CI can be understood as a DC when it enables organizations to adapt and improve processes systematically in response to changing conditions (Furnival et al., 2019). Transitioning from early Lean initiatives to having an organization-wide continuous improvement dynamic capability (CI DC) is a challenge that is not yet understood well (Anand et al., 2009; Secchi et al., 2019).

Recent studies have examined the microfoundations of DCs – defined as the underlying routines, practices, and individual-level actions through which capabilities are enacted and evolve (Chen et al., 2023; Gutierrez et al., 2022; Sunder and Linderman, 2024). They highlight the importance of Lean practices for capability development but provide only partial insight into how such routines evolve across hierarchical levels in large hospital systems. In particular, existing research does not yet explain how Lean practices aggregate into higher-order capabilities that span organizational boundaries and how leaders and employees jointly shape this process (Csiki et al., 2023; Helfat and Campo-Rembado, 2016; Helfat and Martin, 2015).

To address this gap, we employed a comparative case study design (Caniato et al., 2018), combined with a four-year retrospective process study (Langley et al., 2013) comparing two Dutch university hospitals that implemented Lean through contrasting pathways – one top-down and one bottom-up. This is because Kim et al. (2014) highlight that these implementation pathways affect how routines are embedded across organizational levels, ultimately shaping the trajectory of capability formation. Our goal was to uncover how Lean practices ultimately evolve into a CI DC. Specifically, we answer the question: How do organizational routines interact at multiple organizational levels to develop CI as a DC, through Lean practices implementation, in large hospitals?

This study makes four contributions to deepen the microfoundations of DC theory. First, we conceptualize CI as a DC through presenting an empirically grounded model that traces the evolution from Lean practices to organizational routines and higher-order capabilities. Second, the study extends DC theory by demonstrating how CI routines bundle into four critical, interdependent capabilities – Improvement System, Collaborative Synergy, Integrated Accountability, and Learning-to-Learn – that together constitute a CI DC. Third, we reveal the multi-level engagement of leaders and employees in building and reinforcing these capabilities. Fourth, the study illuminates how different Lean implementation strategies (top-down versus bottom-up) shape the speed, scope, and integration of capability development in complex hospital settings. Together, these contributions deepen our understanding of how DCs emerge from Lean practices in knowledge-intensive organizations like hospitals. They offer practical guidance for hospitals seeking to establish a CI DC.

The remainder of this paper first outlines the theoretical background and research design, followed by a comparative analysis of two hospital cases. We then synthesize how Lean practices evolved into routines and capabilities, culminating in a new conceptual model explaining the formation of a CI DC. The paper ends with a discussion of the theoretical and practical implications and proposed directions for future research.

Hospitals are complex systems, with a dynamic environment characterized by a multitude of stakeholders, rapid changes, resource constraints, increasing demands, and advanced technological integration (Fournier et al., 2023; Powell et al., 2024). Implementing Lean in such settings extends beyond tool adoption; it demands transformation of routines related to decision-making and coordination of work (Azadegan et al., 2013; De Mast et al., 2021). Lean, when adopted as an iterative learning system (Ballé et al., 2019; Kristensen et al., 2022; Powell and Coughlan, 2020), enables hospitals to integrate CI into the organizational fabric. Camuffo and Poletto (2023) conceptualize Lean as a management system that evolves through strategic feedback loops, underscoring its capability-building potential.

Despite this, many hospitals take a fragmented Lean approach – through isolated projects rather than systemic transformations (Danese et al., 2018). Without embedding CI into shared routines, outcomes often remain suboptimal (Burgess and Radnor, 2012; Van Zyl-Cillié et al., 2024). Structural barriers – such as disengaged hierarchies and professional autonomy – further limit hospital-wide Lean implementation (De Souza and Pidd, 2011; Fournier and Jobin, 2018). For example, a disconnected hierarchy – where top management sets strategic goals without engaging middle managers or frontline staff—can lead to misaligned priorities and ineffective communication; it weakens Lean by breaking the feedback loops necessary for learning, adaptation, and CI. Sustainable success requires the implementation of Lean as a holistic system that transforms established routines and embeds CI as a core organizational capability (Burgess et al., 2025; Gutierrez-Gutierrez and Antony, 2020). When viewing Lean from such a holistic perspective, there is a lack of understanding of how Lean can be implemented effectively on a hospital-wide scale (Kim et al., 2014; Van Zyl-Cillié et al., 2024). For instance, an unresolved debate persists over whether hospital-wide Lean adoption, and the required routines and capability building, is best achieved through a top-down or bottom-up approach (Dannaphel et al., 2014; Van Beers et al., 2022; Van Elp et al., 2021). These contrasting pathways underscore the importance of examining how implementation strategies shape the mechanisms through which Lean can contribute to developing an organization-wide capability.

Organizational routines are repetitive, recognizable patterns of interdependent actions, that involve multiple actors (Becker, 2004; Feldman et al., 2016; Zollo and Winter, 2002). They comprise ostensive elements, which are the written rules and standard procedures that describe how things should be done, and performative elements, which are the actual actions people take when they carry out those routines in real situations (Biesenthal et al., 2019; Feldman and Pentland, 2003; Feldman et al., 2016). Through their recurrence and coordination, organizational routines bring operational stability (Nelson and Winter, 1982). Over time, feedback and learning enable these routines to evolve, align across functions, and facilitate smoother workflows, creating a platform for continuous adaptation (Zollo and Winter, 2002).

As routines interact and reinforce each other, they bundle into operational capabilities – structured sets of routines that allow organizations to perform activities effectively (Peng et al., 2008). When these capabilities are purposefully renewed and reconfigured to respond to change they evolve into DCs, defined as “the capacity of an organization to purposefully create, extend, or modify its resource base” (Helfat et al., 2007, p. 4). Such purposeful renewal allows organizations to continuously adapt to and sustain competitiveness by making their resources difficult to imitate (Pisano, 2017).

Teece (2007) categorizes DCs into three activity clusters: (1) Sensing: the identification and assessment of opportunities, challenges, and trends, (2) Seizing: the strategic allocation of resources and prioritization of initiatives to capture value from these opportunities, (3) Transforming: the continuous adaptation and reconfiguration of resources and processes to maintain a competitive edge. These activities do not occur in isolation; as routines evolve and integrate, they underpin the dynamic processes of sensing, seizing, and transforming, thereby enhancing organizational agility (Teece et al., 2016; Teece, 2007). In this view, routines constitute the microfoundations that enact and connect sensing, seizing, and transforming activities, enabling the continuous renewal of DCs.

While early DC literature has focused on the macro level, recent work emphasizes the importance of uncovering the microfoundations – the microscopic organizational origins through which DCs emerge from underlying routines, individual actions, and actor-level interactions (Chen et al., 2023; Csiki et al., 2023; Felin et al., 2012, 2015). Microfoundations thus bridge the gap between everyday practices and organization-level DCs, emphasizing how a DC emerges through individual cognition, leadership behavior, and social interactions within an organization (Felin et al., 2015; Helfat and Peteraf, 2015; Salvato and Rerup, 2011; Suddaby et al., 2020).

Building on Chen et al. (2023) who distinguish cognitive, behavioral, and relational microfoundations across individual, group, and organizational levels, recent research continues to highlight three unresolved issues in understanding how DCs develop. First, how specific microfoundations aggregate into organization-level DCs remains only partially addressed, despite growing attention to clusters of routines and system-level DCs (Gutierrez et al., 2022; Keller et al., 2022; Sailer et al., 2023; Sunder and Linderman, 2024). Second, limited empirical work clarifies managerial DCs: how leaders across hierarchical levels co-create and align these routines through their actions (Bojesson and Fundin, 2020; Helfat and Martin, 2015; Saabye et al., 2023). We do know, however, that informal social relationships and relational networks may act as key aggregation mechanisms, underscoring the potential of relational microfoundations in capability formation (Kowalski et al., 2024). Third, the organizational context (which for hospitals includes structural complexity, professional boundaries, and institutional norms) shapes how these mechanisms unfold, yet remains empirically underexamined (Chen et al., 2023).

Tracing how routines evolve and interact across multiple organizational levels in a hospital context may bring us closer to knowing how DCs actually develop. Lean implementation offers a particularly rich setting for examining this process, because Lean practices such as daily stand-ups, plan-do-check-act (PDCA) cycles, and Kaizen events can be viewed as structured behavioral mechanisms that embed improvement logic into everyday operations and routines (Furlan et al., 2019). Moreover, we know already that operational practices lead to DCs if interpreted, adapted, and routinized by actors across multiple levels (Chen et al., 2023). Building on these insights, we conceptualize DC development as a multi-level, evolutionary process in which Lean practices take shape as routines that are enacted, combined, and stabilized across time and organizational levels. The next section elaborates this view by examining CI as a DC in the context of Lean implementation.

CI is pivotal to the evolution of an organization's Lean system (Saabye et al., 2022) and acts as an evolving routine set that enhances adaptability (Galeazzo et al., 2017, 2021). From a DC theory perspective, CI routines serve to renew the organizational resource base (Gutierrez-Gutierrez and Antony, 2020; Gutierrez et al., 2022). Yet, few studies offer a detailed view on how these routines evolve and aggregate into CI as a DC (Anand et al., 2009; Biesenthal et al., 2019; Gutierrez-Gutierrez and Antony, 2020; Schilke et al., 2018). Drawing on Bessant et al. (2001) and Knol et al. (2019), the behavioral infrastructure of CI systems amount to eight CI routines (Table 1). These routines span the trajectory from early-stage Lean engagement to hospital-wide consolidation.

The emergence of CI routines can be understood as a multistage process that underpins the development of DCs within organizations (Anand et al., 2009). In the early stage of Lean implementation, organizations first develop an understand CI routine – where managers articulate the value of CI and identify operational inefficiencies (Bessant et al., 2001; Knol et al., 2019). This routine sets the stage for forming a CI habit, as employees begin to engage in CI practices, such as daily Kaizen Events and PDCA cycles (Franken et al., 2021) that bundle into an organizational routine (Galeazzo et al., 2024). Through a scientific approach to problem-solving, employees systematically identify root causes, critically assess alternative solutions, and implement those with the greatest potential impact (Galeazzo et al., 2024).

In the mid-stage, organizations extend improvement behavior beyond local settings. Align CI and focus CI routines help link local problem-solving to strategic priorities (Knol et al., 2022). Concurrently, shared CI and lead CI routines foster cross-departmental collaboration and leadership involvement, as managers shift from problem solvers to facilitators who coach teams and align improvement goals across units. This stage reflects the growing interdependence of routines and the emergence of leadership as a central aggregation mechanism—a key antecedent of capability formation identified in section 2.2.

In the late stage – when Lean is applied organization-wide – the improvement of improvement routine is built through the continuous enhancement of the overall improvement system, while the learning organization routine formalizes the sharing and consolidation of learning across all levels of the organization (Knol et al., 2019). At this stage, the improvement infrastructure transforms from a set of practices into a self-reinforcing system – an organization-wide capability of sensing inefficiencies, seizing improvement opportunities, and transforming operations (Anand et al., 2009).

This literature review integrates three interrelated domains – Lean implementation in hospitals, organizational routines and microfoundations, and the DC theory – thereby positioning the study at the intersection of Operations Management and Organizational Behavior. Despite important advances in understanding the role of Lean practices in DC development (Gutierrez et al., 2022; Sunder and Linderman, 2024; Sunder et al., 2023), and how CI routines evolve in smaller enterprises (Knol et al., 2022), the literature only partially explains how DCs emerge from Lean-driven CI routines across multiple organizational levels in large, complex, and knowledge-intensive settings like hospitals.

The four specific, interrelated gaps addressed in our empirical study are: First, how CI itself evolves to sense, seize, and transform, and thus becomes a DC. Second, the microfoundations of this CI DC: how Lean practices evolve into organizational routines, interact, and then bundle into higher-order capabilities. Third, the multi-level engagement of leaders and employees in enacting and reinforcing these routines and capabilities across the hierarchy. Fourth, how Lean implementation strategies (e.g. top-down versus bottom-up) influence the aggregation of routines into organization-wide capabilities and a DC.

Three main constructs guided our empirical analysis. First, CI routines that are defined as recurring, patterned practices enacted by frontline, middle management, and top management levels (e.g. understand CI, CI habit, align CI) (Bessant et al., 2001; Knol et al., 2019). Second, higher-order organizational capabilities that develop into stable patterns of collaboration, alignment, and accountability through the interaction and integration of CI routines across functions and hierarchical levels (Csiki et al., 2023). Third, CI as an organization-wide DC that unfolds through sensing, seizing, and transforming activities (Teece, 2007), once CI routines and organizational capabilities become sufficiently connected and reinforced over time. Sensing activities aim to detect inefficiencies and opportunities, seizing activities mobilize resources and structures to act on them, and transforming activities reconfigure the work and embed new practices (Teece, 2007). Together, these constructs underpin our abductive reasoning (Ketokivi and Choi, 2014) for tracing how Lean practices may develop into routines, capabilities, and ultimately a CI DC.

We adopted a comparative case study design (Caniato et al., 2018), combined with a retrospective process study (Van Dun et al., 2022) to investigate how organizational routines interact across hierarchical levels to develop a CI DC through Lean implementation in two large Dutch university hospitals. Process research enabled a multi-level analysis of interactions over time – spanning frontline, middle managers, and top – and their roles in shaping the CI DC. By tracing temporally embedded change episodes (Langley et al., 2013) over a four-year period, we examined how routines emerged, evolved, and bundled through iterative cross-level interactions.

The research is grounded in abductive logic (Ketokivi and Mantere, 2010; Shani et al., 2019) which allowed us to iteratively move between empirical observations and extant theory. This approach is particularly suitable when empirical findings not only confirm but also challenge or extend existing theory. Thus, the analysis was guided by the main constructs, outlined in the previous section: CI routines, capabilities, and DC (including sensing–seizing–transforming activities), while leaving space for theory extension.

We purposively selected two Dutch university hospitals that had embarked on organization-wide Lean implementation three years before the start of our research. Our paper focuses on both cases within the same four-year period (Eisenhardt, 2021; Yin, 2015). We deliberately selected hospitals with different approaches to Lean implementation to ensure a rigorous and balanced comparative analysis. This contrast was central to our theoretical interest in how different implementation pathways (top-down vs bottom-up) influence the development and integration of organizational routines into a DC (Kim et al., 2014). The first author initially accessed case 2 through prior Lean implementation consulting involvement. We then purposively selected the other case based on explicit criteria, including size, complexity, knowledge intensity, Lean maturity, and a contrasting implementation approach, following Eisenhardt (2021).

The hospital selection criteria were as follows. First, the hospitals had to be in the process of adopting an organization-wide Lean implementation, and had to have reached the “transition” stage or beyond (Netland and Ferdows, 2016). The transition stage in Lean maturity marks the shift from exploration to early exploitation, where initial problem-solving and best practice adoption drive rapid performance improvements and increasing employee buy-in (Netland and Ferdows, 2016). We thus verified their status by checking hospital documents related to the Lean implementation. Our within-case visualizations displayed in the results section show that at the start of our research both cases had reached at least the transition stage. Second, the cases needed to differ in their Lean implementation approaches (i.e. top-down versus bottom-up) in order to understand whether the Lean adoption method influences how organizational routines evolve into DC. Table 2 outlines the key case characteristics of both hospitals, revealing their factual similarities at the start of the study: as in the number of full-time equivalents (FTEs), departments, and annual patient admissions. Each hospital had commenced with Lean implementation three years prior to the study; by the end of this study's data-collection period, each hospital had completed a four-year implementation process.

We interviewed employees in a wide range of professional roles (Table 2), to capture a comprehensive understanding of the various actors' perspectives (Yin, 2015). Access to the interviewees was facilitated by a middle manager in each hospital who provided a complete list of employees involved in Lean implementation. Employees from this list were selected by us at random, curbing selection bias. Interviewee selection initially focused on those actively engaged in Lean practices; however additional participants were included based on departmental structure and job function to ensure representation of both supporters and skeptics of the implementation. All invited participants agreed to be interviewed (100% response rate).

The final interviewee sample consisted of executives (i.e. senior hospital leaders with strategic responsibilities), department heads (i.e. senior clinical and non-clinical managers overseeing departmental strategy), team leaders (i.e. operational managers directly overseeing daily team-level processes), employees (i.e. frontline clinical staff such as physicians and nurses involved in patient care delivery), staff (i.e. non-clinical personnel from supporting departments like finance, human resources, strategy, quality, and supply chain), and Lean consultants/experts (i.e. internal hospital personnel formally tasked with facilitating Lean initiatives and external consultants hired specifically for their expertise). Notably, case 1 employed both internal and external Lean experts, whereas case 2 relied exclusively on external consultants.

Demographically, the individuals in our sample were holding at least a bachelor's degree, with an overall male/female ratio of 60/40%. In case 1, several senior leaders with over 10 years of experience participated, while in case 2 the respondents were primarily middle managers and frontline staff, with tenures ranging from less than 5 years to over 15 years; the minimum tenure was 2 years, but most had substantially longer hospital work experience.

Finally, the varying number and length of interviews between the two cases reflected differences in the hospitals' Lean implementation approaches. Case 1's structured, top-down approach required engaging a greater number of individuals across various hierarchical levels, resulting in more interviews overall. In contrast, case 2's organic, bottom-up approach involved fewer formal roles but necessitated longer, more detailed conversations to capture complex informal interactions, contextual richness, and emergent routines.

Data were collected retrospectively, spanning a four-year period, see Figure 1. The data-collection process employed various methods – semi-structured interviews, participant observations, and archival documents including key performance indicator (KPI) data – each tailored to capture distinct aspects of Lean implementation and CI routine development. We consulted the ethics departments of both hospitals, which determined that no formal ethics approval was required since the research did not involve patient data or interventions directly affecting patient care.

At T4, we began with four open-ended interviews with internal Lean experts (two per hospital) to obtain an overview of the Lean journey. These interviews confirmed the differences in each hospital's Lean implementation approach and outcomes: case 1 exhibited an orchestrated, hospital-wide rollout, whereas case 2 demonstrated a more fragmented, unit/department-oriented approach.

In addition, 48 one-hour semi-structured interviews were conducted with key employees with various roles (see Table 2). The interview guide (see  Appendix 1) was developed based on theoretical constructs related to Lean, CI routines, and DCs, and was pre-tested on the Lean experts and an independent researcher to ensure clarity and relevance (Yin, 2015). Topics covered during the interviews included the hospital's Lean strategy, Lean implementation approach, Lean practices, cross-functional collaboration, CI routines and DC, leadership involvement, and how they measured success and performance outcomes. Open-ended questions – for instance, “How do different departments collaborate on Lean initiatives?” and “Could you describe how problem-solving activities are integrated into your daily work?” – helped mitigate confirmation bias while capturing rich, informant-centered data. Interviewees provided candid feedback, as illustrated by the finance staff in case 1 questioning the financial returns of their Lean journey, and the medical leaders in case 2 criticizing the hospital's lack of a holistic approach. This confirms that we had interviewed not only proponents but also critics.

All interviews were audiotaped and transcribed (298 pages for case 1 and 364 pages for case 2). Each transcript was then shared with each interviewee to verify a transcript's accuracy. This led to small corrections in some of the transcripts.

During the interviews, we learned about the stand-up meetings and their central role in the Lean journey of both hospitals. Therefore, after the interviews, thirteen on-site observations of stand-up meetings were conducted using a structured observation framework (Arumugam et al., 2012). In case 1, these observations included five frontline stand-up meetings and three cross-departmental meetings. Case 2 entailed only frontline stand-ups ^[1] (Taher et al., 2016). Each meeting was chaired by a previously interviewed team leader, and the observations focused on real-time leader-employee interactions, problem-solving approaches, and the enactment and adaptation of CI routines. We documented various behaviors and engagement patterns related to CI organizational routines. Specific observational measures included the quality of team collaborations, the nature of interdisciplinary interactions, the types of problem-solving approaches employed, the degree of behavioral engagement in adapting routines, and the extent of initiative-taking by the participants. Detailed field notes recorded the practices related to meeting objectives, daily management processes, and actual process improvements, with a particular focus on how leaders and staff addressed Lean implementation challenges. The observational data were triangulated with the interview findings and archival documents to ensure consistency and robustness in the interpretation of the CI routines and their adaptation in practice (Czarniawska, 2008).

We collected as many hospital-based Lean documents as we could. For case 1, we analyzed 58 documents and, for case 2, 47 documents. These documents included, among others, Lean training materials, Lean practices, presentations, monitoring, and implementation progress data, as well as descriptions of organizational structures and planning and control processes. Publicly available annual reports of both hospitals were also downloaded: to analyze organizational-level and strategic Lean-related changes, and their reported impact over the prior four years.

Finally, we collected archival KPIs data from each case over the entire four-year period. These KPIs related to productivity, patient care quality, efficiency, employee satisfaction, and cost reductions. The interviewees provided internal documents used for KPI reporting and monitoring. In both cases, the frontline and cross-departmental level KPIs were reported daily, weekly, or monthly, depending on the KPI; consolidated strategic level KPI data was only available for case 1 given that case 2 did not monitor a strategic, hospital-wide KPI. Hence, the way performance was measured differed between both cases, reflecting differences in terms of how Lean was adopted top-down versus bottom-up.

Following Gioia (2020) we applied a five-step iterative process to move from first-order, informant-centered codes to second-order, theory-informed constructs, and ultimately to an integrated model of the evolution of a CI DC.

Using ATLAS.ti we coded all interview transcripts to generate first-order codes capturing Lean practices, actions, and sequences within each hospital (Eisenhardt, 1989; Eisenhardt and Graebner, 2007). We applied open coding using informant terms (e.g. alignment of objectives at middle management associated with the quote “Goal-setting sessions now involve managers across levels” and accelerated frontline problem-solving associated with the quote “problems are now raised and solved directly on the work floor”).  Appendix 2 offers illustrative quotes related to the first-order codes. We triangulated these codes with archival documents (e.g. Lean presentations, strategy decks, internal reports) and observational field notes to build a chronological narrative of each case's Lean journey (Langley, 1999; Langley et al., 2013). To validate these reconstructions we conducted one-hour sessions with three key informants per hospital: a Lean consultant, division director, and middle manager (Yin, 2015). During these sessions we jointly reviewed the narratives and clarified whether the timelines reflected key turning points, challenges, and in situ context.

To make each hospital's Lean implementation process explicit, we segmented each narrative into eight episodes – preparation, initiation, pilot, develop, evaluate, scaling, infrastructure, and hospital-wide integration (Van Beers et al., 2022). This temporal bracketing provided the scaffold for tracing how routines emerged, stabilized, and shifted over time (Langley et al., 2013).

First-order codes were grouped into second-order themes aligned with established CI routines (e.g. understand CI, CI habit, align CI) (Bessant et al., 2001; Henrique and Godinho Filho, 2020; Knol et al., 2019, 2022). We coded a practice as a routine when interviewees described repeatable, patterned activities that were embedded in work practices and sustained over time. To capture vertical diffusion of each routine, we tagged each routine by organizational level (frontline, middle managers, and top). Coding and grouping decisions were iteratively reconciled in joint author sessions until consensus: when empirical patterns did not fully align with existing theoretical labels, we revisited and refined the coding scheme. For example, lead CI was reinterpreted not only as leadership behavior but as a routinized coaching practice in which leaders empowered teams through structured facilitation. Also, we observed that middle managers often collaborated cross-functionally, engaging frontline leaders in improvement initiatives. This insight led us to add cross-functional interactions into our analysis. Such abductive refinements ensured that emerging constructs were co-developed rather than retrofitted to theory (Ketokivi and Mantere, 2010).

We compared the routines and performance improvements that were achieved across both hospitals – temporally bracketed by the Lean implementation stages – to identify similarities and differences. A capability was identified when a cluster of routines (1) spanned multiple organizational levels, (2) supported cross-functional or hospital-wide change, and (3) contributed to sustained performance improvements (Csiki et al., 2023). This analysis yielded four emergent organizational capabilities (e.g. Improvement System and Collaborative Synergy), which constitute the aggregate dimensions shown in Figure 2.

In the final step, we interpreted these four empirically derived capabilities through the lens of DC theory (Teece, 2007). Each capability was mapped onto the three core DC activities: sensing, seizing, and transforming (Teece, 2007). This integrative step connected empirical findings with theoretical concepts, resulting in a new conceptual model of the evolution of a CI DC; by integrating different theoretical streams into a (microfoundational) process model of CI DC development we clearly extend existing theory (i.e. abduction). The model was refined iteratively through author discussions, ensuring that theoretical abstraction remained tightly grounded in the data (Ketokivi and Mantere, 2010).

This section presents the findings based on two in-depth case studies: Case 1 implemented Lean through a top-down, organization-wide strategic deployment, whereas case 2 adopted an emergent, bottom-up approach initiated at the departmental level. We first present a detailed within-case narrative for each case, documenting the unfolding Lean implementation and the emergence of specific CI routines over time whereby each episode is identified in bold. The cross-case comparison then identifies common patterns into four overarching capabilities. Finally, the synthesis sketches how a CI DC evolved over time.

In case 1 Lean was introduced as a hospital-wide strategic initiative, led by top management. The transformation was centrally coordinated, supported by structured training, governance structures, and performance monitoring. Figure 3 maps the progress from isolated lean practices to coherent CI routines which evolved into higher-order capabilities. The following case analysis traces this developmental trajectory. Lean was initiated by the executive board as the hospital's operational strategy (episode 1 – Prepare lean). A top manager noted: “Lean is our implementation strategy to implement our strategic objectives”. During this early stage, a rollout plan was developed by the Lean director who was appointed by the top management. The goal was to align leadership and initiate early-stage lead CI and CI focus routines, establishing managerial engagement and strategic prioritization of Lean implementation.

Lean practices (such as training, Kaizen events, value stream mapping (VSM), and PDCA cycles) were introduced across pilot departments, activating understand CI routines (episode 2 – Pilots). These early pilots functioned as safe learning environments, allowing teams to experiment and adapt Lean methods to their specific contexts.

Coaching and Gemba walks, facilitated by Lean experts, initiated lead CI routines started to reshape the role of frontline leaders, from controlling subordinates to gradually facilitate Lean problem-solving (episode 3 – Develop leaders). As a nurse explained “Our team leader is now much more involved through daily Gemba walks.”

After one year, pilot results were evaluated by top management, and a revised rollout strategy was formulated (episode 4 – Evaluate), embedding goal-setting practices with aligned KPIs. This evaluation provided crucial feedback for standardizing successful practices and adjusting underperforming areas A middle manager reflected: “We now run goal-setting sessions across managerial levels”.

The rollout extended to ten departments, integrating goal setting and KPIs into frontline Kaizen, fostering greater accountability in daily operations (episode 5 – Scale). During Kaizen Events, staff increasingly identified and resolved issues, forming a sustained CI habit. A Lean coach remarked: “Employees have taken a big share in highlighting problems and taking responsibility to solve them … they feel proud when they succeed.” Frontline leaders further evolved from problem solvers to facilitators, empowering staff through coaching and questioning. A physician noted: “Problems are now discussed directly on the work floor and often solved the same day.” Gemba Walks reinforced lead CI routines, with frontline leaders offering real-time coaching and maintaining daily visibility. The bundling of understand CI, CI habit, focus CI, and lead CI routines solidified the Improvement System capability.

As implementation spread throughout the hospital, focus CI, CI habit, and lead CI routines emerged across departments (episode 6 – Cross-functional). Interdepartmental Kaizen Events and cross-functional Gemba Walks identified process misalignments and functional silo barriers. One middle manager reflected: “Before initiating the Gemba Walks, we lacked a clear view of our processes and bottlenecks”. In response, frontline CI practices were scaled up to the department level. Standardized tools – A3 reports, visual boards, and shared KPIs—supported align CI routines. Weekly stand-ups enabled middle management to align objectives and extend CI practices to lagging departments. One senior nurse shared: “A multidisciplinary improvement event [A3], including a medical specialist, a doctor assistant, a department head and two nurses from different departments, was started”. These developments embedded a Collaborative Synergy capability and enabled measurable improvements – e.g. lower medication errors and faster throughput times (Table 3). A manager remarked: “We now manage several departments with one set of standard KPIs”.

To sustain momentum, a formal CI infrastructure was introduced (episode 7 – Infrastructure). Middle management led the implementation of standardized methods (standups, PDCA, A3 and visual tracking), which served as a shared language for daily operations. Weekly stand-ups – initially introduced in the cross-functional episode – became embedded in the broader CI infrastructure. This system supported horizontal and vertical coordination, strengthening feedback loops and thus reinforced shared problem-solving.

The final phase of Lean implementation extended CI routines to top management level (episode 8 – Hospital-wide). A hospital-wide PDCA rhythm was introduced, translated into strategic KPIs such as “pain reduction”. A divisional director explained: “Supervisors have stand-up sessions with frontline staff. I meet twice a week with my peers to discuss KPIs and bottlenecks, and we meet weekly with the board for strategic indicators”. This cascaded structure reinforced focus CI routines at all levels, embedding integrated accountability. Finally, the formalization of structured reflection and performance loops – through hospital-wide strategic PDCA cycles – led to the emergence of the improvement of improvement routine. This routine marked a transition from isolated problem-solving efforts to a deliberate, system-wide process of refining the CI system itself. Within six months, this multi-level learning infrastructure enabled the hospital to achieve its primary strategic CI objective – a 72% reduction in post-surgical pain (Table 3).

The combination of multi-level stand-up meetings, strategic PDCA cycles, and structured performance reviews established a collective Learning-to-Learn capability, where CI was no longer limited to operational fixes, and better coordination, but – due to their successes – became also a strategic vehicle for continuous organizational adaptation. Through these Lean practices, employees were not only engaged in solving current problems but also actively involved in evaluating and refining the improvement system itself, developing a learning-oriented culture. As a middle manager described: “Every morning we gather [during stand-ups] with all medical specialists, all nursing teams, and managers at all levels to discuss how to manage the acute patient admissions. We use this information in our improvement team to improve the flow of our intake. The results are visible: we have a thousand more intakes than last year, and our aim is no refusal at the gate.” Altogether, as evidenced by the archival performance data, case 1 showed substantial performance improvement across departments and hospital-wide: patient throughput time decreased by 23%, productivity rose 16%, surgical volume increased 28%, and medication errors dropped 99%, while hospital-wide postoperative pain was reduced 72%. Together this indicates strong operational and clinical impact, driven by the Lean implementation.

In case 2, Lean was initiated through a bottom-up approach, guided by an external Lean expert who reported to divisional management. The implementation began in a few nursing wards and gradually expanded across departments. Lean was introduced with the objective to improve ward efficiency by 10% (episode 1 – Prepare Lean). Figure 4 visualizes how Lean practices evolved over time into CI routines and emerging capabilities. The following traces this four-year journey through eight sequential episodes.

Lean practices – including standups, Kaizen Events, PDCA cycles, VSM, and Yellow Belt training – were introduced across three nursing wards (episode 2 – Pilots). These efforts led to the emergence of understand CI and CI habit routines. Nurses became active problem solvers, and weekly stand-ups with improvement boards chaired by frontline leaders supported lead CI routines. One nursing frontline leader observed: “The improvement board, along with weekly Kaizen Events, makes it possible to highlight all problems.”

In these early Lean stages, the Lean expert deepened lead CI routines by coaching frontline leaders and introducing Lean practices such as Gemba Walks and leader standard work (episode 3 – Develop leaders). Weekly improvement sessions and hands-on involvement of ward leaders further deepened frontline engagement and stabilized routines. As indicated by a senior nurse: “Team leaders make work easier and show they listen to us.”

As Lean practices matured, an internal evaluation by divisional management led to the integration of KPIs into the wards (episode 4 – Evaluate). “We created an improvement board with four categories: patient, safety, employee, and finance”, explained one middle manager. These efforts helped the evolution of focus CI routines, giving more structure to localized improvements. However, learning remained local and limited, as results were not shared beyond the initiating nursing units.

Encouraged by the early successes, Lean practices were replicated in other nursing wards (episode 5 – Scale). Interdisciplinary Kaizen Events and cross-functional Gemba Walks were introduced. Goal setting at department level further embedded focus CI, and lead CI routines remained active. However, improvements continued to rely heavily on individual departments and enthusiastic frontline leaders. CI habit routines evolved but only in three units. Cross-functional physician involvement in nursing Kaizen Events briefly enhanced feedback and collaboration. However, as clearly articulated by a quality staff member: “Cross-boundary improvements remain difficult due to rigid organizational structures and actors beyond control.”

Success of Lean in nursing wards led to the start of Lean implementation in other departments (episode 6 – Cross-functional). While understand CI and CI habit routines successfully emerged in the newly involved departments, the absence of align CI and shared CI routines hindered cross-functional coordination and strategic alignment. As one division head admitted, at that point: “Our strategic objectives are not linked to Lean”. Only three of 34 departments continued Kaizen Events. Without middle management alignment or standardized tools, the Collaborative Synergy capability did not materialize. Efforts to scale and connect Lean beyond individual operation-level departments stalled.

In response to this stalled progress and fragmented implementation, the human resource department launched Lean Green Belt training for frontline and middle managers (episode 7 – Infrastructure). This initiative re-energized understanding CI and lead CI routines. However, the focus remained on developing individual skills rather than aligning these routines at the organizational level. No formal mechanisms were put in place to coordinate CI efforts across departments.

Recognizing the limits of the bottom-up approach, top management eventually launched a hospital-wide strategic CI initiative and established a central Lean office (episode 8 – Hospital-wide). The participation of the top-100 leaders in the strategic CI initiative signaled growing executive commitment and reintroduced lead CI routines at senior levels.

Across these eight episodes, Case 2 illustrates the opportunities and constraints of an emergent, bottom-up Lean implementation. Case 2 demonstrated notable performance improvements at the interdepartmental level – including 10% cost reductions and −30% surgical waiting times, 13% increased bed occupancy, enhanced patient satisfaction (+40%) and nurse satisfaction (+13%), and 12% fewer sterilization errors (Table 3). Although early engagement and local ownership fostered quick wins, the lack of deliberate orchestration, cross-level alignment, and integrative infrastructure constrained the evolution of CI routines into higher-order capabilities and hospital-wide results.

Both hospitals followed divergent paths in developing CI routines and a DC. Case 1 pursued a top-down, strategically coordinated Lean transformation, combined with bottom-up frontline execution, while case 2 adopted a bottom-up, emergent approach initiated within a nursing division. Although both hospitals deployed similar Lean practices (e.g. Kaizen, PDCA, Gemba Walks, stand-ups), their implementation trajectories – and the extent to which CI routines evolved into higher-order capabilities – differed markedly.

Case 1 invested substantial time in designing and aligning its Lean initiatives, dedicating approximately six months to preparing a standardized framework prior to implementation. This deliberate preparation phase created the foundation for learning and capability building across departments. In contrast, Case 2 commenced implementation almost immediately, emphasizing rapid execution over a planned setup. While this approach accelerated early improvements at the departmental level, it constrained the extent of hospital-wide learning and integration. The performance outcomes in Table 3 reflect these differences: Case 1 achieved both interdepartmental and hospital-wide results, whereas Case 2's improvements remained localized, without measurable hospital-wide benefits.

We also compared both cases across the four emergent capabilities identified in our analysis. Each capability represents a bundle of routines, that becomes embedded, coordinated, and sustained over time (Table 4).

In both cases, the Improvement System capability emerged from early CI routines – understand CI, CI habit, focus CI, and lead CI. Case 1 installed these routines through a hospital-wide formal training, Kaizen events, leader standard work, and coaching. PDCA cycles and daily stand-ups reinforced routine adoption across all levels, enabling consistent improvement behaviors. In case 2, similar routines developed in pilot wards under the guidance of a Lean expert. However, lacking a coordinated scaling strategy, these routines remained localized, and PDCA use was inconsistent, preventing system-wide integration and, hence, limited results (see, Table 4).

Collaborative Synergy in case 1 was built through the layering of the Improvement System capability, through align CI and shared CI routines, supported by structured cross-functional Kaizen Events, A3 problem-solving, Gemba Walks, and aligned KPIs. These practices enabled coordinated problem-solving and shared understanding across departments and hierarchical levels. In case 2, collaboration remained informal and largely confined to nurse-physician interactions. The absence of align CI routines, shared metrics, and standard Lean practices such as Kaizen prevented the development of cross-functional coordination, leaving the old, fragmented organizational coordination routines largely in place without huge performance gains.

In case 1, Integrated Accountability developed through the hierarchical layering of its Improvement System, Collaborative Synergy, and a hospital-wide CI focus. Cascading KPIs, scorecards, daily stand-ups, and shared performance reviews helped to align strategic goals with operational activities, ensuring ownership across organizational levels. Case 2 lacked such integrative accountability structures. While HR-led Green Belt training raised individual awareness, the absence of shared KPIs or coordination mechanisms meant accountability remained confined to single departments and disconnected from broader strategic objectives.

Finally, Learning-to-Learn was explicitly embedded in case 1 through strategic PDCA cycles, visual performance tracking, and multi-level reflection practices. Cascaded stand-ups, KPI reviews, and structured CI reviews enabled feedback loops across all organizational levels, allowing teams to adapt routines and improve the Improvement System itself. In case 2, however, learning remained local and episodic. While some reflection occurred within pilot wards, the absence of systematic PDCA rhythms or shared learning structures precluded the emergence of an organization-wide learning capability.

The contrasting cases reveal two distinct pathways for capability development. The top-down case demonstrates how deliberate orchestration, leadership engagement, and governance structures enable routines to integrate vertically and horizontally, allowing the four capabilities to mutually reinforce one another and evolve into a CI DC. The bottom-up case highlights the limits of emergent, locally driven improvement: although engagement and experimentation were high, the lack of integration across functional boundaries barred capability scaling. Together, these findings indicate that CI DC development depends on the presence of Lean-based routines and on their integration across departmental boundaries.

From our comparison of both cases, the organizational development of a CI DC unfolds through an iterative, multi-level process in which CI routines are progressively developed, stabilized, scaled, and reconfigured. This evolution is initiated through a hybrid model, combining top-down strategic intent with bottom-up experimentation. Early preparation stages, such as top management commitment to Lean, Lean coordination structures, and the design of Lean practices, serve to orient the organization toward CI as a strategic priority.

At the frontline, routines such as understand CI, CI habit, lead CI, and focus CI enable sensing by allowing teams to detect inefficiencies and performance gaps in real time through stand-ups, visual boards, and daily PDCA cycles. Seizing occurs as frontline teams develop local structures—e.g. CI boards, Kaizen practices, and coaching – that allow them to prioritize problems, gather responses, and enable team-based experimentation. Over time, as these behaviors become routinized and refined, they lead to a DC related transforming activity: teams reshaping how work is organized and how improvement is embedded in everyday operations. This process gives rise to an Improvement System – a set of repeatable, self-sustaining routines that stabilize improvement, learning, and performance enhancement.

As CI practices expand across functions, cross-functional sensing emerges through lead CI and align CI routines. These routines help teams identify interdependencies and inefficiencies that transcend departmental boundaries. Seizing at this level involves not only mobilizing actors but doing so around a shared improvement logic – facilitated by A3 problem-solving, cross-functional Kaizen Events, shared CI boards, and aligned KPIs. These mechanisms establish shared problem-solving routines, which enable multidisciplinary teams to work on interconnected challenges through a common structure and language. In doing so, the organization develops a Collaborative Synergy capability that allows actors from different units to jointly define priorities, co-create solutions, and take coordinated action. Transforming becomes visible here as routines are no longer confined to departments, but reconfigured to support horizontal and vertical collaboration, enabling the shift from isolated improvements to interdependent systems of continuous alignment and collective learning.

At the hospital-wide level, align CI and shared CI routines become mechanisms that support transforming by changing how improvement is managed and coordinated across different levels of the organization. These routines are embedded into performance scorecards, cascading objectives, multi-level stand-ups, and cross-level review cycles. Rather than relying on top-down directives, these structures translate strategic goals into operational behaviors and create alignment through routine-based coordination. This institutionalizes Integrated Accountability, a capability whereby performance expectations, strategic priorities, and improvement activities are aligned across organizational levels/units, through shared metrics, structured routines, and multi-level ownership.

Finally, sensing, seizing, and transforming come together in the development of Learning-to-Learn: A meta-capability through which organizations continuously refine their improvement processes by embedding feedback, reflection, and adaptation mechanisms that enhance the effectiveness of CI routines over time. Organizations (or rather its employees) do sense when performance stagnates or improvement efforts lose momentum. They can then seize performance-enhancing opportunities by adjusting their CI structures and priorities. This, in turn, allows transformation by changing how improvement is planned and governed. Practices like strategic PDCA cycles, tiered reviews, and flexible CI governance help the system evolve over time. This reflexive capacity enables organizations not just to improve continuously, but to improve the way they improve.

Figure 5 visualizes this process: routines first emerge in local practice, then scale through structure, coordination, and adaptation. When reinforced across organizational levels and units through sensing, seizing, and transforming, they culminate in four interdependent capabilities—Improvement System, Collaborative Synergy, Integrated Accountability, and Learning-to-Learn. Together, these constitute a CI DC that enables organizations to sustain performance improvement while constantly reinventing their own improvement infrastructure.

This four-year retrospective process study compared two Dutch university-based hospitals to deepen our understanding how organizational routines interact at multiple organizational levels to develop CI as a DC. The findings contribute to theory in four key areas: (1) conceptualizing CI as a DC through an empirically grounded model; (2) extending the microfoundations literature by explaining how Lean practices evolve into routines, capabilities, and ultimately a CI DC; (3) revealing the multi-level engagement of leadership and employees in building a CI DC; and (4) illuminating how different Lean implementation strategies affect capability development in complex hospital settings.

Having developed an empirically grounded conceptual model, this study contributes to the theorization of CI as a DC for complex organizational systems. Building on the routine-based view (Anand et al., 2009; Bessant et al., 2001; Gutierrez-Gutierrez and Antony, 2020; Gutierrez et al., 2022; Knol et al., 2019, 2022; Peng et al., 2008), we trace how Lean practices are initiated (Komkowski et al., 2024) and evolve into CI routines and then bundle into four higher-order capabilities – Improvement System, Collaborative Synergy, Integrated Accountability, and Learning-to-Learn – that together form a CI DC.

These layered capabilities reinforce one another and collectively enable the entire organization to sense, seize, and transform, fulfilling the defining criteria of DCs (Teece, 2007), thereby enabling strategic adaptability and sustained performance. Hence, the CI DC aligns with Schilke et al.’s (2018) view of DCs as mechanisms for adapting, renewing, and reconfiguring an organization's resource base to maintain competitive advantage.

This cumulative process departs from existing work in small to medium-sized enterprises contexts (Knol et al., 2019). Here we demonstrate how work routines develop and interact within a large, knowledge-intensive hospital setting. Our findings echo De Jager et al. (2004) and Eldor (2020) who emphasized that a system-wide focus on improvement is a critical routine with which to align all within-organizational actors to build urgency for and vital power for gradual organizational change.

Moreover, the conceptual model resonates with Sunder and Linderman (2024) operational excellence DC framework; it extends it by offering a richer account of how organizational routines and capabilities dynamically co-evolve in complex work environments. By showing how CI routines scale into a system-wide capability, our work responds to calls for deeper theorization of routine aggregation processes (Keller et al., 2022) and contributes a hospital-based extension to earlier models.

Our findings extend the microfoundations perspective of DCs by showing how Lean practices (e.g. PDCA cycles, Kaizen events, daily stand-ups, A3 thinking) function as mechanisms that embed CI logic into day-to-day routines (Furlan et al., 2019; Kristensen et al., 2022). We show how these routines are interpreted, adapted, and routinized across levels/units, ultimately coalescing into organizing capabilities extending the research on routine aggregation (Chen et al., 2023; Csiki et al., 2023; Keller et al., 2022; Momeni et al., 2023; Peng et al., 2008).

The development of the collaborative synergy capability, in particular, illustrates the importance of relational microfoundations (Chen et al., 2023; Kowalski et al., 2024) between departmental and across hierarchical levels – key for organizational learning and adaptation (Helfat and Campo-Rembado, 2016; Sailer et al., 2023). This capability mitigates structural fragmentation in hospital settings, promoting effective cross-functional communication and collaboration, enabling more effective cross-functional and cross-departmental coordination and communication (De Souza and Pidd, 2011; Fournier and Jobin, 2018; Van Beers et al., 2022).

The emergence of the Integrated Accountability capability reveals both structural and behavioral alignment mechanisms that resonate with existing policy deployment frameworks (Witcher et al., 2008; Giordani da Silveira et al., 2017). However, our findings extend this literature by highlighting how alignment is not only top-down but also driven through middle management and bottom-up dynamics (Van Beers et al., 2022).

Lastly, the Learning-to-Learn capability reflects individual and group learning processes akin to those described in the learning organization and DC literature (Margherita and Braccini, 2024; Saabye et al., 2022). It introduces a self-reinforcing loop of experimentation and refinement – where not only improvement problems but also the improvement system itself becomes an object of inquiry and adaptation. This dynamic, recursive learning architecture exemplifies the advanced stage of DC development. Unlike prior studies that primarily focused on routine evolution in isolation (Sailer et al., 2023), our findings highlight the interdependence and dynamic interplay of routines across different organizational levels/units, reinforcing the idea that DC are deeply embedded in the interactions between a great variety of employees and organizational processes (Nayak et al., 2020).

This study contributes also to the emerging body of work emphasizing multi-level agency in the formation of DCs (Helfat and Martin, 2015; Pitelis and Wagner, 2019). Our findings show that DCs are not solely orchestrated and role-modeled by top management but are enacted and stabilized through interdependent actions across all organizational levels. Rather than viewing DCs as the product of mere strategic decision-making at the apex of the organization (Helfat and Peteraf, 2015; Winter, 2013), we demonstrate how frontline employees, middle managers, and top managers collaboratively co-create a CI DC. Each level plays a distinct yet complementary role: frontline actors embed CI into daily work, middle managers interpret and align these routines across units, and top managers provide strategic support, including governance structures.

Hospitals are complex, knowledge-intensive environments where rigid hierarchies and departmental silos often hinder change (De Souza and Pidd, 2011; Fournier and Jobin, 2018). In this context, our results indicate that neither purely top-down nor purely bottom-up approaches are sufficient to support the emergence of a scalable CI DC. Instead, a hybrid strategy – characterized by strategic guidance from top management, a centrally supported improvement infrastructure, and iterative feedback loops among middle managers and frontline teams – proves most effective (Kim et al., 2014; Van Beers et al., 2022).

Such hybrid approaches enable the alignment of local improvement efforts with broader organizational goals, allowing Lean practices to scale into hospital-wide capabilities (Anand et al., 2009). As such, organizations implementing Lean are most likely to develop features of a learning organization (Burgess et al., 2025; Kristensen et al., 2022; Tortorella et al., 2021). This echoes the Lean deployment frameworks proposed by Sunder et al. (2023) and the sustainable implementation roadmap advanced by Van Zyl-Cillié et al. (2024). Key to this alignment is the active engagement of all middle managers and their capacity to translate strategic objectives into context-specific routines, bridging the top-down and bottom-up dynamics necessary for sustained improvement (Heyden et al., 2017, 2018; Reynders et al., 2020).

This perspective challenges the idea that purely top-down strategies are better for organizations in the early stages of CI adoption, as suggested by Galeazzo et al. (2021). Grassroots engagement as a driver of organization-wide CI appears equally important (Secchi and Camuffo, 2016). Thus, empowering frontline staff to initiate improvement efforts, supported by managerial direction, recognition, and facilitation, creates a reinforcing system of CI capability development (Bessant et al., 2001; Nonaka et al., 2016). Our findings on how to build a sustainable CI DC in hospitals show the need for a well-coordinated yet flexible plan, leadership involvement at multiple levels and units as well as enablement of learning mechanisms for all its employees.

Our study offers actionable insights for hospital leaders, healthcare professionals, and consultants involved in Lean implementation. Figure 4 provides a staged roadmap for developing and sustaining a CI DC across organizational level in hospitals. At the frontline, it is advised to begin with tailored Lean and CI training, daily stand-ups, and visual management to develop CI awareness and habits. Middle managers must facilitate cross-departmental alignment through Gemba Walks, standardized templates for CI and reporting, and shared KPIs, shifting from overseeing local initiatives to coordinating hospital-wide improvement. Meanwhile, top management must ensure alignment of department-level initiatives with hospital strategy by establishing cascading KPIs and embedding accountability through dialog-based performance reviews. Altogether, this layered involvement of actors throughout the hospital ensures that cross-functional frontline problem-solving, managerial coordination, and strategic direction converge into a robust CI DC that drives sustained (performance) improvement.

This study is limited to two Dutch university hospitals; findings may not generalize to other national or regulatory contexts. Future research should examine a broader range of hospitals – public/private, large/small, rural/urban – and include external influences such as national policies, market competition, and technological change. Furthermore, while our study focused on the change in internal hospital routines, our findings might not generalize fully to contexts experiencing greater external turbulence such as during COVID-19 (Burgess et al., 2025) or (major) national healthcare system regulatory changes.

Further work should also refine and test the conceptual model by evaluating how specific capabilities and routines contribute to CI DC formation in different contexts (e.g. manufacturing or smaller service organizations). Although our study found links among the four capabilities, the strength of these linkages needs warranting in terms of how they interact and feed each other. Internal factors such as leadership style, resources, organizational history, and hospital size may also shape capability development.

Finally, integrating institutional theory in the study of DCs could clarify how external pressures interact with internal routines to sustain Lean practices over time. The interplay between organizational routines and institutional pressures shapes the sustainability and adoption of Lean practices. Routine theory emphasizes the role of recurring organizational patterns in embedding new practices, while institutional theory highlights the influence of external norms and cultural pressures on internal processes (Slaghuis et al., 2011).

This study shows how Lean practices evolve into organizational routines that bundle into four interdependent capabilities – Improvement System, Collaborative Synergy, Integrated Accountability, and Learning-to-Learn – which together form a hospital-wide CI DC. These capabilities at different hierarchical levels enable continuous sensing, seizing, and transforming, thereby embedding improvement into daily hospital operations and strategic processes and contributing to performance improvements. While exploratory, the proposed model provides a foundation for future studies on how DCs develop in large, knowledge-intensive organizations.

1. Can you please describe your role and years of experience in the hospital?

2. How long have you been involved in Lean or CI initiatives?

3. What does Lean mean to you, your daily work, the patient, and the hospital?

1. How was Lean introduced in the hospital and whose initiative was it?

2. How has Lean been aligned with the hospital's overall strategy?

3. How were the hospital board and senior management involved in Lean implementation, and what role did physicians, nurses, and administrative staff play in the process?

1. Can you describe the key steps in Lean implementation in your department/hospital?

2. Which Lean practices were introduced, and were these practices standardized across departments, or did they evolve differently? (if needed, offer examples like: daily stand-ups, Kaizen Events, Gemba Walks)

3. How were employees/managers trained and coached in Lean?

4. How do employees become engaged in CI?

5. How have Lean practices changed daily work processes?

6. Can you provide an example of how a Lean practice led to a significant process improvement?

1. How do different departments collaborate on Lean initiatives?

2. Have you observed cross-functional problem-solving becoming more common? If yes, could you elaborate on it?

3. How is alignment achieved between strategy and daily operations?

4. How is work coordinated between departments and hierarchies? How did the implementation of Lean influence the coordination?

1. How do employees and leaders recognize inefficiencies or improvement opportunities?

2. How does the hospital use data to guide improvement initiatives?

3. How are improvement ideas selected and prioritized?

4. What role do leaders play in supporting employees in problem-solving and CI?

5. Has Lean influenced strategic decision-making and long-term hospital-wide change? If so, how?

1. How have leaders (top management, middle managers and team leaders) influenced Lean adoption?

2. What role does middle management play in sustaining Lean practices?

3. How has leadership behavior changed due to Lean implementation?

4. Can you describe how problem-solving activities are integrated into your daily work?

5. Can you share an example where lessons from one department were applied successfully to another?

1. What tangible improvements have you observed since Lean implementation?

2. Have there been measurable impacts on patient care, employee engagement, hospital efficiency, quality or financial benefits?

3. How does the hospital/department monitor and measure the impact of Lean initiatives?

4. What challenges remain in embedding CI?

5. How do you see Lean evolving in the hospital over the next few years?

1. Is there anything else you would like to share that we have not covered?

2. Would you be open to a follow-up conversation if needed?