Understanding variation in subunit adoption of electronic health records: facilitating and constraining configurations of critical dependencies

ABSTRACT

This interpretative case study demonstrates how the work system properties and roles of clinical departments are perceived to shape their adoption of an organisation-wide electronic health record. In hospitals, with their heterogeneous and powerful clinical departments, partial or non-adoption by some departments may jeopardise effective overall use. Viewing a hospital’s EHR adoption from a sociotechnical fit perspective, we unravel how clinical departments’ adoption of this organisation-wide system was primarily shaped by their critical work system dependencies rather than their role during its implementation. From the interviewees’ accounts, three orientations emerge: (1) organisation-oriented resulting in unconditional adoption; (2) department-oriented resulting in problematic or delayed adoption with high compromise costs; (3) environment-oriented resulting in at least partial non-adoption or extra design costs. Our subsequent analysis leads to the identification of four types of subunit-level characteristics influencing adoption. Their interplay informs departments in balancing between accepting compromise costs and negotiating design costs. We develop a diagnostic system model of subunit-level adoption that takes the embeddedness of the subunit into account and thereby complements individual- and organisation-level adoption theories. An accompanying heuristic may enable hospital managers to better anticipate the heterogeneity in departments’ adoption.

KEYWORDS:

• adoption
• clinical department
• electronic health record
• sociotechnical fit
• subunit work dependency

1. Introduction

Many hospitals have adopted organisation-wide electronic health record (EHR) systems to deliver better quality care and contain costs (Buntin et al., 2011; King et al., 2014). EHRs are intended to share information among all the healthcare providers involved in patient care. A shared EHR forms the backbone of a hospital’s information flows and is often integrated or connected with other applications (King et al., 2014). Despite effective and consistent organisation-wide EHR use seeming crucial for a hospital’s functioning, the adoption by all clinical departments of a single uniform system is perhaps too easily assumed. Hospital departments are both heterogeneous and powerful subunits. While clinical departments are, to some extent, interdependent, each also has its own goals, speciality developments, workflows, technologies, and interests. Plurality in organisations is known to problematise adoption (Keen, 1981; Kohli & Kettinger, 2004; Markus, 1983). Indeed, in the case of organisation-wide EHRs, full adoption is found to be a highly complex endeavour (Greenhalgh et al., 2009; Jensen & Aanestad, 2007; Van Akkeren & Rowlands, 2007).

Most EHR adoption research explains adoption through either individual or organisational factors (Sadoughi et al., 2019). However, during our EHR research, we observed that many of the practitioners’ emerging ideas about the EHR, and their voiced views on EHR adoption, evolved around departments. For example, one surgeon stated: “we [the department] are very much in favour of this new integrated health record, we are looking forward to it”, whereas a specialist from another department argued, “we don’t see many advantages for our department, rather we foresee problems”. Remarkably, very few studies signal EHR adoption issues caused by the heterogeneity among clinical departments (Ben-Assuli, 2015; Greenhalgh et al., 2009; Park et al., 2017; Van Akkeren & Rowlands, 2007). Even in those studies that do, the nature of this heterogeneity and its implications for EHR adoption are not systematically addressed. In the wider IS literature, there is considerable research indicating that similar systems will be appropriated differently (S. R. Barley, 1986; Lanham et al., 2012; Noble & Newman, 1993), or recognising the possibility of tensions between the organisational value of overall adoption and limited local value (Grudin & Palen, 1995). However, neither of these research streams unravels the nature and role of heterogeneity among subunits in the latter’s (non-)adoption of an organisation-wide system.

Therefore, in this paper, we explore how the work system properties and roles of clinical departments are perceived to shape their adoption of an organisation-wide electronic health record. Within the context of a Dutch hospital, our interpretative case study compared the considerations of leading actors from eight departments concerning their departments’ adoption of an organisation-wide EHR. In this research, adoption is seen as involving the conscious decision to deploy an EHR for operational use within a clinical department’s work system (Alter, 2018; Venkatesh et al., 2003).

We used a sociotechnical fit perspective (Sarker et al., 2019) as a sensitising lens enabling us to use the work system as the level of analysis, whereby the department, as an adopting entity, is regarded as a work system (Alter, 2018). This perspective’s holistic nature fitted our interpretative method and allowed the analysis of both the EHR’s meaning for departments’ work system alignment and the departments’ role as adopters striving for such alignment (Alter, 2018). Alignment is crucial as work systems consist of interrelated “work practices, participants, information, and technology” (Alter, 2006, p. 12), outputs (patient care), and external environments (Jasperson et al., 2005). This interrelatedness emphasises that adopting an organisation-wide system will influence all components and disrupt a department’s work if it causes misalignments. In adopters’ alignment efforts, it has been suggested that “the social and the technical should, whenever possible, be given equal weight” (Mumford, 2006, p. 321). Conceptually, this can be translated into creating room for democratic and participative communication and decision-making that gives prospective users a voice (Sarker et al., 2019). In practice, reconciling heterogeneous voices representing diverse work system alignment needs within one organisation-wide system may be highly problematic.

We contribute to the EHR adoption literature by demonstrating how a subunit-level analysis offers complementary insights into organisation-wide EHR adoption in hospitals containing heterogeneous and powerful subunits. Our study identifies the sources of departmental heterogeneity and the implications for subunits’ (non-)adoption of an organisation-wide EHR system. In a hospital setting, the heterogeneity caused by complex, multi-logic groupings affects clinical departments’ perspectives on EHR adoption. The diagnostic system model that we develop can support managers in identifying departmental differences relevant to the local adoption of an organisation-wide EHR. In the discussion, we argue that some of the identified subunit-level heterogeneity will become more, rather than less, challenging in the future, despite the availability of more advanced technologies, and discuss the model’s generalisability to other sectors.

2. Theoretical background

2.1. Adoption theories’ limitations in explaining organisation-wide EHR adoption

A systematic literature review (Sadoughi et al., 2019) showed that EHR adoption is most often explained at the individual level through the theory of planned behaviour (Ajzen, 1985), the technology acceptance model (Davis, 1989), diffusion of innovations (Rogers, 2003), or the unified theory of acceptance and use of technology (Venkatesh et al., 2003). Only a few studies explain EHR adoption at the organisational level through the technology-organisation-environment framework, institutional theory, stakeholder theory, or social network analysis (Sadoughi et al., 2019). None of the included studies explain EHR adoption from a subunit-level perspective.

While the individual-level adoption theories are mature and influential (e.g., Wisdom et al., 2014), they are also heavily criticised (e.g., Grover & Lyytinen, 2015). They explain individuals’ adoption of information systems in a voluntary context through factors such as perceived ease-of-use, usefulness, social influence, and facilitating conditions. However, organisation-wide EHRs are mandatory, and top management usually implements such systems assuming individuals’ compliance. Sarker and Valacich (2010) demonstrated how technology adoption, as a group-related phenomenon, cannot be adequately explained by individualist views. Individual-level adoption theories explain variations in users’ intentions and behaviour, but do not delve into the sources or processes behind users’ shared perceptions. We examine why members of different clinical departments differ in their shared perceptions of the usefulness and applicability of an EHR system. For example, we would expect users to perceive the usefulness of systems in terms of the work processes their subunit is responsible for.

Organization-level adoption theories have emerged from the strategy and innovation literature. Competitive advantage, transaction functionality, speed (Yu & Tao, 2009), IS value creation, social commitment (Nielsen & Persson, 2017), and quests for control and organisational effectiveness (Malaurant & Karanasios, 2020) have been shown to be influential factors shaping organisation-level adoption (Oliveira & Martins, 2011). A limitation of these theories is their assumption that organisations are monoliths managed from the top. They assume that the organisation’s interests can be unequivocally determined, that top management can unilaterally decide to implement and enforce a standardised organisation-wide system, and that subunits are sufficiently homogeneous, submissive, and instrumental to the organisational goals. This assumption is not realistic in organisations characterised by heterogeneity, decentralisation, and professional autonomy (S. R. Barley, 1986; Keen, 1981; Lapointe & Rivard, 2007; Markus, 1983; Noble & Newman, 1993). In hospitals, departmentalisation is accompanied by a significant degree of decentralisation, granting department-level management substantive authority (Mintzberg, 1979). Moreover, clinical departments can claim decision autonomy by referring to their professional accountability to the medical speciality they represent locally. Consequently, hospital and project managers have to negotiate with departments about organisation-wide adoption (Kohli & Kettinger, 2004; Malaurant & Karanasios, 2020; Petrakaki & Kornelakis, 2016).

In these negotiations, they need to balance transparency and efficiency, through IS-enabled centralised control and standardisation, with customised IS support for heterogeneous work systems on which the departments rely. The stakes may be high since departmental variations in adoption may negatively impact the EHR’s overall effectiveness and, thereby, the organisation’s performance. Gattiker and Goodhue (2004) conceptualise this balancing as trading off two types of costs when a standardised IS is implemented across subunits. Design costs are the additional time and money an organisation spends on designing and implementing an IS, including possible database structure redesign, to meet a subunit’s local needs. In contrast, compromise costs are those incurred by a subunit when the standardised IS lacks the flexibility needed to meet its unique local needs, leading to reduced operational performance or data relevance (Gattiker & Goodhue, 2004, p. 434; Volkoff et al., 2005). In hospitals, we would expect heterogeneous clinical departments to negotiate the balancing of these two cost types with hospital management.

2.2. The role of heterogeneous subunits in electronic health records’ adoption

EHRs are integrated software applications used by healthcare providers to create and share patients’ health status information and register their interventions. As such, these systems are repositories in which digitalised healthcare process information is organised in a way that enables the registering, processing, and sharing of patient data, both on the individual patient level within the daily workflow as well as on the aggregated level (for quality monitoring, reimbursement, teaching, and research purposes). Theoretically, EHRs facilitate clinicians in basing their decision-making on an all-encompassing overview of a patient’s data (King et al., 2014). The expected benefits are that integrated applications, a uniform data model, and shared patient data will improve patient service, quality, and healthcare safety (Boonstra et al., 2014; Buntin et al., 2011). Additionally, EHRs should be cost-effective due to efficient patient data flows (Thouin et al., 2008).

Research shows it is not easy to achieve these benefits, with EHR implementation studies reporting project delays, high implementation costs, and user resistance (Colicchio et al., 2019; Jensen & Aanestad, 2007; Van Akkeren & Rowlands, 2007). Moreover, studies on EHR use once implemented report alert fatigue, and endless system demands continued use of paper, and difficulties in locating required information (Abramson et al., 2012; Carspecken et al., 2013; Colicchio et al., 2019; Flanagan et al., 2013; Saleem et al., 2011). We intuitively expect a substantial part of such EHR-use problems to be symptoms of work system misfits at the subunit level linked to workflow mismatches (Blijleven et al., 2017; D. M. Strong et al., 2014) and complex clinical and organisational adjustments (Avison & Young, 2007; Boonstra et al., 2014). These misfits could be a consequence of departmental work-system heterogeneity and serve as a barrier to compliant adoption by some departments.

In hospitals, an important source of heterogeneity among clinical departments stems from the complex mix of grouping logics applied. This is a consequence of the historical division into medical specialities. Large organisations tend to be departmentalised, mostly following one or two grouping logics such as function, customer, geography, purpose, or operating process (Scott & Davis, 2015). Through this grouping of activities, a department acquires its own distinct and relatively stable work system architecture (Alter, 2006). The grouping logic also influences wider social orders, such as departments’ interdependencies with internal and external parties. A widely applied sociotechnical design rule is to group interdependent activities within departments and to minimise interdependencies between departments. However, hospitals combine diverse grouping logics with some specialities oriented towards patients’ life stage (e.g., paediatrics), some to a body part or organ (e.g., dermatology), and others to a disease (e.g., rheumatology), or a treatment phase (e.g., rehabilitation). This multi-logic grouping of activities leads to differences in the work dependencies in and across departments and, thus, in their information needs and flows. In contrast, the philosophy behind EHRs assumes that subunits can perform their work with one overarching data model and comply with all shared information processing rules and procedures. The seemingly compelling argument is that all this information is derived from and comes together in one common entity, the patient, who can be followed over time (King et al., 2014). However, this argument contrasts with the multi-logic grouping that induces a complex variety of work systems, because of which the EHR may not be aligned with some departments’ work systems.

EHR adoption studies provide some evidence that a hospital-wide EHR is not equally useful for all clinical departments and their work systems (Ben-Assuli, 2015; Blijleven et al., 2017; Eisenberg et al., 2013; Park et al., 2017; D. M. Strong et al., 2014; Van den Hooff & Hafkamp, 2017). For example, ophthalmologists have a unique workflow, distinct from other specialities, due to their high patient volumes and standardised but highly specific workflows (Park et al., 2017). Gynecologists, on their part, face problems when investigating a foetus without a digital medical record number because they have no tools to document their interventions (Eisenberg et al., 2013).

Another study found that a group of orthopaedic surgeons, who had previously enjoyed significant autonomy, felt that a hospital-wide EHR increased managerial control and imposed more administrative tasks (Jensen & Aanestad, 2007). Elsewhere, a group of radiologists who found they had few alternatives but to use the EHR system devised substantial adjustments such as workarounds so they could perform their tasks (Van Akkeren & Rowlands, 2007). Moreover, the latter study argued that institutional forces embedded within the EHR actively influenced the adoption process. The authors demonstrated how governmental requirements to use EHRs for medical information exchange within and among hospitals and pressures from medical professional associations were both external forces that explained EHR adoption. Such external forces may not always align with the internal forces, as Politi et al. (2022) demonstrated in the context of treating critically ill patients in an emergency department. Here, the attempts to establish a single integrated EHR interface for all medical data within the hospital did not fit with expectations from the wider environment to use systems that cater to external stakeholders’ diverse needs. Thus, although we have found some indications of department-level adoption issues, attention to the nature of the heterogeneity among clinical departments and the consequences for organisation-wide EHR adoption remains scarce and fragmented, as our summary in Table 1 shows.

Table 1. Literature implying subunit-level variations in EHR adoption.

Source	Research type	Major findings that imply subunit-level adoption problems
Jensen and Aanestad (2007)	Interviews among a group of orthopedic surgeons	As the orthopedic surgeons enjoyed considerable autonomy, it was important for managers to understand their views and roles in the adoption process. The surgeons experienced the EHR as a control mechanism, creating unwelcome new tasks.
Van Akkeren and Rowlands (2007)	Interviews with radiologists	Radiology practitioners had few alternatives but to adopt the innovation and make the necessary adjustments to perform their tasks. The study argues that institutional forces embedded within the EHR are active in the system’s assimilation process. This case study also explores the role of institutional structures, such as government requirements and industry pressures, in integrating health ICTs by radiology practitioners.
Lanham et al. (2012).	A qualitative study of six practices operating within multi-specialty ambulatory care	Within-practice communication patterns were associated with practice-level EHR use patterns. In practices where communication patterns were fragmented, EHR use was heterogeneous. In practices where communication patterns were cohesive, EHR use was homogeneous.
Eisenberg et al. (2013).	Interviews with obstetricians and gynaecologists	Obstetrics and gynecology have a unique workflow and content challenges. Unique data collection and imaging. The fetus often does not have a medical record number, leaving the provider with no tools to document care interventions. Specific privacy laws regarding adolescent care.
Nguyen et al. (2014)	Systematic literature review	Common negative impacts of EHRs include changes to workflow and work disruption. A lack of sociotechnical connections between the clinician, the patient, and the technology in developing and implementing the EHR and future developments in patient-accessible EHRs.
Ben-Assuli (2015)	Review of empirical studies in emergency care	The unique properties of emergency departments create an environment where EHRs may significantly contribute to health but also face substantial challenges. Using the EHR increased waiting times, treatment times, and physician documentation times, resulting in less patient time.
Park et al. (2017)	A national physician survey of 1199 Canadian surgeons	The barriers that limit increased EMR adoption among Canadian ophthalmologists differ from those of other surgeons. This could be attributed to unique features of the field, including heavy reliance on hand-drawn figures in the documentation, large patient volume, and the high costs associated with independent practice.
Van den Hooff & Hafkamp (2017)	Qualitative data collected on EHR system implementation.	Various misfits between systems and work practices lead to feelings of dissonance among medical professionals. In response to such dissonance, users devise workarounds that entail adjusting routines or changing the use of technology in these routines.
De Benedictis et al. (2020).	A survey among 114 physicians and nurses	This study confirmed that both perceived usefulness and ease-of-use at the individual level, and departmental and inter-hospital normative and regulative forces, are organization-level factors that explain adoption.
Politi et al. (2022).	Analysis of data recorded in system log files on physicians’ substituting or complementing HIE and EMR use in an emergency department	Despite attempts to form a single integrated interface for all medical data, the medical environment often includes multiple systems that cater to different needs for storing and accessing patient data. In the context of treating critically ill patients in the emergency department, this study shows how distributed health data produces information gaps that potentially endanger patients and increase healthcare costs.

To summarise, the setbacks observed in hospitals trying to realise EHR systems’ proclaimed organisation-wide benefits (Avison & Young, 2007; Boonstra et al., 2014) have led to calls for sociotechnical research that addresses adoption issues related to the work system (Black et al., 2011; Carayon & Salwei, 2021). Our study responds to this call with a focus on how the heterogeneity that results from complex, multi-logic groupings affects clinical departments’ perspectives on EHR adoption.

3. Method

We have conducted an interpretative embedded case study within a large Dutch hospital, which allowed us to study departmental adoption in its context through the participants’ eyes (Myers, 2009; Walsham, 1995). While interpretative research is based on the assumption that human experiences and social contexts shape social reality, it still accepts the ontological position (Walsham, 2006) that such shaping is limited by the technology’s physical properties (S. R. Barley, 1986), which accords with our sociotechnical fit perspective (Sarker et al., 2019).

3.1. Case context and department selection

The hospital studied had initiated a program directed at the organisation-wide implementation of an off-the-shelf EHR system guided by a uniform data model developed in-house within which the core patient record comprised around 2700 items, categorised in 34 rubrics. To improve cross-departmental workflows, patient safety, and efficiency of care, the change would include a simplified primary flow process while still allowing for variations in iterations and supporting steps. The EHR was intended to replace 250 applications and reduce 800 roles to a much more limited layered set of roles and authorities. By the summer of 2013, the 47 clinical departments had each submitted a template-guided description of their processes to check and enrich the generic modelling and a more freely formatted set of specific adoption issues, needs, and wants. Especially during the pre-implementation phase, care professionals were represented in the project teams, task forces, and process innovation pilots. Around 8600 employees received information, predominantly through plenary meetings, brochures, newsletters, and intranet documentation. Before and during the Plateau 1 release, training was provided through workshops, an online helpdesk, and workplace support. However, the organisation-wide nature of the EHR and the release of Plateau 1 left little room for customisation: the focus was on providing basic support for the internal hospital processes and service reimbursement.

Our study started in 2013 and covered the pre-implementation stage, complemented by the realised adoption for each department through 2019 (see Figure 1).

Figure 1. The time path of data collection.

For our comparative study (Walsham, 1995), we selected eight departments (coded A – H). This selection was aimed at maximising relevant heterogeneity. At the start of the study, two change managers had already developed a worksheet characterising each clinical department, and this formed the basis for selecting and inviting the participating departments. These change managers expected the following three department characteristics to generate the greatest variety in adoption (Table 2): (1) the department’s existing patient record registration (they distinguished between largely digitalised or largely paper-based); (2) its staffing level (small:< 75 fte, medium:75–150 fte, large:>150 fte); and (3) the grouping logic that the department’s speciality reflects. The change managers’ reasoning can be seen as following a work system perspective, as these three departmental characteristics reflect work system properties. Our first selection criterion on which to sample for variation was thus the existing patient record, which guaranteed differences in costs and benefits associated with a new technical system. The second criterion, size, was expected to bring variation in the departments’ roles during the implementation. The third selection criterion, the grouping logic underlying a medical speciality, heavily affects a department’s information needs, work practices, outputs, and coupling with other departments and external actors.

Table 2. Department description based on the selection criteria.

Department	Patient record digitalisation	Staffing level (small:< 75fte, medium:75-150fte, large >150fte).	Grouping logic
A	High	Small (outpatient clinic, no ward)	Patient’s life stage
B	Partly	Small (outpatient clinic, shared a ward with other departments)	Patient’s life stage
C	Low	Medium (large outpatient clinic, no ward)	Organ/body system
D	High	Medium (outpatient clinic, own ward)	Organ/body system
E	Partly	Large (multiple wards)	Treatment stage
F	Partly	Small (outpatient clinic with a few short-stay beds)	Treatment stage
G	Partly	Large (outpatient clinic, own ward)	Patient’s disease
H	Low	Small (outpatient clinic, shared a ward with other departments)	Organ/body system

3.2. Data collection

The data collection methods included document searches, interviews, meetings, and informal communications. The authors designed the interview guide and conducted most of the interviews. In 2013, during the pre-implementation period, 24 interviews were held with department managers and leading representatives of physicians, nurses, and administrators (McGinn et al., 2011) and with four project managers. Two years later, 12 key players in the departments were again interviewed, along with four project managers (Figure 1).

The interviews were semi-structured, each lasting between 60 and 90 minutes. To explore how specific characteristics affected a department’s adoption, the guide included not only questions about the interviewees’ departments (e.g., What are the distinguishing characteristics of how your department is organized? Are there any specific department features that put a special demand on the EHR? Which ones, and why?), but also about the department’s involvement in the EHR project (e.g., Is your department represented within the program or any of its projects? How?), and the department’s intended adoption (e.g., What was, generally speaking, the initial reaction in your department? And yours personally?).

Finally, to understand the realised adoption in each department studied, we held interviews in five departments and informal communications involving the other three departments between spring 2018 and winter 2018–2019. The post-implementation interviews were short and focused on recording the adoption realised in each department, any customised extras delivered, the emergence of workarounds, and the overall reaction to the EHR. To ensure confirmability (Lincoln & Guba, 1985), we checked our interpretations of the realised adoptions in each department in an interview with the Chief Medical Information Officer (CMIO), a senior medical specialist who had developed a good overview during the project.

Further, before the implementation, we held a meeting every six weeks with project management representatives to ensure confirmability and contextualise the data. Given our aim of examining implementation on the department level, we held feedback sessions to share and discuss interpretations. Furthermore, during the implementation and the period thereafter, meetings with the project management’s representatives continued. The meetings and additional data sources, including newsletters, written reports, and policy plans, enabled us to understand better the departments’ relevant work system characteristics and roles and how these impacted the intended and realised adoption.

3.3. Data reading, coding, and analysis

Holistic case reading. For each department, we developed a holistic storyline, based on our reading of the available materials, to understand their members’ voiced motivations for and reservations against adoption, their intended adoption, and the realised adoption one year after implementation. In discussing these storylines and tentatively mapping emerging differences between departments, we noted how a department’s specific work system characteristics dominated their reasoning about the adoption of the EHR, and much more so than their role in the change process.

Coding of relevant departmental adoption-related characteristics. We unravelled how departments voiced their role and how they evaluated the expected consequences for their local work system. Two of the authors individually coded the data using inductive codes. These first-order codes were combined to create second-order categories involving more general considerations (Gioia et al., 2013; see codebook in Appendix 1 and data structure in the supplementary materials). We went back to the transcripts, the documents, and earlier generated data displays to deepen our understanding of the reasons behind the views articulated on the intended adoption. We then refined the code definitions for the perceived adoption-influencing characteristics. These iterations led to a department-based list of voiced considerations underlying the intended adoption, which could finally be grouped into characteristics related to the departments’ work system or their role in the change.

Establishing departmental adoption. Each department’s adoption was described at two points in time to better understand how this developed. First, in the pre-implementation phase, the same two authors evaluated each interview quote that directly referred to views on the likelihood of the department’s future EHR adoption, i.e., the intended adoption. We summarised these quotes by a department’s leading actors into short department-level statements on the intended adoption. Second, a year after Go Live, we analysed the realised adoption based on the formal and informal data sources specified above.

Identifying decisive facilitating and constraining conditions for adoption through cross-case pattern analysis. We cross-analysed, including additional data checks, the voiced considerations in the departments to derive one tree-like display with all the work systems and role characteristics perceived to influence departmental adoption. Here, how the interviewees explained the departments’ intended adoption and reflected on the subsequent realised adoption enabled us to distinguish between characteristics perceived as dominant conditions for the adoption and characteristics experienced as secondary conditions. Moreover, the cross-case comparison of the tree-like display with departments’ accounts of the patterns of intended and realised adoptions helped to track and carefully interpret each characteristic as a facilitating or constraining condition. From this cross-case analysis, three adoption configurations emerged that each turned out to be characterised by different critical work system dependencies. Through our theory-guided interpretation of this inductive cross-case pattern analysis, we arrived at four types of departmental characteristics that operate as facilitators or barriers at the subunit-level during EHR adoption.

4. Results

Three adoption configurations emerge from the cross-case analysis: organisation-oriented, department-oriented, and environment-oriented. Section 4.1 presents a representative example of each configuration where we describe the alignment of the EHR with the department’s work system, the department’s role in the change process, and the intended and realised adoptions. The supplementary materials contain descriptions for the other departments. Section 4.2 reports the cross-case analysis, based on which the departments were grouped into these three configurations. Table 3 presents an overview of all eight departments’ considerations and their realised EHR adoption.

Table 3. Overview of departments’ considerations and adoption.

	EHR’s alignment and work system component(s)					Role in the change process			Adoption
	Intradepartmental dependencies: hard technical capabilities (Technology)	Intradepartmental dependencies: soft technical capabilities (Participants)	Intradepartmentaldependencies: tasks and activities (Process)	Interdepartmental dependencies (Information)	Environmental dependencies (Environment-Information)	Trust in project	Involvement	Influence on design	Intended adoption	Realized adoption
A	Not voiced as a major issue.	Instrumental attitude towards IT, no IT staff.	Current patient administration is inefficient and unreliable; some misfits with the new EHR.	High: hospital-wide dependencies because of multi-disciplinary care.	Moderate, but this is not a key focus.	Wait and see; attention elsewhere	Passive	No	Compliant	Yes, adopted without customized extras. Happy with the progress.
B	A few interfaces with existing, less-advanced technologies.	Limited, no IT staff.	Current patient administration is inefficient and unreliable.	Moderate: multi-disciplinary care, collaborates with a few other departments.	High: requirements on data reporting and role in region’s patient care.	Wait-and-see	Passive	No	Conditional	Yes, adopted with negotiated customized extras. Workarounds by individual physicians.
C	It will require many interfaces with their advanced digital equipment.	Local IT expert,IT-minded department.	Vulnerable, detailed work design; misfit with the new system.	Low: mono-disciplinary care, independent work processes.	Low.	Trust in the process, but very worried about the outcome	Proactive, persevere	No	Conditional or non-adoption	Yes, but problematic by constraining the workflow design, a collective departmental workaround. A negotiated customized module will be added later.
D	Some interfaces with less advanced technologies.	Vision for the use of IT by specialty; expertise available.	Current patient administration is inefficient and unreliable; interest in uniformity.	Moderate: multi-disciplinary care, collaboration with other departments.	Moderate: through requirements on data reporting.	Optimistic	Proactive has a champion	Feel heard	Enthusiastic	Yes, adopt without customized extras, although not as satisfied as expected.
E	Many interfaces with complex technologies are required.	Some local expertise on IT use; own IT vision, IS under development.	Current patient administration is inefficient and unreliable; interest in uniformity.	Moderate: due to patient referrals to and from other departments.	Low: because it constitutes a transitional stage in care.	Not voiced;attention directed elsewhere	Reactive	Feel heard	Limited	Yes, negotiation resulted in late adoption, with one member being trained to generate customized patient data analyses, partly making up for the lost functionalities.
F	A few stand-alone systems; are expecting a better fit in terms of better-connected systems.	The vision of IT use for the specialty.	Current patient administration is inefficient and unreliable; interest in uniformity.	Moderate: data exchange with various other departments.	Moderate: the department has a central position in regional healthcare.	Some concerns, especially with regard to damage control during the implementation	Reactive	Little	Compliant	Yes, adopted with one customized module. Satisfied department members.
G	‘Band-aid’ applications with few functionalities. Expects a better fit in the sense of more integrated functionalities.	The vision of IT use for the specialty; a clinician with IT knowledge.	Current patient administration is inefficient and unreliable.	Low: mono-disciplinary care; independent work processes.	High: through requirements on data reporting.	High expectations	Proactive, strong drive to be involved	Feel heard	Eager to adopt and overcome recognized barriers.	No, implementation is on hold. Future developments are unclear.
H	Work is in steps, becoming increasingly paperless; the EHR is the next (big) step.	Limited, no IT staff.	Current patient administration is inefficient and unreliable.	Moderate: intensive collaboration with a few related specialities.	Low.	A single system is desirable. Trust that it will come, but it will take a long time.	Proactive, focused on internal preparations; no priority.	Little	Compliant; it is not an issue; it will come.	Yes, adopted without customization.

4.1. Within-case analyses

4.1.1. Example of an organisation-oriented department

Work system. Department A is a small generic subunit staffed by six medical staff members and one psychologist. The department is responsible for a small outpatient clinic staffed with three nurses. There is already a high degree of digitalisation of patient records: “We are a forerunner in working digitally” (A2). Unusually for this hospital, the nurses were also already working with digitalised records. The interviewees emphasised their multidisciplinary patient focus, expressed by one of the doctors as follows: “We are one of the very few departments that focus on our patients and not so much on diseases” (A1). The department does not have a nursing ward, so patients requiring hospitalisation are admitted to wards belonging to other departments. The clinicians from this department will visit their patients in these wards. Their work processes intertwine with and are highly reliant on cooperation with other departments.

Role in the change process. The interviewees admitted that given the department’s small size, their influence was limited, and, as other developments were requiring their attention, the new EHR was not their main priority. They thus seemed to have opted for a wait-and-see role.

Intended adoption. The interviewees welcomed the idea of a hospital-wide EHR and expressed no doubts about adopting the EHR as a department. The department intended to comply and indicated that its operations would benefit from a single shared patient record across departments and professions.

Realized adoption. The two interviewed users were happy with their progress in digitalising patient-related data, as noted by one of them “ … we are catching up now, and I’m happy with it [the EHR]” (A4). Although the department’s members were slightly concerned that their specific, multidisciplinary approach may be in jeopardy because of the system, they adopted it as implemented and did not demand customised extras.

4.1.2. Example of a department-oriented department

Work system. Department C is a medium-sized monodisciplinary subunit consisting of doctors, nurses, paramedics, and technical support staff (120 fte). Its high-volume patient care flow at its outpatient clinic has few interdependencies with other departments. It uses many advanced technologies that are fully integrated into the department’s workflow and on which they are heavily dependent. They treat approximately 300 patients per day, and for each, there is a strictly enforced timeslot enabled by a standardised workflow. Patients complete a pre-defined set of process steps depending on their health issues. The department had previously put considerable effort into designing these work processes, which the interviewees experienced as highly efficient.

Role in the change process. The department had established an EHR team before the project team even requested one. The nurses had a proactive representative, and the department manager was involved in what she called “pushing and pulling”. They put effort into getting their demands and preferences across and felt facilitated by the hospital and project staff during this stage. In contrast, the manager characterised the specialists’ behaviour as follows: “Hold on, I would say due to a lack of faith in the EHR. Of course, many hospitals have already implemented one, with many problems for our speciality. Our specialists have many contacts with colleagues, and they hear how the gains are minimal. There is a big fear…this fear is realistic”. The IT specialist also attributed the fears to bad experiences with earlier software packages (C4). In later stages, they remained involved, as a physician (C2) put it: “ … we are on top of it … We see … 10% of the hospitals’ patients. Still, we are small, so we need to put the effort in to get our voice heard … and our needs are different than those of, for example, internal medicine.”

Intended adoption. Given their existing, well-tailored care processes, they envisaged only minor benefits from the new EHR. The department’s manager expressed this as: “it can save some time in routing the patient records from the depot through the process … but we are particularly concerned that the system will not be able to follow a patient through our process, that is [a misfit with] how we have organised it logistically” (C3). As such, the department was worried that the new EHR might create more problems than benefits. During the pre-implementation phase, it was uncertain whether the new EHR could be aligned with the department’s existing workflow that was integrated with the advanced technological equipment it used.

Realized adoption. The EHR was implemented without any customised extras for the department. Due to the high daily volume of patients, it was crucial that the department could quickly retrieve their patients’ concise, speciality-related medical histories. The system could not meet this requirement. Therefore, the department’s management decided to create a workaround, which all its specialists were obliged to use: “ … it is not even allowed to use it in a different way; these are firm arrangements within our department” (C4). They acknowledge that their view conflicts with having a unified, organisation-wide EHR: “Our way of working goes against agreements” (C5). However, the department also negotiated an additional module to be implemented later to hopefully resolve this mismatch.

4.1.3. Example of an external environment-oriented department

Work system. Department G is a large subunit (230fte) with a ward and outpatient clinic. Here, a digital record is created for new patients, while existing patients still have paper-based records (internal documents). This department is not located in the main building, and their care processes have few work dependencies on other departments. One manager explained: “Some departments operate very isolated within the organisation. We are very isolated … our information systems differ considerably from the other information systems … we don’t collaborate that much” (G2). Given the specialist nature of the department’s treatments, patient registrations have additional requirements to enable the therapies to be reimbursed by insurers.

Role in the change process. Despite this lack of integration, they showed themselves highly committed to helping the change occur. The interviewees did see benefits in the new system for their department: “People intend to use the new system … they can hardly wait because we are faced with the limitations of [the current information system]” (G2). They also saw it as a way to better connect to other departments. The interviewees did not worry about technology-task fit: “ … business as usual, with another information system” (G2) and “this is, for the most part, about finance and bureaucracy, the way we provide care will not change that much” (G1).

Intended adoption. Despite their enthusiastic support and proactive involvement, the department management’s intention to adopt was conditional: “We really do want one [generic] information system. We are true proponents! However, it has to go our way…we will still have to comply with our own [external] rules and legislation” (G4). It was uncertain whether the new system would facilitate these requirements.

Realized adoption. The new EHR proved incapable of supporting the production registration system demanded by the insurers. Departmental adoption was, therefore, suspended until further notice.

4.2. Cross-case analysis

Departments A, C, and G had different critical work system dependencies: towards the organisation (A), within the department itself (C), and towards the wider environment (G). These dependencies influenced their adoption: Department A achieved a relatively unproblematic adoption, although they experienced the benefits as somewhat disappointing; in Department C, we observed the collective creation of substantive workarounds; and, in Department G, the EHR was not adopted. This section presents the cross-case analysis of all eight departments leading to three distinct configurations.

4.2.1. Departmental characteristics informing adoption considerations

The departmental characteristics interviewees voiced concerns about in considering their department’s EHR adoption are presented in Table 4. Based on the shared elements in their narratives, the second column specifies the direction (facilitating, constraining, or mixed) and dominance (indicated in bold) of the characteristic’s impact. The table shows that the dominant conditions for adoption relate to critical dependencies in the work system rather than to the department’s role in the change. Below the table, we explain why some work system characteristics were perceived as dominant influences on the realised adoption.

Table 4. Departmental characteristics perceived to influence adoption.

Voiced impact of characteristics on department’s adoption: facilitating, constraining, or mixed		Considerations influencing department’s adoption as voiced by interviewees: (dominant influence in bold):	Departments
EHR’s alignment with:
Intra-departmental dependencies: hard technical capabilities (Technology)	Facilitating	Local technologies are expected to benefit from EHR-enabled interfaces or integration.	B, G
	Constraining	Local advanced or complex technologies not aligned with offered interfaces or integration.	C, E
Intra-departmental dependencies: soft technical capabilities (Participants)	Facilitating	Local IS expertise and change vision match with EHR.	D, F, G
	Mixed	Mixed feelings; will make do “just like with moving houses”	A, B, H
	Constraining	Local IS expertise and IS change vision are not aligned with EHR.	C, E
Intra-departmental dependencies: tasks and activities (Process)	Facilitating	Current patient administration is inefficient, unreliable, and fragmented.	A, B, D, E, F, G, H
	Facilitating	Current technology insufficiently supports its own work system.	A, B, D, F, G
	Facilitating	Own work system will benefit from EHR-enabled greater uniformity.	D, E, F
	Constraining	Own highly customized workflow is vulnerable to changes.	C
	Constraining	Expect the EHR to be misaligned with workflows or work methods.	A, C
Inter-departmental dependencies (Information)	Facilitating	Involved in multidisciplinary patient care streams.	A, B, D, H
	Facilitating	Treating patients coming from or for other departments.	A, E, F
	Constraining	Mainly monodisciplinary care and independent workflows.	C, G
Environmental dependencies (Environment-Information)	Facilitating	In a supply chain or collaborate with professional networks.	B, D, F
	Facilitating	Legal or professional requirements align with EHR.	D
	Constraining	Requirements of professional bodies not in line with EHR	B
	Constraining	Legal or financial requirements are not in line with EHR.	B, G
Role in change:
Trust in organization-wide implementation, informed by change leaders, change history	Facilitating	Optimistic; high expectations, e.g., based on other hospitals’ achievements; cannot afford another failure.	D, G, H
	Mixed	Wait-and-see; neutral.	A, B, E
	Constraining	Express worries or low trust based on past change history.	C, F
Department’s involvement in implementation	Facilitating	Proactive.	C, D, G, H
	Constraining	Reactive.	E, F
	Facilitating	Passive with a compliant attitude.	A, B
Department’s perceived influence on design	Facilitating	Feel heard, inferred from one’s size or own status.	D, E, G
	Constraining	Experience no or little influence, sometimes attributed to small size.	A, B, C, F, H

First, we see that the pre-implementation accounts from five departments highlight the importance of a fit between the new technology and their existing hard technical capabilities. Department E had started to independently develop its own application: “At E we now have a patient management data system […] which we will keep using. Thus, we will not fully adopt the EHR” (E3). They doubted whether this application could even be connected to the organisation-wide EHR. The key actors in Department C voiced similar concerns about the possible integration of, or interfacing with, the numerous and advanced technologies already employed in their efficiently organised work processes, leading to strong reservations towards EHR adoption. In contrast, in terms of the hard technical capabilities, two departments (B and G) explained that they expected better connections or even the integration of their own systems. The role of soft technical capabilities was less straightforward as it depended on the perceived fit. The existing IT expertise in three departments (D, F, and G) helped envisage the new EHR’s potential. In contrast, the local IT expertise in two departments (C and E) constrained their inclination to adopt as their workflow critically depended on existing advanced technologies.

Second, Table 4 presents a range of conditions related to technology-task fit that were voiced during the interviews. Clearly, perceiving the existing patient administration as inefficient, unreliable, or fragmented is a facilitating condition for EHR adoption in all departments except Department C. Expecting current practices to benefit from the uniformity that the new system imposes was also a positive stimulus for intended adoption (D, E, and F). Department C sensed that having a detailed, painstakingly designed workflow would make it vulnerable to change. They were worried that a new system would cause disruption.

Third, the interviews exposed how a department’s environment can play two counteracting roles determined by: (1) a department’s dependencies on other departments within the hospital (interdepartmental); and (2) its dependencies on the hospital’s environment. Six departments’ reciprocal work dependencies with other departments constituted the main reason for their intended adoption: they deliver multidisciplinary care or share patients with other departments. These departments expected greater benefits from a hospital-wide EHR system than departments that operate in a relatively “stand-alone” fashion, such as Departments C and G. When it comes to environmental dependencies, the picture is less clearcut. Interviewees from four departments voiced the relevance of environmental dependencies (B, D, F, and G), with those from D and F perceiving opportunities to more efficiently deal with supply chain partners or legal requirements. Department B was highly ambivalent regarding the system’s potential to facilitate its external information flows. Both Departments B and G recognised that this new EHR could not support them in meeting externally imposed information requirements.

Regarding their role in the change, the interviewees’ narratives did not reveal any decisive influence of their own department’s role that affected their adoption decision: Department G had trust, was proactive, and felt heard, but external regulations kept that department from adopting the EHR: “for Department G we had to decide against implementation. A proper solution could not be found” (project manager). As Table 4 indicates, our findings also provide no indications of other departments’ sociopolitical roles having a decisive influence on their EHR adoption.

4.2.2. Pattern identification: three adoption configurations

By combining the interviewees’ narratives with our comparisons of the patterns of departmental adoption-influencing conditions and the departments’ adoption, we were able to establish the critical adoption-determining condition(s) for each department (see Table 4). For example, the table shows that the hard technical capabilities of Departments C and E, that were critical resources for delivering their speciality’s patient care, had the most impact on their adoption. Although these departments supported the change direction, they had developed conditional or limited adoption intentions (Table 3). These two departments argued that their existing, highly advanced technology harmed their intended adoption. Their representatives voiced this component as decisive in their intended adoption: “[The existing applications and equipment] are an incredibly huge investment that [Department C] has made, so imagine that you have just adopted these, and then they get thrown out once the EHR is implemented, then you end up totally suckered”. For these departments, the fit with this key component was perceived as critical compared to other alignment issues (such as department E’s references to their “IT expertise and vision for the future”). In contrast, having only limited hard technical capabilities positively influenced Departments B and G in their intended adoption. Department B’s interviewees mentioned that greater integration between applications and a long-desired connection with a specific technology was expected to be realised through EHR implementation. However, they still had a serious concern that this positive influence would be outweighed by the negative effect on the department’s critical environmental dependencies related to national laws and speciality regulations and the integrated workflows established with external partners: “We have many referrals from a large region, and the doctors from our speciality are also based in almost all the hospitals in the region” (B1). Despite the expressed support for the organisation-wide EHR implementation, this led to strong reservations in the pre-implementation stage about their actual adoption.

Our pattern identification process resulted in three adoption condition configurations: (1) organisation-oriented work system, (2) department-oriented work system, and (3) environment-oriented work system.

The first configuration, with an organisation-oriented work system, was related to Departments A, D, F, and H, which had many interdepartmental dependencies. These workflow dependencies with other departments sprang from delivering multidisciplinary care and treating patients from other departments. Such departments are more dependent on interdepartmental information flows than some other departments, and all four departments were dissatisfied with their current systems in this regard. They expected the EHR to enable high-quality interdepartmental information exchange and voiced this as the decisive factor in wishing to adopt the system. Further, the realised adoption proved relatively unproblematic, although the use efficiency was sometimes slightly lower than anticipated [program management for D and H].

The second configuration, with an internal department-oriented work system, was seen in Departments C and E. Interviewees characterised their department as having relatively stand-alone workflows, predominantly pooled (C) or sequential (E), with few interdependencies with other departments. Furthermore, the interviewees voiced serious doubts as to whether their departments’ current technologies-in-use could be safeguarded or further developed under the new standardised system. Both departments were reasonably satisfied with their existing technologies (some still under development), and their relative independence from other hospital units had allowed them to integrate these into their own well-designed work processes carefully. The dependence on, and investments in, their own advanced technologies, interwoven with their workflow and independence from other hospital departments, led to reservations regarding the EHR. Before the Go Live date, they were already seriously considering only selective adoption of the EHR. Although these departments were nevertheless coerced into adopting the EHR, the realised adoption was problematic and delayed. In Department C, a collective workaround was developed to maintain their efficient workflow, and they persisted in their negotiations for a customised module, and this was eventually provided. In Department E, many of the data analysis functionalities of their customised patient data management system were lost with the new EHR, which delayed their processes. They were able to partly compensate for this through a medical specialist who invested in acquiring a high level of EHR expertise so as to be able to build customised flowcharts, and specific order sets to generate the required analyses “deep within” the EHR itself. In mid−2018, both departments were provided with their own “physician builder” – a physician responsible for maintaining and updating their department-specific EHR configurations.

Departments B and G represent the third configuration, with an environment-oriented work system. Both departments had significant environmental dependencies through regional or (inter-)national collaborations, laws, and insurance regulations. These departments were highly supportive of implementing an EHR but said they could only adopt the new system to the extent that it would not hinder them in fulfilling their external requirements. Their fears in this respect showed in their intended adoption: “we have these legal requirements that we must adhere to. You could standardise that, but, well, that clashes with the medical processes” (B2). In terms of realised adoption, it was negotiated that Department G, facing externally imposed production registration requirements that went against the EHR logic, would not adopt the system. For Department B, a customised module was developed to enable it to meet its external demands. Moreover, Department B negotiated that, alongside two applications replaced by the EHR, a third application that the physicians used would be integrated into the EHR. The latter customisation was, however, ultimately postponed by the supplier. Consequently, both a medical specialist and the project manager acknowledged that the outcome was perceived as complicated to work with, leading to the physicians using workarounds. Table 5 summarises our main findings.

Table 5. Adoption configurations based on the critical work system dependencies that informed the department’s adoption.

Departments	Adoption configuration	Dominant adoption-influencing conditions: critical work system dependencies	Secondary adoption-influencing conditions		Realized adoption
			Facilitators	Constraints
A, D, F, H.	Organization-oriented work system	Interdepartmental dependencies that are critical for the workflow (handover coordination).(Facilitator➔ information)	Current systems are inefficient or inadequate(A, D, F, H)(technology-process or technology-information fit)Strong implementation support (D).Proactive involvement(D, H).	Passive involvement (A).Concerns about organization-wide implementation success (F).	Unconditional adoption
C, E.	Department-oriented work system	Intradepartmental dependencies: interfaces with advanced local technologies critical for their workflow.(Constraint ➔ technology).	Proactive involvement (C).	Local technologies (under development in E) better fit the department’s IS vision (C, E)(technology-information fit).Independent workflow or only sequential interdepartmental workflow dependencies (technology-process fit).	Partial, with collective workarounds (C) or delayed adoption (E) after negotiated customized extras.
B, G.	Environment-oriented work system	Environmental dependencies, such as legal or insurance requirements or environmental interdependencies, and alliances at regional or (inter)national levels, are critical.(Constraint➔ environment)	Current systems are inefficient or inadequate (B, G).Proactive involvement and strong implementation support (G).	Passive involvement (B).	Non-adoption (G) or partial with negotiated customized extras (B).

5. Discussion

5.1. A subunit-level model of conditions for EHR adoption

We have explored how a clinical department’s work system’s characteristics and its role in the change process are perceived as shaping their adoption of an organisation-wide electronic health record. As clinical departments constitute a countervailing power to hospital management (W. C. Barley, 2015), they are salient units of analysis for assessing the adoption of organisation-wide EHR systems. We saw from previous empirical studies (Table 1) that subunit heterogeneity complicates EHR adoption but could find no research that models relevant departmental differences and how these affect EHR adoption. Our findings contribute to the EHR adoption literature by demonstrating how departmental adoption of an organisation-wide EHR can be facilitated and constrained by a department’s work system’s characteristics. We now discuss these findings and reflect on how our modelling of subunit-level conditions for adoption in an organisation-wide system implementation may complement individual- and organisation-level theorising on the adoption process.

Our data suggest that specific critical work system dependencies may block or limit a department’s adoption of an EHR system. In particular, this applies to departments whose workflow depends on specialised, advanced technologies that cannot be integrated into or connected with the EHR; or have invested in a complex and vulnerable work design that diverges from the EHR logic. Further, binding requirements imposed by a department’s external task environment through regulations or requirements of a network in which the work system is embedded can also block adoption. Notably, in such cases, we saw that even when the departmental actors had an active role in the change process and aimed for adoption, the system was ultimately not adopted (G) or only after additional design costs had been incurred (B). In contrast, Departments A, D, F, and H, which have critical workflow dependencies with other hospital departments adopted the EHR, even when this involved compromise costs, and the resulting benefits seemed disappointing to some.

While users’ roles during a sociotechnical change have been found to influence adoption (Leonard-Barton & Sinha, 1993; Petrakaki & Klecun, 2015), we failed to identify any decisive influence of a department’s role on its adoption. For example, two smaller departments (C and H) with comparable roles during the implementation differed significantly in their adoption. Departments C and G showed how subunits, one small and the other large, can actively support EHR implementation without realising full adoption themselves. In our view, the EHR’s misalignment with critical work system dependencies (here, obstructing service delivery (C) and reimbursements (G)) amounts to a constraint that cannot be conceptualised in terms of compromise costs (Gattiker & Goodhue, 2004).

Based on the three configurations identified from our data (Table 5) and guided by our sociotechnical fit perspective (Alter, 2006, 2018), we propose a diagnostic system model to help understand differences in departmental adoption (Figure 2; and also see the connected heuristic in Appendix 2).

Figure 2. A diagnostic system model of subunit-level characteristics shaping EHR adoption.

In the model, four types of adoption conditions represent the relevant heterogeneity among the subunits. We propose distinguishing between conditions that affect a subunit’s adoption tendency and those perceived as decisive for adoption. Following this distinction, we define four EHR adoption-influencing conditions at the subunit level:

1. The expectation of improved work system alignment with critical dependencies, which outweighs any compromise costs for the subunit, is a dominant adoption facilitator.

2. The expectation of (1) improved work system alignment or (2) low compromise costs both strengthen the tendency for subunit-level adoption and amount to a secondary adoption facilitator.

3. The expectation of work system misalignment with critical intra-departmental or environmental dependencies will block any possibility of unconditional full adoption and constitutes a dominant adoption constraint.

4. Expectations that there will be (1) no improvement in the alignment with the work system or (2) substantial compromise costs both weaken the tendency for subunit-level adoption, forming a secondary adoption constraint.

Critical dependencies play a decisive role. In terms of adoption, conditions are defined as those where the misalignment between the EHR and the work system components affected would threaten the department’s continuity. Three categories of critical work system dependencies were identified, each with a different meaning for the adoption of an organisation-wide EHR.

Interdepartmental dependencies – Depending on the grouping logic that led to their formation, departments vary in the extent of their reciprocal workflow dependencies on each other (Gattiker & Goodhue, 2004; Volkoff et al., 2005), and this may provoke different reactions to standardising and integrating their patient records across the hospital through an EHR. This ranges from departments whose core activities concern multidisciplinary care (oncology, paediatrics, geriatrics) through those that serve other departments or are dependent on other departments for their patient inflow (e.g., traumatology) to departments that only have to hand over their patients to others (e.g., emergency care), or can even serve the large majority of their patients independently of other medical specialities (e.g., oral surgery, ophthalmology). The more interdependent departments are, the more their managers expect to benefit from an EHR. Our results suggest that the more highly interdependent departments are willing to accept compromise costs by investing in work system adaptation rather than enact their decision autonomy. Here, a cost-benefit approach seems to be steering towards adoption rather than what we saw as the principle of least effort at work (Politi et al., 2022).

Intradepartmental dependencies - Internally, the staff of a clinical department share work routines and any information system needs to be aligned with this reality if it is to be used effectively (Ben-Assuli, 2015; Burton-Jones & Grange, 2012; Eisenberg et al., 2013; Jensen & Aanestad, 2007; Lanham et al., 2012; Laumer et al., 2016; Park et al., 2017). The proposed EHR’s integration of technical systems across departments, in terms of a shared data model and connected applications, faced vast differences among departments in terms of their existing technologies and internal social orders. We reported how departments with low interdepartmental dependencies had already optimised their internal processes, developed or acquired highly customised applications, or used advanced digital equipment supported by their IT vision and expertise. Such departments expected that even their best efforts to adapt their internal work system would still result in misalignments. Here, the principal of least effort rather than a cost-benefit approach may have influenced their adoption process: let’s not replace our optimised system(s) with one where we will be limited in reaping its benefits. In contrast, in other departments, managers felt that the EHR would enable intradepartmental alignment benefits, for example, by replacing the multiple systems they were using synchronously with a single system.

Environmental dependencies - A clinical department also functions within and is part of the wider social order constituted by its external task environment (Van Akkeren & Rowlands, 2007). The external environments of the various departments’ work systems can be very different. As departments vary in their dependency on external actors and structures, these differences weigh more heavily on some than on others. When a department has to meet external requirements that the EHR will not satisfy to be able to continue, this blocks fully compliant EHR adoption. In line with earlier research that focused on one specific medical speciality (De Benedictis et al., 2020; Eisenberg et al., 2013), we show how such requirements can be legitimately imposed by external parties (owners, contracted partner organisations, or financers such as insurers or government) or legally enforced (by law or by regulatory bodies or professional associations with legal sanctioning powers). Our results support the view that having distinct speciality-related responsibilities and legal and budgetary accountabilities encourages departments to develop their perceptions as professionals and to grant their managers a level of discretion when it comes to adoption (Petrakaki & Kornelakis, 2016; Scott & Davis, 2015). Additionally, we argue that in contrast with critical intradepartmental dependencies, misalignment is always a problem for the hospital’s management, who will either have to accept a department’s non- or partial adoption or commit to extra design costs to resolve the misalignment. To retain their external legitimacy, relationships, or resources, such departments need dedicated interfaces, or at least the ability to use compatible data models, to satisfy digital data exchange protocols.

Based on our research findings, only one of the three sets of interdependencies can be dominant in terms of criticality, and the model and related definitions are conceptualised to reflect this. First, a distinctive element in the department-oriented configuration is its relatively stand-alone character within the organisation. As such, by definition, the department will not have critical inter-departmental dependencies. Second, a department that, before implementation experiences both intradepartmental and external critical dependencies will already have aligned the intradepartmental ones with the, financially or legally, imposed external ones. As such, the model assumes the department has an externally oriented configuration. Third, a department experiencing incompatible interdepartmental and external dependencies before implementation will have established mixed arrangements (patient information and care delivery recording systems and procedures) to align with both. With a single uniform system being implemented, non-adoption will not be an option, given the criticality of the interdepartmental dependencies. At the same time, the criticality of their external dependencies, that by definition are imposed, overrides the hospital management’s authority. Consequently, for continued external legitimacy, the anticipated adoption outcome will be either acceptance of partial adoption or extra design costs so that the internal system can meet the external requirements. Despite this apparent clarity, we would caution that our diagnostic system model is analytical and may be less clear in practice, possibly leading to a hybrid situation.

5.2. Theoretical implications

Our research contributes to the general adoption literature, particularly the literature on EHR adoption, by increasing our understanding of specific subunit-level adoption issues.

In Section 2.1, we argued why neither the unilateral assumption used in organisational-level models nor the voluntarist assumption seen in individual-level models might be valid in an organisation-wide adoption setting with heterogeneous and powerful subunits. Such a setting is, perhaps paradoxically, simultaneously both mandatory and pluralistic in nature. However, in the hospital studied, the EHR’s basic model was largely implemented top-down through a “big bang” approach, implying that management expected full adoption. This approach reflects the unilateral assumptions of the organisation-level adoption models (Malaurant & Karanasios, 2020; Nielsen & Persson, 2017). This did not work out as hoped, and our cases demonstrated how negotiating additional compromise and design costs (Gattiker & Goodhue, 2004) would better fit such a setting. Healthcare organisations have been claimed to readily assimilate innovations because of their decentralised decision-making structures and semi-autonomous subunits (Greenhalgh et al., 2004). However, our study suggests subunit discretion can work in two ways depending on the adoption conditions. On the one hand, a subunit’s discretion may offer room for local work system adaptation to deal with compromise costs. Conversely, a subunit can use its local decision autonomy to not (or only partially) adopt and negotiate hard over design costs.

We add to the findings of Gattiker and Goodhue (2004) and Politi et al. (2022) by specifying how the heterogeneous adoption conditions differently inform negotiations concerning the trade-off between a subunit’s compromise costs and the design costs of the higher-level authority pushing the system. Moreover, our results suggest that the existing digitalisation level is not an indicator of a subunit’s experienced compromise costs. The required human change is limited for departments already working with decentralised digital client records. However, the perceived compromise costs in going from customised to standardised system support can still be high. Similarly, a department with a paper-based information flow could either welcome any organisation-wide system as an improvement or might feel that the compromise costs of a uniform system were still too high, given earlier investments in an optimised work system. Finally, although large departments were expected to be more influential in determining the new system, they could not always ensure it aligned with their critical dependencies. As such, in terms of the organisation’s design costs (Gattiker & Goodhue, 2004), the unique needs of a single large department may not outweigh the majority’s shared requirements.

The sociotechnical fit view adopted posits that although perceptions matter when it comes to adoption decisions, the technology limits adopters’ social constructions of the working and merits of a system (S. R. Barley, 1986; Sarker et al., 2019). In our situation, an EHR or similar organisation-wide systems can indeed prescribe the mode, the content, and the temporal aspects of work (Petrakaki & Kornelakis, 2016). Here, we contribute by showing how departmental adoption is blocked when these prescriptions do not align with a department’s critical work system dependencies. As such, our findings also demonstrate why the organisational embeddedness of a subunit prevents the direct translation of organisation-level theories to the subunit level. To summarise, by unravelling work system alignment issues (Alter, 2006, 2018), we have identified three subunit-level sources of heterogeneity that cannot be captured by applying organisation-level adoption models to the subunit level. The distinction between intradepartmental and interdepartmental dependencies, between interdepartmental and environmental dependencies, and between a department’s external dependencies and the organisation’s external dependencies. We contribute by adding environmental work system dependencies to the subunit-level dependencies previously recognised (Gattiker & Goodhue, 2004; Volkoff et al., 2005). In the health sector, with its increasing focus on network medicine, regional organising, and international quality assurance per patient stream, environmental work system dependencies can be particularly critical. We add to organisation-level IS adoption studies (Grant, 2003; Ludwick & Doucette, 2009; S. Strong & Volkoff, 2010) that subunit-level technology’s alignment with their critical work system dependencies is a condition for organisation-wide adoption.

We believe that our findings should be transferable to large service providers other than hospitals that are also characterised by heterogeneous, powerful subunits with a degree of autonomy derived from knowledge intensity and a professional workforce (Mintzberg, 1979; Von Nordenflyght, 2010). For example, in large consulting, advertising, accounting, and law firms, with separate business units or service departments, their professionals’ expert power and accountability may counterbalance central management’s authority. In such organisations, business systems may be introduced, such as ERP, CRM, knowledge management, or workflow management systems. If these IT artefacts are standard and rigid and do not match some sub-units’ critical work system dependencies, adoption problems may be likely. Nevertheless, hospitals may be a relatively extreme case in terms of their multi-logic grouping, which, as argued earlier, creates complex subunit heterogeneity that we have shown to make up a substantial part of the differences in critical work system dependencies.

5.3. Limitations and future research

We recognise that the study’s central sociotechnical concepts, namely a subunit’s work system’s properties, role in the change, and its subsequent adoption, were intersubjectively constructed. We adopted this approach to represent phenomena as experienced “out there” (Leonardi, 2011). Relevant differences among departments were derived inductively and later iteratively categorised, guided by our sociotechnical fit perspective. Thus, our theorising was grounded in both our data and the literature. Nevertheless, we could have overlooked some other subunit-level characteristics that also shape departmental adoption but were not voiced as such by our interviewees. Moreover, since we conducted this study in a single large hospital with semi-autonomous and heterogeneous departments, we acknowledge that other hospitals, especially smaller ones or those working in more centralised health systems, may experience different departmental dynamics when implementing EHRs (Petrakaki & Klecun, 2015). This may limit the transferability of the proposed model of subunit-level adoption conditions, but their rich description makes them amenable to further assessment and adaptation by extension or refinement in future research.

We worked with the premise that, within the time horizon of a project, the properties of a department’s work system architecture are relatively stable (Alter, 2006). Our findings show how expected changes in aligning a work system’s components facilitate and constrain adoption. This offers relevant information for managers that is accessible at the start of a project. Nevertheless, one might question how permanent work system properties are across the adoption stages in long-term projects. Indeed, an implementation process might even affect a subunit’s structural properties (Noble & Newman, 1993), which may need to change because deploying an organisation-wide system does not in itself create integrated and seamless organisational processes (Alter, 2018; Grant, 2003). Moreover, different work system properties might be relevant in different stages of a subunit’s adoption, as has been shown to be the case at the individual level (Sun & Jeyaraj, 2013).

This study focused on clinical departments’ adoption of an organisation-wide EHR. Off-the-shelf EHRs have specific adoption requirements, such as the need for a standardised data model. However, other organisation-wide IT artefacts may have other characteristics leading to different combinations of required design and compromise costs. Future research could focus on determining the design and compromise costs of other types of organisation-wide IT artefacts in professional service organisations that have heterogeneous and powerful subunits (Von Nordenflyght, 2010).

One avenue for future research is to investigate how the constellation of subunit configurations within an organisation influences effective adoption at the organisational level. A related question is how the organisation’s implementation strategy can be adapted to suit the constellation of subunit configurations. For example, could a more differentiated strategy prevent cases of non-adoption, such as those in the hospital we studied?

5.4. Implications for practitioners

Our findings concerning adoption conditions for clinical departments suggest that implementers of EHRs in hospitals should early on explore how the departments’ work systems compare with the EHR’s logic and identify local facilitators and constraints. Our model and accompanying heuristic (see Appendix 2) can guide implementers in identifying departmental work system dependencies and help departmental managers foresee their adoption-influencing conditions, independent of that department’s voiced support and participation during the implementation phase. Such a diagnosis could inform negotiations on trade-offs between compromise costs at the departmental level and design costs at the organisational level, and lead to department-tailored adoption strategies. Our study can highlight to central management that critical work system dependencies can legitimately constrain subunit-level adoption and that implementers therefore need to pay particular attention to the concerns raised by department-oriented and external environment-oriented clinical departments. Tailor-made solutions may have to be negotiated, adaptations of local work systems facilitated, or purposeful workarounds collaboratively designed such that departmental adoption involves the social construction of an aligned work system-in-use with a locally workable balance between technological standardisation and customisation.

6. Conclusions

This interpretive case study has unravelled how subunit-level heterogeneity in critical work system dependencies plays a dominant role in adopting an organisation-wide electronic patient record (EHR) system. The EHR adoption literature predominantly explains adoption at the individual or organisational level and fails to adequately address the nature of subunit heterogeneity and its implications for organisation-wide system adoption. Failure to address the alignment of critical work system dependencies at the subunit level can lead to unforeseen delays or (partial) non-adoption by some clinical departments, which may compromise the hospital’s effective EHR use. This study contributes to the EHR adoption literature by demonstrating subunit-level configurations of facilitators and constraints that influence the adoption of an organisation-wide EHR. We provide a context-specific diagnostic system model that can support hospital managers in identifying departmental adoption conditions when implementing an organisation-wide EHR. Further research could show whether the model is also applicable to other organisations with heterogeneous and powerful subunits that are adopting organisation-wide systems.

Acknowledgements

We thank our former MSc BA students Anneleen van Dijken, Marijke Haaksema, Hans-Peter Palland, Mirjam Postma, and Thijs van de Woestijne, for their support in the data collection and initial analyses. We are greatly indebted to the project members and hospital employees for their trust and willingness to participate.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/0960085X.2023.2225786

Appendices Appendix 1

– Codebook

Table

Code	Definition	Example
EHR’s alignment with work system components
Intradepartmental work dependencies: hard technical capabilities(Technology)	A subunit’s existing advanced or complex technologies (e.g., equipment, software, hardware, operating systems, platform use) prompt a need for interfaces or integration with the new EHR.	“Our patients are connected to machines and we need to measure all kinds of parameters. We want to register that in one system” (E2).
Intradepartmental work dependencies: soft technical capabilities(Participants)	(1. Expertise/vision): Presence of one or more of the following elements: local IT expertise; local IT vision; local IT strategy for the future.(2. Alignment): Perceived alignment of the local IT vision or IT strategy with the EHR implementation.	1. “We have our IT workgroup, and to some extent we already work digitally … and we have also tested our way of working with new computer systems” (C2)2. “I had come to understand, from a different care setting, the benefits of integrating medical photographs in such a system. I already mentioned this when I was a member of our department board: this is our future where computerisation makes our life easier rather than leading to more mouse clicks” (D2)2. “I had come to understand, from a different care setting, the benefits of integrating medical photographs in such a system. I already mentioned this when I was a member of our department board: this is our future where computerisation makes our life easier rather than leading to more mouse clicks” (D2)
Intradepartmental work dependencies: tasks and activities(Process)	Department evaluates that: … its current IT system is inefficient, unreliable, and/or fragmented. … its current workflow is well-designed but vulnerable to changes. … the work processes would benefit from greater uniformity through her. … the current system insufficiently supports the work processes. … the EHR will not match its work processes or approach.	1. “The problem with the current system is that when a patient is under care for a long period you end up with a big mess of sheets of paper and then it’s difficult to find the right one” (B1).2. “We have around 35 steps in our process … and we are particularly interested that the system can follow a patient through our process, that is, how we have organised it logistically” (C3).3. “Patient care benefits from good alarm signals and the standardisation of specific actions to be taken so that they can’t be overlooked” (F3).4. “We look forward to a new system. The limitations of the old system really get in our way” (G3).5. “In our department, the specialist’s diagnosis cannot be phrased in let’s say three words. I am afraid that the underlying arguments will get lost in the new system. A doctor will choose an option. I consider that a disadvantage” (A2).2. “We have around 35 steps in our process … and we are particularly interested that the system can follow a patient through our process, that is, how we have organised it logistically” (C3).3. “Patient care benefits from good alarm signals and the standardisation of specific actions to be taken so that they can’t be overlooked” (F3).4. “We look forward to a new system. The limitations of the old system really get in our way” (G3).5. “In our department, the specialist’s diagnosis cannot be phrased in let’s say three words. I am afraid that the underlying arguments will get lost in the new system. A doctor will choose an option. I consider that a disadvantage” (A2).
Interdepartmental dependencies (Information)	The number and intensity of workflow dependencies with other subunits – ranging from stand-alone to highly dependent. Dependencies exist in three ways: Multidisciplinary care (Yes), Treating other departments’ patients (Yes), Monodisciplinary care (No).	1. “We have collaboration teams; for example in one team we work together with at least eight different professionals, so we really need to be able to deliver multidisciplinary patient reports” (B1).2. “Our patients are hospitalised in various departments throughout the hospital. If similar ways of diagnosis are used, we can benefit from that” (A3).3. “As a department, we do play a role in diagnosing patients from other departments … However, in general, it’s fair to say that we are able to work independently of all the other departments” (C3).2. “Our patients are hospitalised in various departments throughout the hospital. If similar ways of diagnosis are used, we can benefit from that” (A3).3. “As a department, we do play a role in diagnosing patients from other departments … However, in general, it’s fair to say that we are able to work independently of all the other departments” (C3).
Environmental work dependencies(Environment-Information)	The extent and intensity of workflow dependencies in the supply chain (i.e., with healthcare providers outside the organisation) or other environmental dependencies (e.g., professional association, sector association).	We work for a large care region. A number of our patients only get a specific treatment here and then go elsewhere for further treatment. We have many referrals from a large region and also doctors from our speciality are based in almost all the hospitals in the region” (B1).
	The enforcement by laws and regulations of requirements on the collection, processing, and archiving of patient data.	“Regulations require us to enter patient data in a specific way. We just want to have that integrated in the new system” (G3).
Role in change process:
Trust in organisation-wide implementation	Departmental trust in organisation-wide implementation, informed by past change history, current change leaders, or other hospitals’ achievements: Optimistic; high expectations Wait-and-see; neutral Express worries or low trust	1. “Yes, I have confidence in the plan … the structure is well organised” (H2).2. “ … it’s like ‘do we get a new system?’, well we’ll see” (B1)3. “ … this earlier experience [former initiatives to prescribe digitally], makes us very worried … if this happens again, ‘we go down’” (C2).2. “ … it’s like ‘do we get a new system?’, well we’ll see” (B1)3. “ … this earlier experience [former initiatives to prescribe digitally], makes us very worried … if this happens again, ‘we go down’” (C2).
Departmental involvement in implementation	Perceived departmental supportive involvement in implementation: Proactive Reactive Passive	1. “ … we have to get on a roll to prepare. One of us [a researcher] will throw himself at it” (D1).2. “ … not actively approached … but we do invite them [representatives of the project] to our management meetings” (E3).3. “ … there is nobody busy with it, at least I don’t hear anything about it. It’s not a topic on the agenda” (B1).2. “ … not actively approached … but we do invite them [representatives of the project] to our management meetings” (E3).3. “ … there is nobody busy with it, at least I don’t hear anything about it. It’s not a topic on the agenda” (B1).
Departmental influence on design	Perceived departmental influence on design, inferred from size or own status. Feel heard Experience no or little influence	1. “There are clear communication lines. I’m the core-team member [in project] … and together with one of the department staff members, we are the contact persons” (D1).2. “Our requests, from anyone, about ‘how does it look like?’ or ‘how does it work?’ are really being warded off” (A1).2. “Our requests, from anyone, about ‘how does it look like?’ or ‘how does it work?’ are really being warded off” (A1).
Adoption:
Intended adoption	A board and key representatives’ voiced motivations to integrate the EHR in their operations, ranging from intended enthusiastic acceptance to non-adoption.	(enthusiastic) “The bottom line is that it will provide us with great advantages compared to the current situation” (D2).(compliant) “Overall, I think, we are open to the renewal. And, as far as the department is concerned, we don’t moan that much. We’ll see what happens and we will do the best with what we get. […] It has happened before, that we just get something, and then are able to work with it in one way or the other” (A1).(limited or conditional) “We will not make full use of the system, but we will use the connections [with their own system]” (E3).(non-adoption) “There are two options: we’re completely bankrupt … or we have a totally different system [as a department] that we can work with” (C2).(compliant) “Overall, I think, we are open to the renewal. And, as far as the department is concerned, we don’t moan that much. We’ll see what happens and we will do the best with what we get. […] It has happened before, that we just get something, and then are able to work with it in one way or the other” (A1).(limited or conditional) “We will not make full use of the system, but we will use the connections [with their own system]” (E3).(non-adoption) “There are two options: we’re completely bankrupt … or we have a totally different system [as a department] that we can work with” (C2).
Realized adoption	The uptake of the EHR, in terms of the department’s patient data recording and management system being used on a regular basis, and whether this is fully compliant in terms of no additional costs.	(adoption) “ … we are catching up now, and I’m happy with it [the EHR]” (A4).((partial) non-adoption) “For department G, we had to decide against implementation. A proper solution could not be found” (project manager).(department-level customised extras released) “For this department, we now have some templates, so that we [the department] can work in a uniform way” (B4).(local, collective or individual workarounds) “In this section [of the system] we register our own speciality remarks … we do not use it in the general way” (C5). ((partial) non-adoption) “For department G, we had to decide against implementation. A proper solution could not be found” (project manager). (department-level customised extras released) “For this department, we now have some templates, so that we [the department] can work in a uniform way” (B4). (local, collective or individual workarounds) “In this section [of the system] we register our own speciality remarks … we do not use it in the general way” (C5).

Appendix 2

- Heuristic to anticipate heterogeneity in departmental adoption×.

Table

	Facilitators:		Constraints:
	Dominant	Secondary	Dominant	Secondary
	Department’s expectations:		Department’s expectations:
Work system component:	… that improved alignment with critical work system dependencies will outweigh any compromise costs.	… low compromise costs or benefits (improved work system alignment).	… that misalignment with critical work system dependencies will block the possibility to adopt, leading to either design costs or non-adoption.	… compromise costs or limited benefits (i.e., reduction in work system misalignment).
Technology		Current technology is inefficiently or inadequately supporting the work system.Local technologies will benefit from EHR-enabled interfaces or integration.	Current critical advanced or complex technologies are misaligned with the EHR, e.g., lack of interfaces or integration
Participants		EHR aligns with local IS expertise and IS change vision.		EHR misaligned with local IS expertise and IS change vision. Local technologies better align with the IS vision than the EHR.
Information	Depends on other departments (s) for patient input or output, and thus the information needs to be well-aligned.	Current patient administration is inefficient, unreliable, and fragmented.	EHR’s data model constrains the department’s digital prescribing.	A substantial part of the data model is superfluous and will create inefficiencies.
Process	Work processes make it critical to use the same EHR throughout the hospital.Interdepartmental dependencies that are critical for the department’s workflow will be better supported.	Work processes will benefit from EHR-enabled greater uniformity.		EHR misaligned with (currently optimised) workflows or work methods. Limited interdepartmental work dependencies (pooled or sequential). Mainly monodisciplinary care.
Environment		Support for work dependencies in the supply chain. Support for network collaboration. EHR aligns with legal or professional requirements.	Legally or financially imposed requirements do not align with the EHR.	Environmental interdependencies in alliances on regional or (inter)national levels misaligned with the EHR.

*The conditions specified in the cells are based on this study’s results and may be context-specific.