Effective system use expectations of hospital
departments during implementation of an EHR
Msc BA Change Management University of Groningen Faculty of Economics and Business
February, 2018
Supervisor: dr. J. F. J. Vos Co-assessor: dr. M.A.G. van Offenbeek
By: Thijs van de Woestijne Eeldersingel 14 9726AR Groningen t.van.de.woestijne@student.rug.nl
Student number: S2030020
3
Acknowledgements
4
Table of Content
Acknowledgements ... 3
Abstract ... 5
1.
Introduction ... 5
2. Theoretical background ... 7
2.1. Theoretical lens: representational theory perspective ... 8
2.2. More specific theoretical approach: theory of affordances ... 9
2.3. Content, process and context of change ... 10
3.
Methods ... 11
3.1. Specific empirical context ... 11
3.2. Case study approach ... 13
3.3. Data collection ... 13
3.4. Data analysis ... 16
4.
Results ... 17
4.1.
Within-case analysis ... 17
4.2.
Cross-case analysis ... 26
4.2.1.
Evaluation of affordances ... 26
4.2.2.
Evaluations of change process ... 30
5.
Discussion ... 32
References ... 37
Appendix I: Codebook ... 41
Appendix II: Interview protocol ... 45
5
Abstract
Studies about effective Information System (IS) use are often conducted on an organizational level or individual level of analysis, while a departmental level of analysis might be particularly relevant for studying effective IS use. This study uses a representational theory perspective to analyze effective use expectations of hospital departments before the ‘go-live’ of an Electronic Health Record (EHR) in a large teaching hospital. Affordance theory is used to analyze differences between departments in evaluations of system functionalities. Possible effects of both change content and process are analyzed. Analysis shows that departments differ in requirements and context leading to different perceptions of system affordances. An affordance can be expected to cause concrete benefits for one department, and expected to cause counter-beneficial outcomes for another department. This study contributes to affordance theory by showing that a relationality might exist between department and IT artifact. Before ‘go-live’ of an EHR, perceptions of affordances are partly a result of the relationship between system content and departmental context and requirements. Patient information needs, data exchange needs, patient turnover and task delegation differ per department and possibly affect affordance perceptions.
Keywords: Effective system use, representational theory perspective, theory of affordances, Information Systems, EHRs.
1. Introduction
Healthcare costs are rising and innovation in healthcare is necessary to maintain quality of care and reduce inefficiencies. A potential mean to innovate is the implementation of Electronic Health Records (EHRs) in hospitals. Implementing an EHR potentially improves quality of healthcare (Zhou et al., 2009; Miller & Sim 2017). An EHR is an information system (IS) that enables longitudinal collection of patients health information digitally, streamline healthcare data flow and share patient information across different healthcare settings (Gunter & Terry, 2005; Rashbass, 2001). It is a mechanism for integrating healthcare information which is collected in both paper and Electronic Medical Records (Gunter & Terry, 2005). Potentially an EHR system reduces costs (Hillestad et al., 2005), increases patient safety (Koppel et al., 2005), and leads to more organizational integration (Barki & Pinsonneault, 2002). Nevertheless, achieving benefits from implementation of a technological system is a difficult endeavor (Markus, Axline, Petrie, & Tanis, 2000; Staehr, Shanks & Seddon, 2012).
6
might be different than an organizational goal for using that system. Therefore, effective system use might be different depending on the level of analysis. Individual effective use may differ from departmental or organizational effective use.This study focuses on effective EHR use of departments in a hospital. There are two reasons why studying effective use on a departmental level is particularly relevant. First, departments in an organization can be affected by an information system in dissimilar ways (Wagner, Newell & Piccoli, 2010). To make effective use of systems, the system needs to be in good fit with organizational characteristics such as culture and structure (Markus & Tanis, 2000). The culture and structure of departments might differ, resulting in different fits with the system between departments. Whether implementing an information system leads to benefits for a department might partly depend on characteristics of that department (Gattiker & Goodhue, 2005). Second, organizational goals for using a new system might conflict with departmental goals. For instance, the organizational goal of standardization for implementing an EHR can conflict with the preference of departmental members to exercise a level of discretion (Petrakaki & Kornelakis, 2016).
While there are reasons for studying effective system use on a departmental level, to the knowledge of the researcher no studies are conducted on effective system use of departments. Literature is either focused on the individual level of analysis (Agarwak & Karahanna, 2000; Marcolin et al., 2000; Benight & Cieslak, 2011) or the organizational level of analysis (Ahmadi, Nilashi & Ibrahim, 2015; Burton-Jones & Volkoff, 2017). By studying effective system use on a departmental level this gap in the literature is addressed.
The theoretical background for studying effective use consists of an elaboration of the representational theory perspective of effective use (Burton-Jones & Grange, 2013) and the theory of affordances (Volkoff & Strong, 2013). The representational theory perspective is used as scientific lens to determine what constitutes effective system use. The theory of affordances is used as a more specific theoretical approach to analyze effective use of EHR functionalities. The assumption of this theory is that information systems come with multiple functionalities which are a necessary but not a sufficient condition for effective use. An affordance is a potential for action which may or may not occur (Markus and Silver, 2008). Differences and similarities between departments in evaluation of system functionalities are analyzed.
7
studied. The process of change is analyzed to discover differences in capabilities to leverage affordances. This study focuses on the departmental context instead of the organizational context. The research question therefore reads:“How and why do the content of an EHR and the implementation process of an EHR lead to different expectations of effective system use between hospital departments?”
The implementation of an EHR in a large teaching hospital in the Netherlands is analyzed. The analysis is conducted during the implementation period till the moment of ‘go-live’. During the implementation period, future users are being prepared to work with a new system (Markus, 2004). Departmental members make evaluations about the outcomes of the new system in comparison to the outcomes of the old system (Joshi, 1991). Departmental members evaluate effectiveness of the new system for their department as opposed to the situation before the ‘go-live’ of the system.
This article contributes to effective system use literature by taking a departmental level of analysis. In previous literature affordances are treated as outcomes between organizational features and IT artifact (Pozzi, Pigni & Vitari, 2014) or between user and IT artifact (Bernhard, Recker & Burton-Jones, 2013). Analysis showed that departments can have vastly different requirements, which shines a different light on how and why affordances are actualized. The next session presents the theoretical lens of representational theory and the theoretic approach of affordances in more detail. Thereafter, main results of the within-case and cross case are presented. The article ends with a number of propositions about effective system use of EHRs.
2. Theoretical background
Unlike the large amount of studies on IS use or acceptance, studies to effective IS use are rather scarce (Burton-Jones & Grange, 2013). The scientific field for studying effective use finds its roots in the work of DeLone and McLean (1992; 2003) who studied and defined information system success. Measuring success by the amount of use of an IS does not measure whether or not implementation of the system leads to organizational benefits: “Simply saying that more use will yield more benefits without considering the nature of this use and context is clearly insufficient” (DeLone & McLean, 2003, p. 16). Effective use is not only using a system, but using a system in a way that helps to attain desired goals (LeRouge, Hevner & Collins, 2007; Burton-Jones & Grange, 2013). Various goals exist during EHR implementation and different stakeholders within a hospital can have very different goals (McGowan, Cusack & Poon, 2008; Seidel, Recker, Vom Brocke, 2013). Looking solely at the attainment of goals is therefore not enough to analyze effective use. Analysis should also focus on which system functionalities are and which functionalities are not useful for specific user groups (Markus & Silver, 2008; Seidel et al., 2013).
8
& Grange, 2013). It is not sufficient to analyze effective use in a specific context in detail. To complement this abstract perspective, the more specific theory of affordances is presented. This theory provides insights in how functionalities of an IS can or cannot be actualized into concrete results (Strong et al., 2014). The last part of the section contains literature about influences of the content, process and context, which are pillars for change success (Walker, Armenakis & Berneth, 2007).2.1. Theoretical lens: representational theory perspective
The central idea of the representational theory perspective is that the most basic function of all information systems is the representation of a real-world phenomenon (Burton-Jones, Recker, Indulska, Green, & Weber, 2017). The purpose of having an information system is that the IS is providing information about a certain domain which otherwise had to be obtained without an IS (Burton-Jones et al., 2017). People use information systems because obtaining information from an information system is more efficient than obtaining information without one (Wand & Weber, 1995). An information system is useful when the representation obtained is truthful (Burton-Jones & Grange, 2013). IS use is more effective if representations are easier to obtain, if representations are faithful, and if representations are understandable (Burton-Jones & Grange, 2013). To be faithful, a representation must be complete; the system represents all parts of the domain which are relevant to user needs (Wand & Wang, 1996). EHRs are systems that provide decision support, communication tools and many other functionalities (Van Ginneken & Van Eekeren, 2010) but this is all based on data classified in tables, graphs and fields which is more or less faithfully representing a real world domain (Parson & Wand, 2008).
9
2.2. More specific theoretical approach: theory of affordancesIn line with this theoretical lens is the theory of affordances, which can be used to analyze effective system use in a specific context (Burton-Jones & Volkoff, 2017). Looking at potential goals of implementing EHRs alone does not give an adequate picture of effective system use. Perceptions of increased patient safety might be due to usage of a particular part of the system, while working around other parts of the system might possibly decrease patient safety. By using affordance theory, the parts of the system which are used effectively and the parts of the system which are not used effectively can be categorized (Strong et al., 2009).
The concept of affordances derives from ecological psychology and has become of growing interest in IS research (Pozzi, Pigni, & Vitari, 2014). The concept is developed by Gibson (1986), an ecological psychologist, who defined affordances as ‘what is offered, provided, or furnished to someone or something by an object’ (Strong et al., 2014, p. 55). Recently the concept of affordances gained attention in IS research because of its usefulness for explaining the consequences of IT use in organizations (Pozzi et al., 2014). Affordances in IS research are defined as “the potential for behaviors associated with achieving an immediate concrete outcome and arising from the relation between an artifact and a goal-oriented actor or actors” (Strong et al., 2014, p. 69). In short, affordances are possibilities for action (Markus & Silver, 2008). An affordance is a functionality of a technological system which is or is not actualized. For analyzing effective system use it can be used to see which components of an EHR can be leveraged to gain benefits and which components cannot be leveraged for benefits (Bygstad, Munkvold, & Volkoff, 2016).
The term affordances is adapted in IS research by Hutchby (2001) to conciliate the two opposing views of constructivism and determinism (Zammuto, Griffith, Majchraz, Dougherty, & Faraj, 2007; Leonardi, 2011). Constructivism assumes that technology is dependent on social constructs (Zammuto et al., 2007). Technology is given meaning by consciousness and shared understandings (Myers, 1997). Determinism assumes an opposite relation between the social and the technological and treats technology as an objective force which determines organizational aspects as routines and structures (Orlikowski, 1992). The concept of affordances recognizes the bidirectional relationality between technology and actor (Strong & Volkoff, 2013). Technology and actor influence each other: ‘people's practices and their use of technology in a setting is shaped, but never fully determined, by the setting's physical and social characteristics’ (Fayard & Weeks, 2014, p. 237). Affordance theory suggests that although IS and organization features may exist independently of each other, their value for explaining organizational form and function comes from how they are enacted together (Leonardi, 2011; Zammuto et al., 2007). Therefore, patterns of technological system use can only be understood by referencing both organizational characteristics and technological features (Majchrzak & Markus, 2013).
10
relation between them (Strong et al., 2014). For an affordance to become useful, it needs to be recognized by a user or group of users as functional. A spoon is useful for people who recognize a spoon as helpful in attaining the goal of eating, while the same spoon might not be useful for people who do not recognize its functionality, such as infants. The content of a system might be evaluated differently between different groups of users, depending on whether or not they perceive affordances as helpful in attaining benefits. Second, not only do the users need to recognize the potential benefits of using an affordance, they should also be capable to actualize the affordance into a benefit (Bernhard, Recker, & Burton-Jones, 2013). A group of users should be prepared accordingly to work with a system effectively. Third, the technological and organizational context gives rise to a variety of mechanisms that may act as conditions that initially enable or constrain the actualization of the affordance (Bygstad et al., 2016). Thus, actualizing affordances of a system depends partly on context factors. All in all, an affordance can be described as a necessary but not sufficient condition for effective use, depending on whether or not the affordance is actualized (Markus and Silver, 2008).For studying effective use of an EHR system in the context of different departments during the implementation phase, affordances mean the developed functionalities1 of an EHR system that can be actualized to realize outcomes (Burton-Jones & Volkoff, 2017). From literature about EHR systems in the Netherlands (Van Eekeren & Polman, 2010; Huysman, 2012), and previous literature about affordances of an EHR (Strong et al., 2014) three possible affordances are presented. This is far from a complete list of affordances but these three functionalities can be viewed as most applicable to this specific case (see section 3.1 for list of affordances). These affordances are viewed as most important EHR affordances (Van Eekeren & Polman, 2010) or particularly relevant to the case.
2.3. Content, process and context of change
To leverage affordances of an EHR, both system content and process of implementation are important. Furthermore, different contexts in which the change takes place may lead to dissimilar evaluations about future effective system use. Between departments an EHR might have different effects on departmental outcomes (Wagner et al., 2010). Also the implementation method might influence effective system use expectations of departments.
Most studies about the influence of IS content on effective system use are carried out on an individual level of analysis. Fit between user task and technology is analyzed (Goodhue & Thompson; 1995; Marcolin et al., 2000; Barki, Titah, & Boffo, 2007). These studies acknowledge that use does not necessarily lead to positive outcomes. Only when the technology is in good fit with the task it supports, a new IS has a positive impact (Goodhue & Thompson, 1995). Task in this model regards to
1 Affordances can also emerge during use of the technological system from the recognition of system users of
certain parts of the system which are not previously recognized or designed by developers (Van Osch &
11
the goals and deliverables of actors within an organization (Lyytinen & Newman, 2008). However, tasks and deliverables of users between departments might differ. Therefore, the question resides whether there are differences between departments within a hospital in their evaluation of system content.There are two ways in which the implementation process influences expected effective use, the amount of learning and adapting of the system (Burton-Jones & Grange, 2008; 2013; Barki et al., 2007). Learning can have two forms. Learning to work with the system and learning about the system (Barki et al., 2007). By learning to work with the system users become more capable and skilled to leverage an affordance into a concrete outcome (Strong et al., 2014). Learning about the system is the amount of information and overview users have of the system. By learning about the system, users know better what is and is not possible while working with the system (Burton-Jones & Grange, 2013).
The second driver for effective use is adapting. This is based on the assumption that there are no systems that give a totally correct representation of a domain (Burton-Jones & Grange, 2013). An actor can adapt the surface structure or physical structure to engage with the deep structure more easily, or can adapt the deep structure to make it more truthfully represent a domain (Burton-Jones & Grange, 2013). By adapting an IS it can become more in line with tasks users have to perform (Barki et al., 2007). The question is whether differences exist between departments in the possibilities to learn and adapt the system, causing different expectations of departments about effective system use.
The level of analysis in this article is the departmental level. Previous studies that researched the fit between content and context were carried out on an organizational level of analysis (Soh, Kien & Tay-Yap, 2000), were organizational-specific requirements were compared with capabilities of EHR packages. However, within organizations various practices exist which can be divided by departmental borders, influencing how IS systems are adopted (Wagner et al., 2010).
3. Methods
In this section, methodological issues of the study are addressed. The section starts with an elaboration of the specific empirical context in which the research is conducted. The remaining of the section describes how the research is conducted and why certain methods are chosen.
3.1. Specific empirical context
12
input for the remaining of the research. The list was based on literature of EHR systems, analysis of documents provided by the specific case, and conversations with employees of the implementation team from the specific case. This list was far from complete and gives a simplified image of the large and complex EHR system but was necessary to increase the focus of the study.The most clear hospital goals for implementing an EHR are increased patient safety and improved efficiency (Huysman, 2012). Patient safety increases because medical data does not have to shift between different data records anymore causing a reduction of data errors (Van Luxemburg & Van Eekeren, 2010). Efficiency increases because medical data is easier and quicker to obtain and data can be used multiple times (Van Luxemburg & Van Eekeren, 2010). Another goal for most hospitals for implementing an EHR is to have a more patient centered approach (Lorenzi, Kouroubali, Detmer, & Bloomrosen, 2009; Huysman, 2012). EHRs potentially cause quicker communication between healthcare professional and patient and makes new communication methods possible (Lorenzi et al., 2009; Baron, Fabens, Schiffman, & Wolf, 2005). Thus, effective use of an EHR can be defined as using an EHR in a way that helps to increase efficiency, patient safety, or patient centeredness. The question resides whether or not these organizational goals for implementing an EHR are equally important for departments to adopt an EHR.
The first and most basic affordance is the structural record keeping of data in a common database. An EHR supports uniformity of entering data in an integrated record for all data which is relevant for treating a patient (Van Eekeren & Polman, 2010). This data is stored in a common database and can be used multiple times and exchanged with internal and external stakeholders (Strong et al., 2014). Potentially this affordance increases patient safety and efficiency in processes (Strong et al., 2014; Huysman, 2012)
The second affordance is order management; the electronic communication between the healthcare professional and medical supporting processes (Van Eekeren & Polman, 2010). Many different specialisms and other hospital actors contribute to the treatment and care of patients. This results in an order relation. For instance, the action diagnosing implies an order relation between medical specialist and laboratory (Van Eekeren & Polman, 2010). Implementing an EHR makes the order relation more visible by an audit trail of what was done, by whom, and when (Strong et al., 2014).Having structural, electronic communication between healthcare professionals potentially reduces medical errors and leads to more effective communication.
13
between patient and healthcare professional changes because patients become more involved in the care process (Strong et al., 2014).These three goals and affordances have been used as input for interviews with healthcare professionals. By having three pre-set goals and affordances, it was possible to make comparisons between hospital departments. The method of how the research was conducted is described in the following paragraphs.
3.2. Case study approach
A single-case, in-depth qualitative research in a large teaching hospital was conducted to explore how the process and content of change influence expectations of future effective use of multiple departments. A single-case study was appropriate because the question asked was ‘how’ and ‘why’ differences exist (Yin, 2013). Another reason why a single-case study was appropriate is that the context was an important part of the study. If general circumstances of the phenomenon are analyzed and boundaries between phenomenon and context are blurry, a single-case study suits the research (Yin, 1994). A third reason why a single-case study was chosen is because the case represents a rare event. Implementation of an EHR in a large hospital can be seen as a rare event and this is in line with case-study research (Yin, 2013). Because the studied phenomenon was not fully addressed in literature, the research goal was theory development (Aken, Berend. & Van der Bij, 2012). The research process was therefore based on the model of building theories of Eisenhardt (1989).
This explorative research took place during the implementation of a new Electronic Health Record (EHR). During the period of research, future users of the system received training and the ‘go-live’ of the system took place. Data was gathered from three months before ‘go-‘go-live’ till a week after ‘go-live’. During this period, the researcher was actively present at the hospital.
3.3. Data collection
14
Selection of participantsBecause analysis was done at a departmental level, it was important that enough departments were selected and that these selected departments were significantly different from each other so that departmental context factors could be highlighted (Yin, 2013). The first step in the selection process was to create a list of departments suitable for research, resulting in a list of eight departments. These departments had been previously researched in a study at the departmental level of analysis and therefore found suited for this research. Healthcare professionals with a leading function at this eight departments were contacted and asked if their department was willing to participate. Five of these contacted persons reacted positively. These departments differed in terms of work processes, size, type of medical specialists, and other factors.
The selection of participants for semi-structured interviews at these selected departments was non-random. It was important that at least one, but preferably two participants per department had a leading or coordinating role because these participants were expected to have a good overview of the effect of the new system for their department. Therefore, members of the department with a leading function were asked to participate. The goal was to include participants with different degrees of involvement in the change process. The method for selecting these participants was to ask the member with a leading function to select two other departmental members. One with a large involvement in the change process and one with limited involvement in the change process. The assumption was that the participant with a leading function had a better view of who was suitable for an in-depth interview. This resulted in the selection of five participants with a leading function, three with a coordinating function, four superusers, and four participants with limited involvement in the change process. Furthermore, included participants had different roles. Medical specialists, nurses, medical administration, and one manager were included in the list of participants. This selection method resulted in the selection of 15 participants coming from five departments, with different degrees of involvement in the project and from different professions (Table 1). High involvement in the change process means that the participant had a leading role within the department in realizing the change. Medium involvement means that the participant was a superuser. Low involvement means that the participant only received training.
Type of data
15
The interviews were semi-structured. An interview protocol was designed in advance (Appendix II & III). By conducting an interview protocol in advance, standardization of the results enhanced (Van Aken, Berends, & Van der Bij, 2012). Similarities between interviews increased and the same type of topics were asked (Emans, 2002). Furthermore, by using a protocol, replicability of the study improved (Eisenhardt, 1989). However, the protocol was not strictly followed, to leave room for other topics to emerge during the interview and to focus on important issues of the department. Some departments had very different issues than others. Moreover, involvement and role influenced the way the interview proceeded. More involved health care professionals had a better view of the process and therefore more detailed questions about the process could be asked. Nurses had a different preparation for working with the EHR than medical specialists, which led to different interviews. By using semi-structured interviews, more relevant questions were asked. Nevertheless, it was made sure that certain topics were always treated during an interview. The protocol of the semi-structured interview was based on literature, primary data observations, input from EHR specialists working at the EHR implementation team, and secondary data (see Appendix II for interview protocol).Other forms of primary data collection were day-to-day observations, observations during trainings, observations during demos and superusers meetings, observations at the researched departments the days after ‘go-live’ and conversations with employees of the EHR implementation teams. The role of the researcher was shifting from more participatory (observations during trainings) to more observatory (observations during demo’s and superusers meetings) (Babbie, 2010). There was no clear theoretical framework or pre-coding scheme for these observations. Therefore, observations were unstructured in nature (Jupp, 2006). There were multiple goals for collecting these data. A first goal was to use these data as input for the selection of affordances and goals. A second goal was to gain a better understanding of what constitutes effective system use. During training, day-to-day observations and conversations, knowledge was gained about how the system should be used to increase patient safety, efficiency and patient centeredness, e.g. healthcare professionals need to use standardized buttons instead of free text to increase efficiency of data sharing and uniformity of data input. A third goal was to increase information about departmental expectations of effective system use, by observing departments after ‘go-live’.
16
There were multiple sources of secondary data. The most information came from classified documents about the implementation of the EHR. Analyzed documents were for instance an action plan, a communication and implementation report, and organizational scope decision reports. Furthermore, non-classified data was analyzed, found on the intranet page of the hospital and in EHR magazines for future users. In total twelve reports and documents were analyzed. Secondary data was analyzed by making summaries of the written reports and documents. This resulted in a document of notes containing 2375 words. Secondary data was used in two ways. Firstly, a list was created of possible goals for using the system, which served as input for the eventual selection of three goals. Secondly, knowledge about change process increased which served as input for semi-structured interviews with healthcare professionals. For reasons of confidentiality, most secondary data was not appropriate for publication in the result section.3.4. Data analysis
Analysis of semi-structured interviews consisted of multiple steps. The first step was to get the raw data from recordings on paper. All the interviews were recorded and transcribed leading to a document of interview transcripts of 92202 words. Interviews were transcribed literally. The transcripts were then send back to the interviewee who could check the transcript and comment on the transcript. This resulted in three changes in the transcripts. Transcripts were read multiple times to increase familiarity (Eisenhardt, 1989; Yin, 2013).
A second step was to form deductive codes based on literature presented in the theoretical background of this paper. Deductive codes were used to compare affordances and goals from literature and interviews (Miles & Huberman, 1994). An example of a deductive code is order management: the electronic communication between the healthcare professional and medical supporting processes (Van Eekeren & Polman, 2010). Extra codes were added based on user perception about the element of the process or content of change; positive or negative. In this way it became easier to group positive and negative evaluated aspects of the system and implementation method.
The next step was to code transcripts using the program Atlas.ti with the deductive codebook. During coding, inductive codes were formed and adjusted in the codebook. An example of an inductive code is change task: perception of the degree of change in tasks and processes after EHR implementation. The codebook can be found in appendix I. Most interviews were coded twice. This was done because the researcher started coding before all interviews had taken place, to have an overlap between collection, coding and analysis of data (Glaser & Strauss, 1967; Eisenhardt, 1989). After all interviews were conducted and all transcripts were written down, all interviews were coded again with a selection of most prevalent inductive codes.
17
Per department, information was categorized, using the family manager function of Atlas.ti. During cross-case analysis, differences, similarities and patterns between departments were unraveled (Eisenhardt, 1989). The method for pattern seeking was to select particular codes in Atlas.ti. A list of quotes of that particular code was thereby generated with information about which department the quote originated from. With this method, comparisons between departments were made. Patterns were revealed by comparing departmental context factors and evaluations of the system and implementation method.Table 1. Overview of research participants
4. Results
This section comprises of a within-case and a cross-case analysis. In the within-case analysis, departments are treated as unique cases, and fit between affordances and departmental characteristics is given. Effects of the implementation method for the particular department is summarized. In the cross-case analysis, patterns of effective system use expectations are unraveled. Possible influences of departmental requirements on expectations of effective system use are presented.
4.1. Within-case analysis
In this section, results are presented from the analysis of five researched departments. Expectations about future effective use, evaluations of affordances, and positive and negative aspects of the implementation process are given.
Department A is a large unit of 269 fte. divided in four sub-units. One sub-unit has twelve medical specialists, one sub-unit has eight medical specialists and there are two sub-units of six medical specialists. The sub-units have different nursing wards and one sub-unit is sharing its nursing ward with department B. The department has one department team for customization of department specific content. One member of the department is also member of the team that makes decisions about the
Department Code Occupation Role Involvement Time
A A1 Assistant Superuser Medium 37 min.
A2 Medical specialist Leading High 52 min.
A3 Nurse Coordinating Medium 43 min.
B B1 Medical specialist Future user Low 47 min.
B2 Manager Leading High 59 min.
B3 Nurse Superuser Medium 39 min.
C C1 Medical specialist Leading High 62 min.
C2 Medical specialist Superuser Medium 50 min.
C3 Physician in training Future user Low 39 min.
D D1 Medical specialist Leading High 50 min.
D2 Medical specialist Future user Low 32 min.
D3 Paramedic Coordinating High 51 min.
E E1 Medical specialist Leading High 31 min.
E2 Medical specialist Future user Low 40 min.
E3 Medical specialist Coordinating and superuser
18
generic content of the system. Another departmental member is detached for 50% to an EHR building team.Interviewees of department A especially mention an efficiency gain as a potential merit of EHR implementation. “So yes, I see a lot of benefits. Especially time savings of course, once information is stored, you don’t have to go through it again” (A1). Patient safety is only mentioned twice during the interviews and there is no agreement whether the EHR will lead to more patient safety. Maintaining the patient relation is more a goal than changing the relation. Maintaining is the goal because interviewees are worried digitalization might negatively influence contact between patient and healthcare professional. “you have to be very aware that you don’t only look at the screen […] that you involve the patient” (A3).
In department A, system content appears to fit departmental context and requirements. “Yes a couple things are made generic. Everybody works with allergies, medication, patient history and a problem list. Yes, you try to look from the viewpoint of [department A], […] well, we can work fine with that” (A2). Multiple departmental characteristics might have positively influenced system content evaluation. First of all, one integrated system has benefits for a department with sub-units that all have different electronic systems. “Now we have a part digital and a part paper, and there are three different digital systems. So sometimes you have to write or type things four times. Later we can do that all in once. I think that is... that makes me very happy” (A3). Second, the department engages in frequent internal and external data exchange. Structural input of data and order management are in good fit with this requirement. Nurses and physicians constantly need to communicate about patient results. “There are a lot of nice things in [the system]. Like this, a nurse enters data and I can see it directly. I don’t have to ask the nurse all that” (field observation). Furthermore, interviewees expect an increase of efficiency in external data exchange: “Of course it works well if you write down patient information … and its directly in the letter, or a score… And you have to note a lot of parts, because it goes to national data registration. […] All that data has to be delivered so you have to do it very structural” (A2). Third, the department collaborates intensive with other departments. A common database with uniformity of input is expected to increase effectiveness of interdepartmental collaboration.
19
The implementation process is perceived as top-down. However, because the department has delivered one member to the decision team for the generic content of the EHR, interviewees feel the department has a certain level of influence. “There’s a department team, […] led by our Chef de Clinique. And our Chef de Clinique is also part of [decision-making] team, [departmental member] has quite some influence. So through her I have the feeling we’re being heard” (A3).Adapting department specific content of the systems appeared relatively easy at department A, within the limits of the ‘vanilla’ content of the system. A departmental characteristic which might have influenced ease of adaptations is that the department has detached one member to an EHR building team. “If something needs adjustments? Well then we have [colleague X] right? I don’t say everything is possible. I mean, sometimes we too have something which isn’t possible or allowed, because [system] has the last call in that, but then you’re going to look how you can organize it” (A3).
Problems with learning occurred with receiving information about the system. Healthcare professionals at department A have limited knowledge about work processes of other professions within their department. “I think we don’t know from each other, what someone views on his screen. I think it would be good to know. Yes, everybody works with the system differently, but eventually you have to work together” (A2). A day after ‘go-live’ a nurse at the outpatient clinic said that she had no idea how the system looked like for a physician. Limited information of other professions might decrease effective use of the order management affordance. “We have to set up assignments for them, orders, but I have no idea how they see that order, and how they handle those orders. And I think to understand each other… because nursing staff will become frustrated if a physician doesn’t enter the order quickly enough, and the physician thinks, I still have half an hour” (A2). Leveraging the order management affordance into a concrete outcome of more efficient communication might be negatively influenced by a lack of information about other professions.
Department B is a small sized unit with 57,96 fte. It has an outpatient clinic and shares its nursing ward with department A. As one of the few departments of the hospital, there is a surgical room within the department. Most surgical procedures are therefore executed within the clinic of the department. Moreover, medical specialists can execute research and radiation treatments themselves: “our clinic is a bit more special. You’re doing research and treatments yourself. At other departments radiology does that, but here we do that ourselves, because the physician also has a radiation diploma” (B2).
20
safety” (B2). However, similar as department A, changing the relation between healthcare professional and patient is not seen as a goal, and interviewees do not expect changes in this relation.The system content is evaluated positively by the department. The affordance of structural data input can be leveraged into a concrete result of higher patient safety. “What we do now is that we all type in a sort of notepad-like domain. And what you get later is that you have boxes with the basic data of patients, so medical history, medication list, and that’s visible for every user. That makes it more reliable. It can’t happen anymore that physician A prescribes a medicine three times a day and physician B two times a day and that subsequently nobody knows what’s true” (B1). Also the affordance of order management is recognized and perceived to lead to concrete outcomes. For instance the orders. I think that one is a very good one, that it’s more clear what everybody wants. In the current situation it happens that it is unclear, that people think ‘I have said that right?’ but that things are carried out a little different. For instance medical administration with planning. Well I think those ambiguities will happen no more” (B2). The affordance of standardization of roles and data input is less recognized as a functionality of the system that possibly leads to positive outcomes for the department. An interviewee noted that entering data during consult is not a change in tasks for department B. “What I do now is when I talk to a patient, I directly type in such a way that patients can read what I type” (B1). The stricter authorizations coming from standardization of roles and data input in which only medical specialists are authorized for certain patient related tasks, is seen as logical, because it is congruent with law and regulations. According to interviewees, privacy of patients increases because of stricter authorizations. On the other hand, stricter authorizations are expected to cause an imbalance of tasks, according to an interviewee. Our nursing staff does a lot of tasks in the current situation, like a physician. […] soon they aren’t able to do that much anymore. So that’s going to be a tricky time. Is it still workable yes or no or should a repair in certain competencies take place? (B2).
The process of implementation is evaluated less positive by departmental members. Interviewees experienced flawed communication between the department and EHR implementation teams. “If things needed to be discussed or we wanted to know something, email was used. And often you received a mail back that it wasn’t the right person you asked, that you had to go to another team… yes communication definitely could have been better” (B3). Flawed communication led to problems with system adaptations and information receiving. “You talked about it very long and you asked ‘do you get it?’ and they say ‘yes it’s going to be okay’ and eventually it’s not okay” (B3).
21
could not plan surgical operations which take place here […]. So physicians here are never trained to plan a surgical procedure” (B2).There was also a problem with information received about the HER. Information received was incomplete. For nurses it was still unknown what their specific tasks would be after ‘go-live’. “[the nurses] are going to do way less. But then you… you’re kind of a reception lady, but you want to do something else instead of smiling politely behind a desk. And the difficult thing is, we don’t know yet. This morning a colleague said ‘yeah we don’t know we have to find out’. But in a month we already have the system. It’s weird right that we don’t really know what people are going to do?!” (B2).
Striking about department B is that members did not expect a change of their work processes after implementation. According to interviewees, the only real change is that there will be a new system were people should get used to. “Does that has consequences for the department? Yes at first with the habituation period of course it does. But eventually I don’t think we’re going to work very differently. Maybe a little in the way I register, but now I’m also just preparing my consult, then make a note, a letter, and send it, and [after ‘go-live’] I’m going to do exactly that. It looks different, but…” (B1). A possible explanation for the perception that work processes will change very little is that the department is relatively developed in digitalization. “We actually work digitally for three years now. So the step towards [EHR] is not very big for us. We write every report immediately and we don’t have any paper statuses anymore. So in that respect it’s easier for us I think” (B3). Therefore the expectation is that learning tasks to gain a better fit between task and technology is less difficult for department B.
Department C is a medium sized surgical department of 12 medical specialists. They have an own ward and outpatient clinic. In total 70 people work for the department. The treatments and surgeries provided are variable and cooperation with other departments is necessary. According to interviewees, the department has a culture that stimulates change. “So it’s in the culture of the department. Now it’s the new EHR but when quality demands where increased, a safety management system needed to be introduced, the department was first in line too to say we’re going to implement it like this and that” (C1). Participation in the change process is high. The department has both members in decision-making teams and construction.
22
EHR, certainly an integral EHR such as [system] makes certain work processes more uniform, causing more uniform ways of working, causing more cohesion between departments” (C1).Interviewees have different expectations whether or not the system increases efficiency, because two affordances are considered to have opposing effects on efficiency. The affordance structural data input is evaluated by all interviewees as positive. Structural data input potentially causes easy, quick and more careful data exchange. “Now I’m typing the lab values from one screen to the other. If you can transport that in once, yes than you are a winner, than you have it there” (C3). However, administration time increases because of role standardization.
Standardization of roles and data input is not perceived as an affordance that causes departmental benefits. Interviewees do not think that medical specialists are better able to enter patient data. This will not lead to higher patient safety. “You’re the source, you have to put it in. Well, my secretary sits next to me, that’s registering on the source too. I think a secretary is better able to register than a physician. That’s going to be a problem here” (C1). One interviewee mentioned that entering data during consult has negative consequences for the relation with the patient. “Currently, I’ll try to have a conversation with the patient, and on the moment he leaves the room I do my administration. But the time I have per patient decreases. And I have to click stuff during the process, because it has to be done with the patient. That’s weird right?” (C2). The expectation is that administration time will eventually stay equal because negative effects of standardized roles and data input are compensated with positive effects of structural data input. “Certain things we will get on our plate. That’s going to be the new task division, [after ‘go-live’] we have to enter extra codes, indicators and letters. So in that way administration increases. But that works nicely with [new system]. If your notes are correct, you can make the letter in one go” (C3).
The implementation method is perceived as positive by interviewees. They do not think that the implementation method is too rigid or too top down. “I think it’s good. It is a pro-active manner, it looks like a professional method, they clearly communicate that there is no way back, don’t counteract, because that’s the case in a big hospital, digging your heels […] so I think it’s really well done and one of the qualities of [the hospital]” (C3). A departmental characteristic which might influenced the perception of the implementation method is that communication lines between EHR teams and department were clear. “I’m [decision-making] team member and leader of our department team, and together with one departmental member, an assistant who has lots of experience with [the system], we are the contact persons. He knows a lot of the content and I know a lot of hospital policies” (C1).
23
Department D is a medium-sized unit with 101,26 fte. The department has a number of specific characteristics. First of all, the department is working in a stand-alone matter, having limited data exchange with other departments. Second, patient turnover is very high. Healthcare professionals treat many patients a day for often minor interventions. Third, there is a labour shortage of medical specialists in the region for this medical specialty, resulting in patient waiting lists and difficulties with delivering departmental members for EHR teams. “[the department] has as peculiarity that, unlike other medical specialties, there is labour shortage, and in [this area] even a big shortage of [medical specialists]. So if I join the [decision-making] team, we lack a specialist that we cannot replace” (D1). Fourth, healthcare professionals operate with many different technological tools. Integrating all these technological tools is an almost impossible endeavor. “A difficult job, yes. Because […] getting all the imaging in one viewer, that’s a task several companies have failed to do. They all said yes we can, but they all crashed on that database” (D2).Interviewees think that the EHR system will cause acute problems for the department. “There are two options. Or we’re completely bankrupt […], with all patients outside. Or we have a totally different system as [department D] were we can work with” (D3). Efficiency will decrease, because medical specialists expect to have more administrative tasks and patient treatment numbers will decrease. Waiting lists that are already present are expected to grow. According to interviewees this will cause medical selection of patients and therefore patient safety may decrease. Furthermore, interviewees expect a negative influence on the relation between healthcare professional and patient because the healthcare professional will become unable to have a real conversation with the patient and loses his or her time merely on administration.
24
that himself. And before you are at the stage that you have delegated that again… therefore it’s too rigid, because they are focusing on having the base in order, but for us the base in order is twenty years back” (D3). Thirdly, there is a misfit between the order management affordance and parallel work processes of department D. Parallel work processes are necessary because of high patient turnover and increase efficiency of work processes. The new order management is expected to force a sequential work process. Previously laboratory, nursing staff and physicians did several tasks synonymously for a patient. “Many examinations are carried out during consult. There are many clinics where they say come back another day for your results. Here it is like, well I need this examination, well okay go to the [technical examination assistant] which does the examination and directly you see the results on your screen” (D2). After ‘go-live’ every task needs to be sequentially executed or ordered by the medical specialist. “You give him a note and the patient goes to the desk and they take care of it […] but later nothing happens before you do the order. So you have to do the order before the patient can go to the next station” (D3).Not only change content, also change process is evaluated negatively. “The whole implementation method was and is a very inconvenient happening” (D1). Necessary adaptations of the system are difficult to make and there appears to be little influence of department D in the implementation process. “Like the surgery requests. After the surgery department we deliver most surgeries. We addressed the issue last year December, ‘hey guys can we please have some insight and think along in the process?’ Didn’t happen, and now since September it has escalated and now we can finally submit surgery request. Far too late of course” (D3). Furthermore, the department does not know where to go for necessary adaptations. “Oh the whole time you have to scale up problems. You have meetings with a person, and you cannot come further because he’s not allowed to do certain things […] you have many meetings at which you know beforehand that you will talk about issues which you will not solve” (D1). Adapting the system to make it more in line with departmental work processes and requirements is difficult: “It takes an enormous amount of energy to adapt” (D1).
Another problem is that the department has received little information about the system. “But that you have in any way an overview of what’s possible in the system, I can absolutely state that is not the case. We also had no single way to test whole patient trajectories. […] you see fragments” (D1). At department D, limited system overview caused insecurities about the effects of the system on departmental work processes and limited knowledge about adaptation possibilities.
Department E is a small sized unit with 7 medical specialists. The unit has an outpatient clinic but does not have a ward of its own. The unit is not an official department but is part of a large department that consists of various disciplines2. Medical specialists at department E are interested in divers and detailed information about the patient. A characteristic of the unit is that they treat certain patient
2 The researched unit in this study is a sub-unit of a large department with multiple sub-units. For matters of
25
groups throughout the hospital. “Which means that during service, we’re responsible for all [patient group X] patients in the hospital” (E2). The department is therefore cooperating with and depending on other departments extensively for data exchange.Interviewees of department E expect positive effects of the system on patient safety. Using the system causes a decrease in errors, for instance in medicine prescription. The expectation is that efficiency will not gain. “But it is time-consuming right, you have to click a lot and very precisely. For the quality of care that’s fine, it’s great, but it is time-consuming, you have to fill in a lot” (E3). Departmental members do not think the system will have an effect on the relation between patient and healthcare professional, especially not within their department. According to interviewees, clientele at department E does not want a different relation with the healthcare professional. According to all interviewees, implementing an EHR is necessary because the technological systems are too old-fashioned and it is time to innovate. “So I don’t even think it’s a necessary evil. I think its maintenance which is long overdue which we catch up now, and I’m happy with [the EHR]” (E1).
There are multiple instances of good fit between the standardized content of the system and requirements of department E. First of all, one integrated system with structural input of data is in good fit with strong cooperation with other departments. This department treats patients staying at various departments throughout the hospital. Clear communication of data is key. “Sometimes you’re called by for instance ward K1 and they say; ‘well I have patient X here and patient X has a [certain score]’. And then I think; A: who is patient X, B; what does patient X eat, C; what does Patient X use. All that kind of stuff and then again you have to… and now in [the new system] you can just see it. And that’s a real improvement” (E2). Second, an integrated system with structural data input is in good fit with the amount of patient information medical specialists at department E need. “We have a sort of generic function in the sense that we look very broadly and therefore we want all information. An internist in general but we in particular are kind of neurotic and focused on detail and therefore we want information to be correct” (E2). Third, the order management affordance fits the high degree of checking protocols by medical specialists at department E. With order management it is possible to check whether or not an order is given from a medical specialist to a nurse. “Yes it’s good.. for quality it’s good, it’s very precise, you go through everything precisely. The physician gives an order and the nurse carries it out and then you can check what the nurse did, and when, and if the task is done. And you have proof that the physician told it” (E3).
26
The department has a participation method to gain involvement of departmental members and educate departmental members about the change. Selecting certain healthcare professionals to become superuser is part of the departmental strategy to make the change towards the new system easier. Whereas at other departments often people who want to become superuser are appointed, at this department there was a certain strategy for making particular departmental members superuser. “Those are people who find it difficult to change. They are now superuser or even order set builder. But I thought, every extra exposure they have… look, they are not stupid or clumsy, they are often extremely well organized people who find it difficult to let their own system go. If you invite them to think along about the new system, you win at both sides, because they already can think well in a system” (E1). Furthermore, there is experience with implementing this particular system in another hospital, there is a clear change leader, and there are structural meetings about the system with the whole department. As part of a larger department, the unit has resources to put time and effort in the EHR implementation within the department.This participation strategies influences how well departmental members receive information about the system. “Well what I said about the weakly gatherings, and I had a walk-in consultation hour […] I still have that but now you see the appetite for information is appeased and everyone knows what the deal is” (E1). However, it does not necessarily lead to more adaptations of the system or better learning to work with the system.
4.2. Cross-case analysis
In this chapter, differences and similarities between departments in expectations of effective system use are presented. First, differences and similarities of system affordances perceptions are analyzed. Next, evaluations about the implementation process are compared. Table 2 summarizes the results of the analysis of the content and process of change.
4.2.1. Evaluation of affordances
27
Table 2: Summary of analysis
Evaluation of affordances Factors influencing expectations of effective use
Dep. Structural input of data Order management Standardized roles/data input Process factors Departmental requirements Departmental context A Leads to more efficient
internal and external data exchange
In line with internal data exchange
Possibly creates problems with work processes
Representative (+) Adapting easier by having departmental member at building team (+)
Extensive data exchange between professions and with external parties (+)
Four sub-units with different systems. One integrated system reduces administration time (+) B Increases patient safety
by less ambiguous recording Increases patient safety by improving clarity of communication within the department
According to law and
regulations. No big change for this department but chance of inefficiencies
No representative and unclear communication lines make necessary adaptations difficult (-)
Specific requirements: surgery function. Build in too late so that learning specific functions was not possible (-)
Little (perceived) change of tasks, roles and protocols (+)
C Leads to more efficiency by reducing administration time. Leads to more patient safety by reducing errors
No results found Leads to more administration time and less patient contact
Representative increases learning about the system (+) Clear communication lines (+)
Cooperation with other departments (+) no necessity of availability of all patient information (-)
Relatively long time per patient, decreases with standardization of roles. Negative effects on relation with patient (-)
D Common database relatively very time-consuming Imposes sequential working processes which leads to inefficient work process.
Shifting tasks to the medical specialist leads to acute efficiency problems
No informed adaptations possible due to limited information about the system (-)
Easier adaptations after department received
permanent contact person (+)
Need to link all
technological devices with EHR
Need to divide tasks of medical specialists (-)
High patient turnover (-) Labour shortage (-)
E Increases efficiency of data exchange with other departments
Increases efficiency of checking protocols and data
Sometimes impossible due to limited number of medical specialists who have to be “on two places in the same time”
Participation strategies to cope with uncertainty and increase information about the system (+)
Need of elaborate patient information (+)
Need of checking protocols (+)
Responsibility of certain patient groups throughout the hospital, good fit with affordances of EHR (+) Small unit (-)
28
A similarity between departments is the expectation that structural data input potentially increases patient safety. Medicine and allergy information is entered more careful, causing a reduction of medical errors. However, there are indications that some departments are better able to leverage the affordance into a concrete result than other departments, because some departments are more engaged in medication prescribing. “That you can check more precisely.. especially the medication. I’m (department E profession) so for me medication safety is very important, and with [the system] you can check that very well I notice” (E3). “Medicine interactions, that’s not my area. So I don’t want that, I don’t want to be involved in that. I don’t want to make mistakes with [medicines]. So I would prefer, when patient data is known, for instance medicines at a pharmacy, that it is channeled to the [hospital]. And I don’t believe that’s too hard, especially since we automate”. (C2).The affordance is evaluated differently in terms of efficiency gains. Some departments have a high degree of data exchange between professions (A) or with other departments (E). These departments can leverage the affordance of structural data input into a concrete outcome of more efficient data exchange. Department D is less dependent on data exchange with other departments and operates in a more stand-alone matter. At department D, structural data input is not expected to cause more efficient data exchange. “For that matter we’re totally separate. So that’s one of the things that I face. A vital important issue for a surgeon, internist et cetera, is normally of total unimportance for us. So that’s.. I think other specialisms can work more generic, but that’s not the case for us” (D1).
Another difference is the effect of working with a common database. For most departments (A, B, C & E) a common database has both positive and negative effects but does not lead to problems. At department D working with a common database does create a problem, presumably because of high patient turnover and limited time per patient. Working with a common database relatively costs more time at this department, causing a reduction of patients after ‘go-live’ at department D.
The second evaluated affordance is order management, the electronic communication between healthcare professional and medical supporting processes. Potentially order management increases order clarity and efficiency in tracking errors. The affordance order management is evaluated differently by departments, depending on departmental requirements.
29
The last analyzed affordance is standardization of roles and data input. In this case, standardization of roles and data input means that access to EHR functionalities is restricted, based on profession, and patient information is entered during consult. Certain patient information cannot be entered by administrative staff anymore but has to be entered by medical specialists. Potentially this affordance increases patient safety. Privacy increases because unauthorized healthcare professionals cannot see certain patient information. Furthermore, the assumption is that medical specialists are more capable of entering trustworthy patient information. The relation with the patient changes because data is entered together with the patient. During a demo implementers showed how this affordance can be leveraged: by directing the screen towards the patient, the patient can have direct insight in the care process. According to a trainer, patients get direct insight in their care file, which improves effectiveness of transmural communication. However, these assumptions about standardizing roles and data input are contradicted by departmental evaluations. Departments do not perceive the affordance positively.A similarity between all departments is that the EHR is not expected to change the relation with the patient. “I don’t see an influence of the EHR in that. You have your own type of physician – patient relation right? And it will never be tantamount because you have different knowledge than a patient” (C3). Furthermore, departments do not expect entering data with the patient leads to concrete benefits. An often heard statement of interviewees of varying departments is that entering data with the patient can lead to a worse relation because the computer can become a wall between healthcare professional and patient. Another similarity is that departments do not expect standardization of roles to increase patient safety. Medical specialists are not expected to be more capable of entering patient information than administrative or nursing staff. Furthermore, standardized roles can cause inefficiencies in work processes of departments and increase administration time of medical specialists.
