Information and Organization 27 (2017) 100–115 Contents lists available at ScienceDirect Information and Organization journal homepage: www.elsevier.com/locate/infoandorg Affordance potency: Explaining the actualization of technology affordances MARK Chad Andersona,⁎, Daniel Robeyb a b Department of Business Informatics, College of Informatics, Northern Kentucky University, Highland Heights, KY 41099, United States Department of Computer Information Systems, Georgia State University, Atlanta, GA 30301, United States AR TI CLE I NF O AB S T R A CT Keywords: Affordance potency IT artifact IS theory Healthcare Given the importance of information technology (IT) in effecting organizational change, scholars have strived for many years to theorize the ways in which IT can produce the changes intended for it. Recent arguments claim that most information systems (IS) research has taken a limited theoretical focus on the information technology (IT) artifact, which arguably should be at the core of the IS discipline (Benbasat & Zmud, 2003). This research engages directly with the IT artifact by evaluating the use of an electronic medical records system and its relation to actualization of technology affordances. We conducted a case study at a large urban acute care hospital in the Midwestern United States with registered nurses working on inpatient care units as the clinicians of interest. Through interviews with nurses and other clinical stakeholders, observation of nurse's work practices on three patient care units in the hospital, and direct examination of the medical records system, we develop theoretical insights into the role of IT in work practices. The novel concept of affordance potency is introduced as an integral theoretical construct in our model of affordances, helping to explain actualizations of IT in use. Our contribution provides a nuanced yet powerful way of understanding the nature of IT artifacts and their relationships to technology users and work practice. 1. Introduction Information technology (IT) is typically designed to enable change in individuals, organizations, and societies. However, the relationship between IT and the change it is intended to facilitate is inconsistent, and much of information systems (IS) research is focused on understanding this issue through a variety of theoretical lenses (Robey & Boudreau, 1999). Recently, a number of IS scholars (Leonardi, 2011, 2013; Majchrzak, Markus, & Wareham, 2016; Markus & Silver, 2008; Robey, Anderson, & Raymond, 2013; Strong et al., 2014; Volkoff & Strong, 2013) have considered affordance theory, drawn originally from the field of ecological psychology, as a viable lens for theorizing IT-enabled change. Affordance is a foundational theory in ecological psychology and has been utilized in other fields like human-computer interaction (HCI) (Norman, 1988; Sutcliffe, Gonzalez, Binder, & Nevarez, 2011), engineering (Brown & Blessing, 2005; Maier & Fadel, 2009), and education (Dalgarno & Lee, 2010; Gamage, Tretiakov, & Crump, 2011). The recent interest in affordances within the IS field stems from a call to understand the material nature of IT and its relationships with its social context of use (Orlikowski & Iacono, 2001). Our study lies within this stream of research by offering new theoretical insights into IS affordances and their implications for IT-enabled work practices. The term affordance was coined by psychologist James Gibson, who postulated that, “The affordances of the environment are ⁎ Corresponding author. E-mail addresses: andersonc16@nku.edu (C. Anderson), dan.robey@gmail.com (D. Robey). http://dx.doi.org/10.1016/j.infoandorg.2017.03.002 Received 12 July 2013; Received in revised form 2 March 2017; Accepted 31 March 2017 Available online 06 April 2017 1471-7727/ © 2017 Elsevier Ltd. All rights reserved. Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey what it offers the animal, what it provides or furnishes, either for good or ill” (1979, p. 127). IS scholars adopting the concept have adapted it by focusing on technical artifacts as the part of the environment directly relevant to IT users. For example, Markus and Silver define a functional affordance as, “the possibilities for goal-oriented action afforded to specified user groups by technical objects” (2008, p. 622). More recently, Strong et al. define an affordance 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” (2014, p. 12). Thus, affordance theory offers a promising perspective on IT-enabled change by focusing attention on aspects of the relationship between an IT artifact and its users. The context of this research study is healthcare, and the IT artifact studied is an electronic medical records (EMR) system used by clinicians to provide and document patient care within an acute care hospital. Our aim is to use this context to further develop the concept of IS affordance and its relationship to work practice. We build on existing affordance research by introducing the concept of affordance potency as a novel theoretical construct within the nomological network of affordance theory. Affordance potency helps to explain the influence of IT's material properties on work practices. We begin by reviewing affordance theory and developing the concept of affordance potency. Following this theoretical development, we describe the research methods, present the findings, and refine our contribution in a discussion of the findings. We conclude with implications for future research and practice. 2. Affordance theory A number of scholars have issued calls for the study of technology affordances as an inroad to understanding IT-enabled social change. In sociology, Hutchby (2001a, 2001b) posits affordances as a theoretical mechanism for understanding the constraining influences of technology on human action. In organization science, Zammuto et al. suggest that affordances can provide an effective lens for explaining the “increasingly symbiotic relationship between IT and organization” (2007, p. 752). Leonardi and Barley (2008) hypothesize that material features of IT act as both affordances and constraints on individual user behavior, and argue for developing a language of affordances that allows us to better explain and predict technology-occasioned change. Affordances have also assumed prominence in IS research directed at understanding and solving societal-level problems (Majchrzak et al., 2016). One approach to responding to such calls is to use affordances to complement existing theories used in IS research that limit their treatment of the IT artifact. As Majchrzak et al. advise: “By looking at technologies as sets of affordances and constraints for particular actors, IS researchers can explain how and why the “same” technology is used or has different outcomes in different contexts, thus enriching general and substantive IS theories” (2016, p. 272). For example, Giddens' (1984) structuration theory provides little guidance for understanding the theoretical role of IT artifacts. To compensate, Jones and Karsten (2008) suggest that affordances could provide a way for IS researchers using structuration theory to construct a more consistent theoretical account of the IT artifact. In similar fashion, Markus and Silver (2008) note that the concepts of “structural features” and “spirit” in Adaptive Structuration Theory (AST) (DeSanctis & Poole, 1994) have seen little use by other researchers. Markus and Silver therefore suggest that, in addition to “technical objects” and “symbolic expressions,” “functional affordances” could serve as an effective substitute for “structural features” and “spirit” in research using AST to understand IT-enabled change. As seen in these examples, a theory of affordances may not stand apart from existing theories of IT-enabled change, but rather serve to enrich and deepen the explanatory power of theory by more precisely theorizing IT artifacts in relation to their social contexts of use. In the spirit of enriching theories of IT-related change, we propose a new concept of affordance potency within the nomological network of affordance theory and explain its added value for understanding IT-related change. We adopt the definition of affordance from Strong et al., mentioned earlier: “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” (2014, p. 12). Thus, an affordance is dependent upon the relationship between other concepts, in this case an artifact and an actor, and therefore does not exist independently of those concepts. However, relational views of affordance have not been universally held by scholars, as some have argued that an affordance is a property of the environment.1 Initially, ecological psychologists understood affordances as dispositional properties of the environment (Heft, 1989; Michaels, 2000; Reed, 1996; Turvey, 1992). More recently, Stoffregen (2003) and Turvey (2009) argued that affordances should be conceptualized as emergent properties of the inseparable animalenvironment system. Fayard and Weeks (2014) offer yet another view of affordances as both dispositional and relational. We are partial to Strong et al.'s definition, which resonates with Chemero's (2003) assertion that affordances are “relations between the abilities of organisms and features of the environment” (2003, p. 189). This relational definition of affordances is also consistent with other IS researchers (Leonardi, 2013). Affordance theory in IS research also tends to distinguish between information systems and social contexts within which information systems operate. A relational view of IS affordances maintains the relevance of the social context while emphasizing the distinct role of IT artifacts in generating affordances. While the social context also consists of material artifacts, such as physical human bodies and workspaces, it usually refers to abstractions like work routines, authority relationships, job responsibilities, and incentive systems. A convenient way to approach social context is to focus on “work practices,” the recurring activities through which individuals and groups direct their work-related efforts. The phrase is meant to refer in a general way to recurrent situated activities rather than to stand on its own as a theoretical concept. Work practices may or may not be treated as “routines” (Feldman & Pentland, 2003) and our intent is not to contribute directly to a theory of routines. Rather, we use the term to denote the application of a 1 Divergent schools of thought emerged in part because Gibson's (1979) initial definition was ambiguous on whether an affordance is a property of the environment or a property of the relationship between an animal and its environment. 101 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey practice lens (Orlikowski, 2000) involving IT applications implicated in actual work, as in many earlier studies (e.g., Chu & Robey, 2008; Leonardi, 2012; Schultze & Boland, 2000; Vaast & Walsham, 2005). By considering the relationship between IT artifacts and work practices, researchers may explain the consequences of IT use on social context. Using the relational framing of affordance, scholars can avoid deterministic approaches to studying IT “impacts” and move to more nuanced studies of the interplay between technology and work practices (Kling, 2000; Robey & Boudreau, 1999). 2.1. Affordance actualization While most scholars follow the lead of Gibson (1979) in theorizing affordances as behavioral possibilities or opportunities, few explicitly address how particular affordances are selected from the many potential affordances available at one time. Since there are typically multiple behavioral opportunities available to an actor in the environment, it is worth developing explanations for the choice to act or not act on a particular affordance that leads to the actualization of the affordance. One reason not to actualize an affordance is a failure to perceive it. As part of his original idea that affordances existed independently of their actualization, Gibson states that, “The observer may or may not perceive or attend to the affordance, according to his needs, but the affordance, being invariant, is always there to be perceived” (1979, p. 139). By this reasoning, the set of affordances available can exceed the affordances perceived to be available at any given moment. This logic became part of Norman's (1988) basis for introducing affordance theory to the HCI field. Norman suggests that technology designers should make technology affordances explicit to potential users so that intended affordances would be easier to perceive. A second reason not to actualize an affordance is their disconnection from user goals. In Gibson's original conceptualization of affordance he argues that, in contrast to the Gestalt notion of demand character or valence, which bestows meaning on physical objects based on the needs of an observer, “The affordance of something does not change as the need of the observer changes” (1979, p. 138). Stoffregen makes the same argument when he states, “Affordances are defined independent of the goals of the animal. The opportunities for action are what they are regardless of what we may want or desire” (2004, p. 82). Therefore, the set of available affordances may include many that do not correspond to a need or goal and consequently, those affordances are less likely to be actualized. The link between affordances and goals has also been studied in IS research. Leonardi (2011) found that when users were unable to achieve their goals with their current set of routines and technologies, they changed either one or the other. For Leonardi, the decision whether to change the routine or the technology is based on the user's perception of constraints and affordances. When a user perceives the technology as a constraint, the user will look to change the functionality of the technology. When a user perceives that the technology offers an affordance, the user will look to change the routines to take advantage of that affordance. Therefore, when people evaluate technology they see either affordances or constraints based on how the technology fits with their current goals. Strong et al. (2014) offer the most recent and most complete explanation of affordance actualization, which they define as “the actions taken by actors as they take advantage of one or more affordances through their use of technology to achieve immediate concrete outcomes in support of organizational goals” (2014, p. 15). In their study of the implementation of an electronic health records system in a multi-site medical group, Strong et al. (2014) looked at goals and affordances at the individual and organizational levels. As clinicians drew upon the affordances of the system to achieve their individual goals of patient care, they contributed to the achievement of organizational goals to provide more effective health care at lower cost. Affordance actualization is therefore conditioned by both the perception of the affordance and the relationship of perceived affordances to current goals. However, affordance actualization observed in practice is not fully explainable through perception and goals alone because there are times when a user is both aware of an affordance and has a current goal that the affordance could support, and yet the affordance is not actualized. Strong et al. (2014) acknowledge this possibility through their conception of affordance actualization as a process entailing “journeys” stretched over time as users interact with shifting technology features. This process of actualization resembles Leonardi's (2011) concept of “imbrication” in which distinct elements, both technical and social, are intertwined in practice over time. Strong et al. identify three factors (abilities and preferences of the individual, features of the system, and characteristics of the work environment) that both support and restrict an individual's affordance actualization. They suggest that future studies using their affordance lens, “would contribute to understanding what it is about IT and its design, about organizations and actors, and about their relations that matters in IT-associated organizational change” (Strong et al., 2014, p. 80). Our own study contributes to this need by suggesting that affordances possess a “potency” that is a product of the three factors mentioned above. 2.2. Affordance potency Our development of the concept of affordance potency is rooted in earlier studies of the complexity of affordances in relation to users' choices of behaviors. For example, Warren (1984) conducted experiments on stair climbing and found that the affordance of stair climbing is based on a ratio of riser height to leg length. Increasing the ratio makes the affordance more difficult until “a critical point is reached at which the stair is too high to afford bipedal climbing, and the climber must shift to a quadrupedal (hands-andknees) gait” (1984, p. 686). Conversely, the affordance of stair climbing is least difficult at an optimal point, “yielding minimum energy expenditure per vertical meter of travel - a candidate for the best fit between climber and stair” (1984, p. 686). Consequently, while all affordances are behavioral opportunities, affordances that enable the same type of behavior may not be equivalent in the amount of effort it takes to carry out the behavior. McGrenere and Ho (2000) call the range over which an affordance is, on the one end, maximally easy to undertake, and on the 102 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey other end, no longer possible to undertake, the “degree of affordance.” They further suggest that when designing technology, affordances can be designed with a “degree of perceptual information,” which is the clarity of information available to perceive the affordance. Together the degree of affordance and the degree of perceptual information produces a two-dimensional continuum which they argue determines the usability of a technology. We extend the idea that affordances differ in their power or utility to enable user goal achievement. Just as pharmaceutical drugs may vary in their potency for treating medical conditions, we posit that technology affordances vary in their potency to help users perform their work practices. Specifically, we define affordance potency as, the strength of the relationship between the abilities of the individual and the features of the system at the time of actualization, conditioned by the characteristics of the work environment. This means that, while technology designers may attempt to create affordances based on anticipated user abilities and design system features within an expected context, the potency of affordances is always based on the abilities of the specific individual and the existing system features within the actual context of use. Therefore, affordance potency, along with perception of the affordance and the goals of the individual, is a necessary condition for affordance actualization. We define the strength of the relationship as the ease with which the individual can actualize the affordance, which is a function of the requisite energy, mental and/or physical, needed to carry out the actualization. An affordance with strong potency would therefore require less energy to actualize than a similar affordance with weak potency. Higher energy requirements will translate into greater difficulty and therefore, all else being equal, an affordance with weak potency will be less likely to be actualized than an affordance with strong potency. Affordance potency will be unique to each individual since it is partially based on the abilities of the individual and it can change over time as abilities, features, and the environment change, but it will always exist at a specific level at the time the individual chooses to actualize the affordance. Strong et al. (2014) suggest that future research on affordances should include the study of difficulties individuals encounter during affordance actualization. Affordance potency responds to that suggestion by explaining why individuals encounter difficulties during actualization and how that influences their decisions to actualize an affordance or to find ways to work around the difficulties. That this notion has not been well-developed in IS research is surprising, given that many IS researchers focus on improving system features. The tendency, however, is to design features into systems with a particular set of users and contexts of use in mind. Alternative uses are not entirely unexpected, but are generally considered outliers to be dealt with infrequently and therefore the difficulties experienced by “non-typical” users in actualizing an affordance are not sufficiently considered in the design of systems. To develop the idea of affordance potency, we undertook an empirical study of EMR use in clinical work practices. The following section details the methods for the empirical study. 3. Method 3.1. Research design For Gibson (1979) and other ecological psychologists, affordances represent an important ontological premise for the study of perception; that meaning can exist in the environment independent of meaning that is internally or socially constructed. In IS research, critical realism (Archer, Bhaskar, Collier, Lawson, & Norrie, 1998; Bhaskar, 1978; Mutch, 2013) has been used to support this ontological position through a philosophical perspective that takes “…a realist view of being in the ontological domain whilst accepting the relativism of knowledge as socially and historically conditioned in the epistemological domain” (Mingers, 2004, p. 91). Volkoff and Strong (2013)argue that affordances are the generative mechanisms of critical realism, and we find a realist view to be an effective philosophical position from which to study and understand technology use that can be consistent across settings at times while also being amenable to varying user appropriation. Accordingly, the research design used in this study provides a balanced focus on objective system features, user goals and abilities, and the social context of IT use. We employed a qualitative case study design (Eisenhardt & Graebner, 2007; Yin, 2003). The study's focus on healthcare was shaped by the first author's background as both a clinician with experience providing care in hospital patient units, and as a vendor of health IT systems designed for hospital patient-care units. He also trained nurses and other clinicians to use those systems. The broad goal for this research was to study the role of IT in the coordination and delivery of health services, with a specific focus on direct patient care in hospital care units. Among the many IT applications in use on hospital patient care units, EMR systems are the most comprehensive and information-intensive systems in place today and are of particular interest given federal incentive programs to promote their adoption and use. We sought a study site in the geographic area where the first author had worked as a clinician and IT vendor, expecting that familiarity would improve our chances of gaining access to an acceptable site, which it did. We were directed, by personal contacts, to a hospital that had recently implemented an EMR system facility-wide in a short period of time, which required them to start with a non-customized package and adapt it later to their specific needs. After speaking with the hospital's director of IT about their implementation process, it appeared that the hospital would provide a potentially rich setting for studying the consequences of IT for the coordination and delivery of health services. For the remainder of the paper, the pseudonym “Urban Hospital” is used to refer to the study site. Urban Hospital is a 555-bed, not-for-profit acute care facility located in the heart of a major metropolitan city in the Midwestern United States. It is a general medical and surgical hospital with a staff of over 1000 physicians and more than 4000 employees. Since 2001, the hospital has been consistently recognized as one of the top 50 hospitals in the U.S. by U.S. News and World Report. The EMR system implemented at Urban Hospital on the patient care units was the EpicCare Inpatient Clinical System, and will be referred to as “Epic.” 103 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey 3.2. Data collection A combination of interviews, observation, and document analysis were employed for data collection in order to strengthen triangulation of evidence and broaden data gathering opportunities (Eisenhardt, 1989). Semi-structured interviews were conducted to allow subjects latitude in expressing their thoughts and opinions while still being guided to the topics of interest. Interview guides were initially constructed based on concepts from the literature and then modified to incorporate new concepts that were identified as the data collection progressed. The initial interview guides are included in the Appendix. All interviews were audio recorded so that the interviewer could focus on the respondent's answers and adjust questions as necessary without having to take detailed notes. Two nurses asked not to be recorded, so for those interviews notes were taken during the interviews and written out in detail immediately afterwards. Epic was implemented throughout the hospital, but Urban, like most hospitals, contained a mix of different patient care units, each with unique characteristics and a certain level of autonomy in how they chose to use technology. Through negotiation with the hospital's nursing administration, access was granted to two medical/surgical units (3West and 5South) and one intensive care unit, the cardiovascular ICU (CVICU). This yielded a diverse set of nursing practices in different physical settings. Access was also granted to conduct interviews and observations on both day and night shifts, as work practices on patient care units vary between days and nights. Onsite data collection was conducted during the summer of 2010. The first author spent a total of 14 days and four nights over a one-month period at Urban Hospital engaged in data collection and interviewed a total of 51 subjects, 33 of whom were registered nurses (RNs) who were the clinician users of interest in the study. Data collection began with interviews of the hospital's nursing administration and system support personnel with the expectation that understanding the system implementation process, features, and intended purpose would provide a basis for gathering data from the clinician users. Hospital administrators included the Chief Nursing Officer (CNO) and the nurse managers of each unit for which access was granted. Interviews with the nurse managers served the dual purpose of identifying their expectations for and perceptions of the use of the system by the nurses on their unit as well as establishing procedures for interviews with the unit nurses and observation of activity on the unit. System support personnel included the five-member Information Technology team, which provided backend support to Epic and its integration with other related systems; and the five-member Clinical Informatics team, which provided direct support to the clinical users of Epic. The interviews with Information Technology and Clinical Informatics personnel were conducted in groups while all other participants were interviewed individually. Three educators who provided ongoing system education and training to clinicians were also interviewed for their perspectives on user abilities related to Epic's features. Nurses were selected as the clinician users of interest for the following reasons. Patients admitted to a hospital are assigned a room on a patient care unit and nurses staff those units to provide primary care to the admitted patients for the length of their stay. Nurses, therefore, spend more time providing and documenting care to admitted patients than other clinicians, they must ensure coordination of care across shifts, and they communicate regularly with physicians and other clinicians who interact with the patients. There are several types of nurses on patient care units including Registered Nurses (RNs), Licensed Practical Nurses (LPNs), Certified Nursing Assistants (CNAs), and Patient Care Assistants (PCAs). Registered nurses have primary responsibility for implementing the physician's plan of care for the patient while the other types of nurses carry out parts of that plan under the supervision of the RNs. Therefore, RNs have the broadest range of responsibilities on the unit as they can perform all of the activities of the other nurses in addition to their unique roles. Based on these factors, registered nurses formed the core of the data collection effort with a total of 33 RNs interviewed across the three patient care units included in the study. For the remainder of the paper they will be referred to simply as nurses. Table 1 provides a summary of all study participants. In addition to interviews, 22 of the nurses were observed as they engaged in their daily work practices of providing patient care and interacting with physicians and other clinicians on the patient care units. Observation sessions always took place after the nurses had been interviewed and had consented to their participation in the study. The first author was given the freedom to move about the units to observe nursing activities and could ask the nurses questions when clarification was needed on observed behaviors. The first author was also given access to both the Epic production system and the Epic “sandbox” environment, which allowed for self-testing of system functions to evaluate features that were discussed during the interviews. In studying the relationships between IT and users, there can be considerable value in engaging directly with the IT artifacts rather than relying solely upon reports from users and IT staff. Observations of physical artifacts, such as portable workstations on wheeled carts, in the hospital units provided additional insight into IT features and their importance to work practices. This combination of data collection methods provided a rich set of data and a thorough understanding of the case environment. Table 1 Study participants. Role Unit # of participants Registered nurse Registered nurse Registered nurse Nursing administrator Nursing educator IT and clinical informatics 3 West 5 South CVICU Multiple NA NA 13 14 6 4 3 11 104 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey 3.3. Data analysis Recorded interviews were transcribed and loaded into nVivo software along with field notes for data analysis. The guidelines for qualitative data analysis provided by Miles and Huberman (1994) were followed in the analysis of the research data. An initial round of coding to identify descriptive and interpretive codes resulted in a total of 87 unique codes (e.g. medication administration record, medication cart, notification flags, barcode scanner). nVivo's tree node structure was utilized to facilitate a code hierarchy with descriptive codes organized under specific interpretive codes with the coding structure evolving over time through iterative periods of coding separated by periods of reflection on both the data analysis process and the results of that analysis. The first round of coding continued until all data had been analyzed and there was reasonable certainty that all relevant text had been coded and no new descriptive or interpretive codes were forthcoming in the data. At that point a second round of coding was initiated. This second round of coding was meant to produce what Miles and Huberman (1994) call pattern codes or meta-codes, which are more inferential and explanatory in nature than either descriptive or interpretive codes. This second round of coding was conducted to elaborate the influences of the Epic system on nurses work practices at Urban Hospital. The second round focused on the codes that contained the most references in the data (e.g., charting, medication pass, handoff). By concentrating on the categories that were given the most attention, we expected to obtain richer, more extensive data for the identification of robust pattern codes. Identified pattern codes were then compared to the literature on technology affordances. Through this process we iteratively developed the components of a model of affordances relevant to the social context of our case including the novel concept of affordance potency. The empirical basis for the model is presented next, followed by a discussion section in which the implications of affordance potency for IT-related changes in work practice are further explored. 4. Results 4.1. Social context of work practice: inpatient nursing Urban Hospital provides both inpatient and outpatient care services. Inpatient care is defined as the admission of a patient for treatment that will require at least one overnight stay in the hospital, while patient services that do not require an overnight stay are considered outpatient care. Nurses' work practices differ significantly between inpatient and outpatient care and the decision for this study was to focus on inpatient care services. Like other hospitals, Urban Hospital organizes its inpatient care around patient care units, and patients admitted to the hospital are assigned to a room on a particular unit. Urban Hospital has a mix of patient care units with each unit configured for the care of a particular type of patient or condition or set of conditions (e.g., maternity, pediatrics, medical/surgical, cardiac care, etc.). These units are physically separate from each other and have dedicated staffs of nurses and support personnel who provide 24-h care for the patients admitted to the unit. Patients admitted to a patient care unit spend most of their hospital stay on the unit and therefore the nurses assigned to the unit become primarily responsible for their care under the authority and direction of physicians. These patients require round-the-clock monitoring and care and therefore the units are staffed 24 h a day on rotating shifts. On the units studied at Urban Hospital, nurses work 12-h shifts from 7 a.m. to 7 p.m. (day shift) or from 7 p.m. to 7 a.m. (night shift). Each unit is headed by a nurse manager and one nurse per shift is normally designated as the charge nurse who functions as the shift supervisor. The remainder of the results presents an account of selected work practices of nurses across three patient care units (3West, 5South, and CVICU) at Urban Hospital with each work practice selected to focus on a particular affordance that uniquely illustrates the concept of affordance potency. Nurses engage in various activities during their shift including assessment, charting, handoff, medication pass, patient education, care plan development, admissions, and discharges. The findings are organized into three of these specific nursing work practices: charting, handoff, and medication pass. Each of these work practices was a pervasive part of nursing within the units studied at Urban and was directly affected by the Epic system. Although many affordances exist and are actualized in any work practice, we focused on one specific affordance for each work practice because that affordance was both associated with the primary goal of the work practice and provided a compelling example of how affordance potency is conditioned by its component parts. Table 2 lists the affordances, their associated work practices, and the three elements that contribute to affordance potency. Table 2 Work practices, affordances, and contributors to affordance potency at urban hospital. Work practice Affordance Technology features User abilities Social context Charting Documenting Care Handoff Coordinating Care Typing skill, comfort with computers, comfort with patient care Typing skill, communication preferences Care unit structure, patient perceptions Shift change, unit expectations Medication Pass Validating Medications Work stations on wheels (WOWs), Epic software Dear Staff freeform textbox with character limit WOWs, barcode scanners and labels Nursing judgement, equipment use Patient room layout, admin expectations 105 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey Fig. 1. Epic mobile workstation on wheels (WOW). 4.2. Charting and the affordance of documenting care Prior to the implementation of Epic, the documentation of patient care at Urban Hospital was accomplished using a paper-based, three-ring bound patient chart, and “charting” was the term generally used to denote the process of documenting patient care. With the implementation of Epic, the patient chart was digitized and access was made available through both stationary and mobile computer terminals. Of Strong et al.'s (2014) eight named affordances, “capturing and archiving digital data about patients” would include Urban's process of charting in Epic. The devices on which the nurses could chart varied by unit. The CNO explained that, “When we went live, we didn't really have the time, nor did we have the construction ability to get computers in every room. There was just not enough time and there's really quite frankly, not enough space in most of our rooms.” Consequently, most patient care units were equipped with mobile computers called workstations-on-wheels (WOWs) and some fixed desktop computers in the hallways and at the nurse's station. The ICUs, in contrast, had desktop computers at the patient bedside because ICU rooms are larger and the hospital had previously decided to install computers in the ICU rooms to support another information system. WOWs were the computers used most frequently by nurses on the 3West and 5South patient units. Fig. 1 shows a WOW that includes a computer base with Wi-Fi, a flat screen monitor, a keyboard and mouse, a barcode scanner for scanning patients and medications, power cord, and a battery pack, all built on a rolling frame that could be height-adjusted. There is also a wire basket to hold supplies and a desktop surface on which to set objects or to write. Both the WOWs and the in-room computers allowed nurses to chart at the bedside making documentation more timely and accurate. One nurse explained, “I can chart right there when I'm in the room as opposed to when we did paper — you'd have to remember, write something down, and then go and chart it.” Another nurse described how the WOWs changed her work practices. “I take their medications in with me and I do a good assessment and I usually chart it all in the room so that I don't miss anything.” Another benefit of a computer-based patient chart was that nurses and other clinicians had access to relevant data without having to locate the patient's physical chart or other paper documentation. In describing the expected impact on the nurses, the CNO said, “They don't have to hunt for a chart anymore; they don't have to hunt for a MAR [medication administration record] anymore; they don't have to hunt and find all the papers from registration.” The Epic software also provided flexibility in how information was accessed and documented, which was either a benefit or a challenge depending on how the system features fit with a clinician's workflow. “There's a lot of different places to chart the same thing in Epic and I think that can be good and that can be not so good, and the good is that you can find it about anywhere you want. The not so good is that nurses have a harder time getting into the flow of, what do I need to do next because they're all over the place instead of systematically doing it in order.”(Senior Clinical Informatics Analyst) 106 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey The change from paper charts to the use of Epic on WOWs and in-room computers was designed to increase the potency of the affordance of documenting care because Epic enabled more ways to access information and document care than a paper chart. However, the availability of multiple options could also reduce potency if nurses did not sufficiently understand the structure of the electronic health record. Another challenge for some nurses was that using computers in the patient rooms created a tension between caring for their patients and documenting that care on the computer. The CVICU Clinical Coordinator said, “I think the biggest thing that concerns me is the amount of patient contact that you have…looking at a computer screen.” And other nurses expressed concern for what the patients thought of their use of the computers. “I try to work in the room with the computer, but I try not to document in front of the patient because I've had so many patients say to me that nobody really listens to them. It's very easy to be typing and talking to them, but their feeling is that you don't really care or listen, you've just got to check your boxes.”(5South Nurse) In this example, the expectation for documenting care at the bedside was not always actualized in practice because the potency of documenting care was reduced by the nurse's discomfort with documenting care in the presence of the patient. Issues with system features, like the login process, also created new difficulties for nurses in patient care and the documentation of that care. Login time was slow and also fluctuated depending on network load. An ICU nurse explained: “The logging in now is ridiculous. As soon as I walk in my patient's room I log in and then I address the patient. It's just something that you have to get going or otherwise you're standing there and not accomplishing anything while you're waiting for it [the bedside computer] to log in.”(CVICU Nurse) Thus, while the potency for documenting care was high because the chart is always available at the bedside, the nurse had to first get into the system, which wasn't always as easy as flipping open a paper chart. This impediment related to a system feature designed to ensure security lowered the potency of documenting care on Epic. User abilities were influential in the transition from paper to digital charts. Users had to switch from writing in the paper chart to typing in Epic, and relative typing ability significantly affected the time it took a nurse to chart in Epic. “It takes a lot longer sometimes to get things done in an EMR than it does to have that piece of paper that I can set on the bed and put my check marks on…Typing skills are huge. If a person knows how to touch type they went through much less distress than if it's a person that hunts and pecks… I think at first from some end-users it could have been as much as doubling their documentation time if not more.”(Inpatient Clinical Team Lead) General computer knowledge and comfort with computer systems was also a contributor to a nurse's ability to chart with Epic. “I'm 58 and so I did not come up through computers…and this is where I find that I'm not as competent as nurses half my age…I can't always see, okay, have I done all the parts of the chart that I'm supposed to?...When its paper you can flip through and you can see it. When it's electronic, there's a little bit more to going through because it's just not quite as easy for me.”(5South Nurse) Typing ability and general computer skills clearly influenced the potency of documenting care as some nurses struggled to make the transition from paper to digital patient records. In summary, Epic at Urban was designed to facilitate electronic creation, storage and retrieval of confidential patient records. When placed onto mobile carts and/or in fixed locations within the patient care units, the system hardware was close enough to patients to afford nurses the potential to both care for patients and access/record information about those patients as a continuous work practice. However, the actualization of documenting care varied from nurse to nurse and unit to unit and that variability can be traced to the potency of the affordance as influenced by differences in individual nurse's abilities, features of the implemented Epic system, and characteristics of the patient care unit. 4.3. Handoff and the affordance of coordinating care Coordinating care is necessary in inpatient care settings to provide continuity across multiple care providers and/or encounters and multiple work shifts. As the main care givers on patient care units at Urban Hospital, nurses were primarily responsible for maintaining coordinated care through the exchange of information between nurses starting and ending their shifts. The specific procedures for coordinating care across shifts on each unit were at the discretion of the unit's nurse manager, but the general expectation was that nurses ending their shifts would remain on the unit until the nurses starting their shifts had been sufficiently advised on their patients. The procedures for coordinating care between shifts at Urban were collectively referred to as “handoff.” Nurses began their shift by getting their patient assignments. Relevant information on assigned patients was obtained from the patient chart, but the outgoing nurses typically had additional knowledge that was either not in the patient chart or not easy to locate in the chart. For this reason, outgoing nurses were required to “give report” to incoming nurses on the current state of the patients in their care. Prior to the implementation of Epic, giving report on 3West and 5South was done either verbally or by tape recording. With verbal report, nurses sat down face-to-face and discussed each of the patients being handed off. The outgoing nurses typically had notes accumulated during their shift which guided their reports to the incoming nurses. Some patient care units had special “report rooms” where nurses had privacy to give report, while others used a common room, like the nurse's break room, where all nurses congregated at the same time for verbal report. 107 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey “We were face to face for half an hour before and after shifts and it just seems archaic now the way we used to do it. In fact we laugh about how we all used to get together in one room and shout over one another and I don't know why we did that either. We had a whole floor but we'd all be in the break room trying to shout over each other. In fact we were just talking about it the other day, can you believe how we used to give report? That was so crazy!”(5South Nurse) Another drawback to verbal report was lengthy time needed to give a full report on every patient. An alternative method for giving report, which addressed this issue, was the use of tape recorders. Before the end of the shift, the outgoing nurses audio recorded their reports on each patient using a tape recorder. The incoming nurses then listened to the taped reports on their assigned patients when they arrived. The advantage to tape recording was that the outgoing nurses could make the recording anytime during their shift, which reduced the time required to verbally handoff between shifts. The incoming nurses were also able to listen to the recordings at their own pace and refer back to them if necessary after the outgoing nurses had left. These features of recording technology were expected to increase the potency of coordinating care. However, the drawback to tape recording was that information could only be accessed sequentially, and searching the tape for specific information was inefficient. Thus, the benefit of being able to listen to the tape recordings did not strengthened the potency of coordinating care as much as anticipated. With the implementation of Epic and the recognition that existing handoff procedures were limited in their potency to enable coordination of care, some of the nurse managers looked to Epic for ways to change their units' handoff procedures. Nurses' handoff comments were intended to be “off the record” and not retained as part of the patient's permanent chart, and Epic did not have a feature designed specifically for handoff. However, it did have a function called “Dear Staff”, which was a freeform textbox that, unlike the rest of the documentation system, did not become part of the patient's permanent medical record. The Dear Staff function enabled “off the record” communication between all medical staff, but there was a 2000-character limit on the information that could be stored in the textbox and only one instance of the textbox was available per patient for the length of the patient's stay. All staff wishing to use the Dear Staff function shared the same limited space and could delete, modify, or reuse any information previously entered by any other staff member. While not a perfect fit with the handoff process, some of the nurse managers decided to try Epic's Dear Staff function for handoff on their units. Anything that was previously exchanged in the verbal report or included in the tape recordings could be written in the Dear Staff textbox. This preserved the communications as “off the record,” thereby allowing the nurses to speak more colloquially. For example, one nurse explained: “I had a guy that wouldn't get out of bed and in my electronic handoff I said, ‘I've used every skill that I have to get this man out of bed other than dynamite!’ And I can use words like that because it's not part of the permanent record.” The Dear Staff feature was therefore transformed from a space designed for “off the record” communication between all medical staff to a tool that specifically supported the unit nurses' needs for handoff reporting. The Dear Staff content could be modified as often as necessary, so as relevant information was identified during the course of a shift, the nurses could update the content of each patient's textbox to serve as their handoff report for the following shift. A nurse describing her new routine said, “I usually start updating my reports around one o'clock and then, after I make that first update, if anything new comes up I will immediately update my report.” This had the effect of increasing the potency of coordinating care because handoff content could be available whenever the incoming nurses wanted it. “I don't have to come in like old times and say, where's this nurse? I can't find her so I guess I can't start report, I've got to wait. Because that's happened before, I've been ready to get report and they were in another room doing a dressing change or giving pain meds or whatever so you have to sit and wait.”(5South Nurse) Because it was a freeform space, the Dear Staff textbox provided functionality similar to the tape recordings where nurses had the freedom to structure their information as they saw fit. However, one significant difference was that information in the Dear Staff textbox was carried over from shift to shift and could therefore be reused. With tape recording, the expectation was that a nurse started from scratch and included the patient's entire history in each recorded report. “You never have to go through the same story over and over again. You just have to update. Maybe you just have to add a couple of sentences at the end of your shift instead of having to go through the whole thing…I mean tape recording would sometimes take 45 minutes and now updating a handoff takes a few minutes.”(3West Nurse) This had the effect of increasing the potency of coordinating care because nurses could focus their attention on the content that needed to be updated without having to repeat the patient's entire history at the end of each shift. Of course, reuse also had a downside when nurses failed to keep information current. With tape recording or verbal report, the incoming nurse knew that the information was current because it was either just recorded or conveyed verbally. With Dear Staff, the incoming nurse did not necessarily know what information was current and what information was out of date because the system didn't track or time-stamp changes made in the textbox. This reduced the potency of coordinating care, as expressed by one of the nurses: “People don't update the report as thoroughly or as appropriately as they should, because they say, “today had an x-ray and had two units of blood” and that was on day shift. Then it goes to evening shift and they don't change it, so it's saying this shift got two units of blood and you're thinking, okay they just had the blood and then you find out, no that was close to 24 hours ago. So the thing about “today”, “tomorrow”, words like that don't get changed and then it can get confusing.(5South Nurse) The 2000-character limit also became a problem for nurses when patients were in the hospital long enough for their Dear Staff 108 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey textbox to reach that character limit. A nurse explained the difficulty with the character limit as, “Once you have a patient that's been here so long it gets hard to go into much detail on the handoff because you have to go back and figure out what to delete so that you can keep going. You try to delete things that aren't necessary or try to abbreviate things.” The character limit thus also weakened the potency of coordinating care because nurses had to expend effort in manipulating the content rather than simply focusing on providing relevant information. The potency of coordinating care with the Dear Staff feature was also reduced for nurses who were not adept at typing. Although nurses were expected to use Dear Staff for handoff, some verbal communication was often necessary at shift change. Nurses with weaker typing skills therefore took advantage of the opportunity to revert to giving just verbal report. “I'm busy and I have six patients, I might give you verbal because I haven't had time to type it up…I will give verbal if I don't feel that I can type it up fast enough.” In summary, a system feature that was initially designed for general “off the record” communication by all medical staff was appropriated by nurses to better enable the affordance of coordinating care. The decision to change was the result of a perceived problem with the potency for coordinating care using face-to-face or tape recording handoff procedures. The new handoff procedure using the Dear Staff function was more potent because nurses could generate the handoff report during their shift, which reduced the overlap between shifts to enable coordinated care, and relevant handoff information was more easily accessible after the shift transition. However, the limitations of the Dear Staff feature (e.g., 2000-character limit, free form nature of the textbox, no time stamps on individual pieces of information) along with nurses' abilities (e.g. weak typing skill) had the effect of reducing the potency of the affordance. Coordinating care was therefore actualized in different ways as conditioned by the potency of the affordance in use. Potency was affected by the length of stay of patients, technical features of Epic, and the skills of the nurses actualizing the affordance. 4.4. Medication pass and the affordance of validating medications One of the important patient care tasks assigned to nurses on inpatient care units is the administration of medications to their patients. At Urban Hospital this procedure was known as “medication pass” and was performed at four-hour intervals during the day, beginning at 9 am. When physicians gave orders for patient medications, they were added to the patient's medication administration record (MAR) and ordered from the hospital pharmacy. It was the MAR, included in the patient's chart, which nurses used to guide their medication passes. An important affordance for the medication pass process was validating medications. Before passing medications a nurse needed to confirm that five criteria were present: right patient, right medication, right dose, right route, and right time. The right patient was determined by the identification bracelets worn on the wrists of all patients during their hospital stays. Prior to Epic, the nurses were expected to visually check the bracelet to confirm the patient's identity before passing any medications. The labels on the medications were then visually checked against the information on the MAR to confirm the medication, dose, route, and time. Once a match on all five criteria was confirmed, the nurse could then safely administer the medication to the patient. With the implementation of Epic, the hospital moved to an electronic validation process for medication pass. Specifically, Urban began using barcodes to uniquely identify both medications and patients, and the MAR was incorporated into Epic. The patient identification bracelets were printed with a barcode, and the pharmacy applied a barcode to all medication doses that it dispensed. This change was expected to strengthen the potency of validating medications because the computer would be able to verify the five “rights” and inform the nurse if the medication pass was valid or not. As described earlier, the mobile WOWs included barcode scanners to be used for medication pass, requiring the nurses on 3West and 5South to take a WOW into the patient's room to administer medications. A 5South Nurse Coordinator explained, “Now I have to go get the computer, look at the pain medication, go to the Pyxis [medication dispenser], and get the pain medicine. Then I have to go back and get the computer, roll the computer into the room, scan the patient...” Compared to the prior process, the potency of validating medications was weakened by the extra effort required to consult the WOW and maneuver it into the patient's room to complete the medication pass. Once at the bedside, nurses were required to first scan the barcode on the patient's identification bracelet, which automatically opened that patient's MAR in Epic. Some of the nurses did not realize that they had to scan the patient before scanning the medication, and this led to an early increase in system reports of non-compliance with proper scanning procedures. A nurse on CVICU described the difficulties she had with the new medication pass process: “You go in and you scan the medicine and if you haven't scanned the patient first it pops up a little box. If you scan the patient at that time, well then that still reports you as not scanning the patient on their reports. So they write you up over it.” In such instances, the inflexible technology features led to error reports even where no medical errors were made, which further reduced the potency of validating medications. However, the benefit of the new system was that it alerted nurses to possible medical errors before they happened. After scanning the patient first and then the barcode, the nurse received a confirmation from Epic confirming that the scanned medication was correct. If any of the criteria were inaccurate, Epic was designed to display a flag to warn the nurse of the discrepancy. The warning occurred even when an early medication was justifiable, according to a nurse on 5South: “I've done this before where the pain medication was five minutes early. You can give a pain medication by nursing judgement five minutes early if a patient needs it, but it [Epic] will alert you that it's too early.” Nurses had the ability to override the flag in order to continue with the administration of a medication that they determined to be appropriate, but the flags also identified genuine errors that needed to be avoided. A CVICU nurse explained that, “Once or twice I've been saved from making a medication error, nothing drastic, but I think it's prompted me by saying that the amount has been changed or it was the right drug but they changed the dosage…So it's nice to know that it will catch 109 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey those mistakes.” The system features therefore increased the potency of the affordance of validating medications, which resulted in a reduction in medication errors. The new procedures for medication pass became the standard at Urban Hospital, and nursing administrators tracked nurses' compliance with the prescribed procedures through Epic's capture of system events. Since Urban was legally liable for medication errors, nursing administrators realized greater control by preventing deviations from Epic's mandatory work process. The CNO explained the hospital's policy on compliance with the medication pass procedures: “There's only one way to do barcode med administration and the policy says you have to do it at the bedside, so people who don't do that are automatically in the discipline system. Now there are some exceptions to that in an [emergency] code situation or if the scanners are down, but every month I get a report and so does every manager and every director. And I can tell every nurse in this hospital how many doses of medication they gave, how many times they bar-coded the medicine and how many times they bar-coded the patient. So if people are in what I call the 50/50 club, we have a very different conversation than the 95/5 club. Meaning 95% of the time I did it versus 50%. We gave them some leeway in the beginning, they were learning, but now, if you're in the 50/50 club, you're probably not going to have a very long tenure at [Urban] Hospital.”(Chief Nursing Officer) The goal of 95% compliance conceded that there would be occasional circumstances that necessitated a deviation from the ideal standard of 100% compliance, but exceptions were to be limited. However, there were a number of issues with the scanning process that made some of the nurses feel that even 95% compliance was unrealistic. One example, noted above, was inflexibility of the system to deal with the case of nursing judgement to give pain medications as needed instead of at a specific time. A 5South nurse explained, “It [the prescription] says you can have ‘one to two’ tablets and you just give them one and then three hours later you're going back to give them the second one. It's still within their time where they can get the second one, but because it's a new scan of the same medication, it's coming up saying that it might be too early.” The issue of timing also pertained to the administration of IV-based medications where adjusting the flow of the medication could throw off the timing of subsequent doses. The pharmacy entered times for each dose, usually every hour from the start time, but as a 3West nurse explained, “Inevitably we'll never get it done because we'll have to decrease it and give it over two hours instead of one hour so then it makes the meds late and overdue…and I don't think there's a way of getting around it because Epic won't allow you to not time it.” In other words, the inflexibility of Epic to enable adjustment in medication timings by nurses on the unit had the effect of reducing the potency of validating medications by generating inappropriate errors. Another issue in the new medication pass process was that in some cases, the packaging of the medications made it difficult or impossible to scan the barcode. For examples, many medications were prescribed as one or two pills that were packaged in a small wrapper that was smaller than the barcode label. This made it difficult or impossible to scan the barcode. Also, bagged medicines might crinkle the barcode because the container was soft and pliable. A 3West nurse stated: “So you know some of our meds won't scan and we get dinged for pharmacy not being able to label it correctly.” Here, the physical characteristics of the medications themselves had a weakening effect on the potency of validating medications, making it harder to achieve the goal of 95% compliance with the medication pass standard. In some cases, nurses intentionally bypassed the scanning process because they felt that following it would negatively impact patient care. A nurse explained her decision to bypass the scanning process, “if you're in a hurry and it's an emergency you just give the med, you don't worry about the scanning stuff. You take care of the patient before you do the computer.” The system features that enabled medication pass and the affordance of validating medications were relatively modest (bar code scanners), and they were relatively easy to employ as long as the WOW carts could be moved close enough to the patient. Even so, the potency of the affordance depended on the ability of nurses to obtain clean scans and follow the exact process as regulated by Epic. Since nurses needed to vary the timing of medication administration in the interests of patient care, minor deviations (5%) were tolerated, but nurse administrators clearly did not tolerate nurses that regularly worked around the affordance of validating medications. Validating medications thus enabled both nurses' goals to administer medications accurately and administrators' goals to reduce liability for medication errors. 5. Discussion As formally defined earlier, affordance potency is the strength of the relationship between the abilities of the individual and the features of the system at the time of actualization, conditioned by the characteristics of the work environment. We selected three work practices and one of the affordances within each of those work practices from our study to illustrate the concept of affordance potency. Table 3 summarizes our results for the affordances of documenting care, coordinating care, and validating medications and notes characteristics of system features, user abilities, and the work environment (context) that either strengthened or weakened the potency of those affordances along with representative quotes. We started with the work practice of charting and the affordance of documenting care. The potency of documenting care was intended to be strengthened through the implementation of Epic where information was accessible on multiple devices instead of being limited to a single paper chart. In practice, however, the potency of documenting care varied from nurse to nurse based on their relative ability to manipulate the electronic record and the ease with which they shifted attention between providing direct care to patients and recording the delivery of that care. The technology itself was also varied in its influence. The computer login process took extra time and also fluctuated based on the network load, and it always took longer to get into the system to find information that it would have taken to flip open the paper chart. The environment also impacted the potency of documenting care. The administration's expectation was for nurses to document care at the bedside, but patients' perceptions regarding care could alter the 110 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey Table 3 Affordance potency at urban hospital. Affordance Documenting care Coordinating care Validating medications Features that strengthen potency Abilities that strengthen potency Context that strengthens potency Mobility of WOWs, flexibility of Epic software Strong typing skills, comfortable with technology, skilled at patient care Support from supervisors Consistency of system to evaluate the five rights Familiarity with scanner and scanning procedures Expectation of compliance by administration Examples of strengthened potency “I can chart right there when I′m in the room as opposed to when we did paper you'd have to remember, write something down and then go and chart it.” (3West Nurse) Dear Staff feature accessible at any time, “off record” capability Strong typing skills, comfortable with technology Transitions between shifts are unpaid for either the incoming or outgoing nurse, clear expectations for structuring handoff report content “You never have to go through the same story over and over again. You just have to update. … I mean tape recording would sometimes take 45 min and now updating a handoff takes a few minutes.” (3West Nurse) Features that weaken potency Slow login, bulkiness of WOWs, complexity of software Abilities that weaken potency Context that weakens potency Weak typing skills, discomfort with computers, discomfort with patient care Patients that see the technology as a barrier Examples of weakened potency “It takes a lot longer sometimes to get things done in an EMR than it does to have that piece of paper that I can set on the bed and put my check marks on…Typing skills are huge. If a person knows how to touch type they went through much less distress than if it's a person that hunts and pecks.” (Clinical Team Lead) Freeform structure and character limit of Dear Staff textbox, open editing with no time stamps of Dear Staff textbox Weak typing skills, preference for verbal communication No fixed standards for content structure of handoff report, no restrictions on reverting to verbal handoff “Once you have a patient that's been here so long it gets hard to go into much detail on the handoff because you have to go back and figure out what to delete so that you can keep going. You try to delete things that aren't necessary or try to abbreviate things.” (5South Nurse) “I went to give a medication, I looked at the medication, it sounded like, looked like, acted like, but it was the wrong medication. I went and scanned it and [Epic says] there is no medication for this … This is what they call a near miss! … the computer just saved me from making a med error.” (3West Nurse) Rigidity of system to not accommodate for nursing judgement, cluttered patient rooms Unfamiliarity with med pass procedures Barcodes not always scan able, variations in timing of med delivery “So you know some of our meds won't scan and we get dinged for pharmacy not being able to label it correctly.” (3West Nurse) nurse's decision to adhere to that expectation. These variations in the actualization of the affordance of documenting care can be understood through the concept of affordance potency. At each moment of actualization the current state of the system's features, the abilities of the nurse, and the environment come together to produce the potency of the affordance. Coordinating care was the focal affordance within the work practice of handoff, which was enabled through Epic by nurses' appropriation of the Dear Staff feature. The Dear Staff feature was attractive because of the limited potency of coordinating care through either verbal handoff or tape recording. The Dear Staff system feature was not designed for the handoff process, but the nurses found that it would more effectively enable their goal of maintaining quality care across shifts while minimizing the time required to transfer necessary information between shifts. Here the potency of the affordance of coordinating care was shaped by the material characteristics of the Dear Staff feature (e.g., a 2000-character limit, freeform structure, only one instance per patient), the abilities of the nurses (e.g., typing skill and different ways of structuring information in the freeform textbook), and the environment (e.g., time constraints to fill in the textbox before shift change). Once again, the actualization of the affordance, in this case coordinating care, was influenced by the variations in potency of the affordance. The barcode-based medication pass process was designed to function within a set of standardized parameters to maximize the affordance of validating medications, but those parameters were not always aligned with use in practice. Deformed labels made it impossible to scan some barcodes, and timing issues with certain medications produced errors that showed up as non-compliance on administration reports. In some cases, errors resulted from misunderstanding how to perform the barcoding process (e.g., the requirement to scan the patient's ID before scanning a medication), but in other cases nurses deliberately chose to bypass the barcode system if they felt that patient care would be compromised (e.g., when the patient required pain medication before a scheduled time). However, since barcode scanning was associated with a reduction of potential medication errors, as detected by Epic, the potency of validating medications generally met nurse's goals to provide accurate patient care. Our aim in this research is to offer a more complete explanation of affordance actualization. Clearly, technology features are subject to a range of actualizations across users and work environments, previously reported in the IS literature as workarounds or unintended consequences (Azad & King, 2008; Harrison, Koppel, & Bar-Lev, 2007). Our evidence affirms that affordances can be actualized in a variety of ways by users seeking to attain specific goals. However, users' goals and perceptions must be understood in relation to specific technology features, users' abilities, and the wider organizational context in which technologies are used. While previous studies have generally treated affordances in binary fashion (McGrenere & Ho, 2000), our study reveals the importance of an additional concept that we have labeled affordance potency. Rather than simply being either available or unavailable for actualization based on the perceptions and goals of the user, affordances are available at different levels of potential 111 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey Fig. 2. A Model of Affordances (shaded elements are the focus of this research). based on the current state of implemented system features and the user's abilities within the actual context of use. For example, documenting care using Epic was improved through the availability of WOWs that enabled documentation at the bedside, but was also influenced by user abilities (e.g., typing skills) and the context in which the activity occurred (e.g., caring for patients and meeting their expectations for attention). The potency of documenting care was limited due to the physical requirement to access an available workstation and log in; yet it enabled greater efficiency in accessing patient charts and updating them. Handoff also varied in its potency to enable the affordance of coordinating care based on the amount of information to be conveyed for each patient. Because of the character limit in the Dear Staff feature that nurses appropriated for handoff, a variety of actualizations were observed with different results. For example, some nurses reverted to verbal handoff because their typing skills were weak while others expressed confusion about the timing of Dear Staff entries made during the prior shift. These limitations in potency could be traced to both human abilities and technology features. Validating medications was perceived to be a high potency affordance of the barcode-based medication pass process because nurses and administrators expressed confidence that the system could catch medication errors. System features imposed discipline on the work process of administering patient medication, with a resultant reduction in potential errors. Nonetheless, the strict discipline could be bypassed by nurses who did not understand the system or by nurses who chose to administer dosages (e.g., pain medication) over a longer time period than authorized by Epic. Variability in the quality of the barcodes and the need to get a barcode scanner close enough to the patient also reduced potency and frustrated nurses' attempts to provide adequate patient care while meeting hospital standards for compliance with required practices. To encapsulate our insights, we construct a conceptual model of affordances, as shown in Fig. 2. The left side of the model depicts how technology is typically designed and constructed to achieve a particular purpose, which is to enable a set of planned affordances that are a relationship between designed system features and anticipated user abilities within an expected context. For example, designers of Epic sought to enable the affordance of validating medications through the development of algorithms that work with barcodes and the electronic MAR to verify the five “rights” of medication pass. Designers envision their system resulting in the consistent actualization of that affordance when used by an imagined “typical nurse” with the ability to use the barcode scanners in the prescribed way within a clinical context where hospital administrators have established standards for medication pass behaviors. However, situated affordances may vary significantly from planned affordances because of differences in the way system features are implemented and variations in the abilities of real users within the actual work context. In particular, these situated affordances, as behavioral opportunities for action, possess a potency that is a product of that variability which, in addition to user's perception and goals, influences the actualization of those affordances. The horizontal arrows in Fig. 2 represent sequential relationships, not causal dependencies. They simply suggest the order in which IS solutions are typically designed prior to actual use, and that affordances are actualized through the milieu of work practice. This is, therefore, a process model rather than a variance model in which elements to the left are necessary but not sufficient precursors of elements to the right. The solid vertical arrows within the left and center boxes of the diagram represent the components of affordances at each stage of the model where together, system features and user abilities form affordances within a context of use. In addition, while planned affordances could be conceptualized as possessing potency (e.g., arguments could be made that one designed system feature will lead to a more potent affordance than another designed system feature), only situated affordances possess potency in the context of actual work practice. The dashed vertical arrows in the right-hand portion of the diagram illustrate that, in addition to potency, actualization of situated affordances are influenced by user perception and goals as theorized in prior research (Norman, 1988; Strong et al., 2014). Consequently, actualizations may deviate from intended outcomes as users vary their work practices in response to changes in system features (Leonardi, 2011, 2012). Such deviations are often reported as reinventions, tweaks and workarounds in IS research (e.g., Boudreau & Robey, 2005). The concept of affordance potency contributes to studies of unintended use by tracing the antecedents of variant practices back to the relationship between user abilities and system features. Given that the social context of use (work practices) help to shape user goals, resulting appropriations of IT features can be better understood as the actualization of 112 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey affordances with greater or lesser potency. As shown, our model does not represent a complete theory of work practice or IT-related organizational change. Rather, it is a mid-range theory similar to that provided by Strong et al. (2014). Such mid-range theories might be fit into more encompassing theories that are already used in IS research. For example, routines (Pentland & Feldman, 2008); learning (Boudreau & Robey, 2005), and imbrication (Leonardi, 2011) have all been used to explain how changes in IT affect work practices. Rather than extend our analysis outward to any one of these theories, our contribution is to develop further the “affordance module” that focuses on the potency of affordances that are referred to in a variety of theories. We specifically augment the affordance actualization theory of Strong et al. (2014) by adding the concept of affordance potency, a concept that helps to explain more completely how and why specific affordances are actualized within the context of work practice. 6. Conclusions Through a qualitative case study of EMR use on three hospital patient care units, we illustrated the elements of affordance theory and assembled a model explaining the progression from intended design based on expected social context, to actual social context, and to affordance actualization. By invoking the novel concept of affordance potency, we contribute to the nomological network needed to understand the position of affordances in changing work practices. We used three specific affordances to demonstrate the process of affordance actualization as articulated in prior research (Strong et al., 2014). As a mid-range theoretical contribution, the nomological network of affordances may be incorporated into grander theories (e.g., organizational learning, routines, politics) to explain IT-related organizational change. Although our analysis depends on qualitative research methods, our ontological assumptions are realist in nature; we do not ascribe to a purely interpretive approach. Accordingly, the research method includes direct engagement with the EMR artifact and associated hardware (e.g., WOWs, barcode scanners) in the physical work setting. We combined this engagement with the technology with methods designed to elicit nurses' and administrators' experiences with that technology. Through interviews and observation, we obtained detailed accounts of changes in work practices. This inclusive approach to collecting data allowed us to theorize about the relationship between the IT artifact and its social context, which is inherent in our concept of affordances. The study is not without limitations. The data for this study were collected during a single month period and relied in part on retrospective accounts of work practices and other activities occurring prior to the data collection period. Although perceptions and interpretations may change over time, Leonard-Barton suggests that, “studies have shown that the participants in organizational processes do not forget key events in these processes as readily as one might suppose” (1990, p. 250). Given the salience of EMR implementation at Urban Hospital, we do not regard this limitation to be serious enough to invalidate our analysis. Nonetheless, a true longitudinal research design could result in more accurate accounts of events as they occur. Leonardi's (2012) ethnographic study is exemplary in this regard. Affordance theory is clearly undergoing development rather than being fully formed. We traced the relevant controversies in ecological psychology to gain a sense of issues that need to be addressed in IS research. Based on our review of this literature, we adopted the relational view of affordances that appears to be favored in IS research (Robey et al., 2013; Strong et al., 2014). Our data permit detailed analyses of relationships between human abilities and technology features, thus affirming the value of this relational concept. However, affordance theory needs to be refined through future research in different types of work settings. Since IS researchers are advantageously positioned to understand IT artifacts, we see a great opportunity for them (as opposed, say, to less technically savvy social scientists) to develop affordance theory further. Advancing affordance theory would help IS researchers to theorize the IT artifact more directly than has been the case in past research (Orlikowski & Iacono, 2001). Our study offers implications for practice. Specifically, the insights regarding affordance potency may assist healthcare administrators and clinical decision makers in their evaluation of new information systems for their organizations and the policies they implement regarding the use of those systems. Taking an affordance perspective should lead to an evaluation of possible changes in work practice enabled by the system in conjunction with clinicians' skills. Studies often show that end users adjust their work practices in both intended and unintended ways (Markus & Robey, 2004), suggesting that a closer awareness of the adjustments to work practice deserves scrutiny. First-line supervisors, in particular, would benefit from sensitization to affordances that emerge from the relationship between human skills and technology. In addition, we focus on a neglected category of the healthcare workforce: nurses. Most IS studies in healthcare organizations have focused primarily on physicians and their use of health information technology (Davidson & Chismar, 2007; Kohli & Kettinger, 2004; Reardon & Davidson, 2007). However, nurses are typically hospital employees (as opposed to physicians who practice privately), and their work is subject to monitoring and evaluation. We found that the Epic system generated data useful to hospital administrators regarding medication pass, suggesting the greater role played by EMR beyond direct patient care. While our focus is not on organizational-level affordances, we detected some tension between the goals of nurses to provide effective personalized care to their patients and the goals of the hospital to exert control over medical care more broadly by monitoring staff compliance with Epic's procedural disciplines. Future research could examine how well individual actualizations of affordances align with the goals of the organization (Strong et al., 2014). Advances in healthcare are associated not only with improvements in clinical outcomes but also with organizational efficiencies related to human resource management and patient workflow. Appendix A Nurse 113 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey • What are your work roles and responsibilities? • Describe the current system and your use of the system. • How much experience have you had with the system? • Have you had experience with similar systems? If so how does the current system compare to those systems? • What kind of training have you received on the current system? Was it sufficient? • What features or functions do you use the most? • Are there features of the system you are aware of but don't know how to use? • Are you satisfied with the current system? Why or why not? • Does the system function in the way it is supposed to? If not, what functions do not work properly? • Are there features not currently available that you would like to see in the system? • How has the implementation of this system affected work on the patient care unit? • How has the implementation of this system affected the users? Administrator • What are your work roles and responsibilities in relation to the system? • Why did you purchase a new system? • Describe the process that was used to select the current system. • What systems did you consider when making the selection? • Why did you choose the current system? Where there any features which were particularly important in the selection process? • How important was ease of use and training requirements in the selection decision? • Are you satisfied with the current system? Why or why not? • Are the users satisfied with the current system? • How has the implementation of this system affected the patient care unit? • How has the implementation of this system affected the users? Support Personnel • What are your work roles and responsibilities in relation to the system? • What kind of training have you received on the installed system? • How would you describe your understanding of the system? • Describe the system and its installation. • Describe the user training for the system. • Describe the current maintenance of the system. • What features of the system are most important for users to know about and use? • Are there particular features that users have more difficulty understanding and using? • Does the system have any recurring problems? • Are there features the system does not have that users have asked for? References Archer, M., Bhaskar, R., Collier, A., Lawson, T., & Norrie, A. (1998). Critical realism: Essential readings. New York, NY: Routledge. Azad, B., & King, N. (2008). Enacting computer workaround practices within a medication dispensing system. European Journal of Information Systems, 17, 264–278. Benbasat, I., & Zmud, R. W. (2003). The identity crisis within the IS discipline: Defining and communicating the discipline's core properties. MIS Quarterly, 27(2), 183–194. Bhaskar, R. (1978). A realist theory of science. Sussex: The Harvester Press Limited. Boudreau, M.-C., & Robey, D. (2005). Enacting integrated information technology: A human agency perspective. Organization Science, 16(1), 3–18. Brown, D. C., & Blessing, L. (2005). The relationship between function and affordance. Paper presented at the ASME 2005 International Design Engineering Technical Conferences & Computers and Informaiton in Engineering Conference. (Long Beach, California). Chemero, A. (2003). An outline of a theory of affordances. Ecological Psychology, 15(2), 181–195. Chu, T. H., & Robey, D. (2008). Explaining changes in learning and work practice following the adoption of online learning: A human agency perspective. European Journal of Information Systems, 17(1), 79–98. Dalgarno, B., & Lee, M. J. W. (2010). What are the learning affordances of 3-D virtual environments? British Journal of Educational Technology, 41(1), 10–32. Davidson, E. J., & Chismar, W. G. (2007). The interaction of institutionally triggered and technology-triggered social structure change: An investigation of computerized physician order entry. MIS Quarterly, 31(4), 739–758. DeSanctis, G., & Poole, M. (1994). Capturing the complexity in advanced technology use: Adaptive structuration theory. Organization Science, 5(2), 121–147 (doi:citeulike-article-id:327342). Eisenhardt, K. M. (1989). Building theories from case study research. Academy of Management Review, 14(4), 532–550. Eisenhardt, K. M., & Graebner, M. E. (2007). Theory building from cases: Opportunities and challenges. Academy of Management Journal, 50(1), 25–32. Fayard, A.-L., & Weeks, J. (2014). Affordances for practice. Information and Organization, 24, 236–249. Feldman, M. S., & Pentland, B. T. (2003). Reconceptualizing organizational routines as a source of flexibility and change. Administrative Science Quarterly, 48(1), 94–118. Gamage, V., Tretiakov, A., & Crump, B. (2011). Teacher perceptions of learning affordances of multi-user virtual environments. Computers & Education, 57, 2406–2413. Gibson, J. J. (1979). The ecological approach to visual perception. Boston, MA: Houghton Mifflin Company. Giddens, A. (1984). The constitution of society: Outline of the theory of structuration. Univ of California Press. Harrison, M. I., Koppel, R., & Bar-Lev, S. (2007). Unintended consequences of information technologies in health care — An interactive sociotechnical analysis. Journal 114 Information and Organization 27 (2017) 100–115 C. Anderson, D. Robey of the American Medical Informatics Association, 14(5), 542–549. Heft, H. (1989). Affordances and the body: An intentional analysis of Gibson's ecological approach to Visual perception. Journal for the Theory of Social Behaviour, 19(1), 1–30. Hutchby, I. (2001a). Conversations and technology: From the telephone to the internet. Malden, MA: Blackwell Publishers Inc. Hutchby, I. (2001b). Technologies, texts and affordances. Sociology, 35(2), 441–456. Jones, M. R., & Karsten, H. (2008). Giddens's structuration theory and information systems research. MIS Quarterly, 32(1), 127–157. Kling, R. (2000). Learning about information technologies and social change: The contribution of social Informatics. The Information Society, 16, 217–232. Kohli, R., & Kettinger, W. J. (2004). Informating the clan: Controlling physicians' costs and outcomes. MIS Quarterly, 28(3), 363–394. Leonard-Barton, D. (1990). A dual methodology for case studies: Synergistic use of a longitudinal single site with replicated multiple sites. Organization Science, 1(3), 248–266. Leonardi, P. M. (2011). When flexible routines meet flexible technologies: Affordance, constraint, and the imbrication of human and material agencies. MIS Quarterly, 35(1), 147–167. Leonardi, P. M. (2012). Car crashes without cars: Lessons about simulation technology and organizational change from automotive design. Cambridge, MA: MIT Press. Leonardi, P. M. (2013). When does technology use enable network change in organizations? A comparative study of feature use and shared affordances. MIS Quarterly, 37(3), 749–775. Leonardi, P. M., & Barley, S. R. (2008). Materiality and change: Challenges to building better theory about technology and organizing. Information and Organization, 18(3), 159–176. Maier, J. R., & Fadel, G. M. (2009). Affordance based design: A relational theory of design. Research in Engineering Design, 20, 13–27. Majchrzak, A., Markus, M. L., & Wareham, J. (2016). Desiging for digital transformation: Lessons for information systems research from the study of ICT and societal challenges. MIS Quarterly, 40(2), 267–277. Markus, M. L., & Robey, D. (2004). Why stuff happens: Explaining the unintended consequences of using IT. In K. V. Andersen, & M. T. Vendelo (Eds.), The past and future of information systems (pp. 61–93). Oxford: Elsevier Butterworth, Heinemann. Markus, M. L., & Silver, M. S. (2008). A foundation for the study of IT effects: A new look at DeSanctis and Poole's concepts of structural features and spirit. Journal of the Association for Information Systems, 9, 609–632. McGrenere, J., & Ho, W. (2000). Affordances: Clarifying and evolving a concept. Paper presented at the Graphics Interface 2000. Montreal: Canada. Michaels, C. F. (2000). Information, perception, and action: What should ecological psychologists learn from Milner and Goodale (1995)? Ecological Psychology, 12(3), 241–258. Miles, M. B., & Huberman, A. M. (1994). Qualitative data analysis: An expanded sourcebook(2nd ed.). Thousand Oaks, CA: Sage Publications, Inc. Mingers, J. (2004). Real-izing information systems: Critical realism as an underpinning philosophy for information systems. Information and Organization, 14(2), 87–103. Mutch, A. (2013). Sociomateriality — Taking the wrong turning? Information and Organization, 23, 28–40. Norman, D. A. (1988). The psychology of everyday things. New York, NY: Basic Books, Inc. Orlikowski, W. J. (2000). Using technology and constituting structures: A practice lens for studying technology in organizations. Organization Science, 11(4), 404–428. Orlikowski, W. J., & Iacono, C. S. (2001). Research commentary: Desperately seeking the 'IT' in IT research — A call to theorizing the IT artifact. Information Systems Research, 12(2), 121–134. Pentland, B. T., & Feldman, M. S. (2008). Desiging routines: On the folly of desiging artifacs, while hoping for patterns of action. Information and Organization, 18, 235–250. Reardon, J. L., & Davidson, E. (2007). An organizational learning perspective on the assimilation of electronic medical records among small physician practices. European Journal of Information Systems, 16(6), 681–694. Reed, E. S. (1996). Encountering the world: Toward an ecological psychology. New York, NY: Oxford University Press. Robey, D., Anderson, C., & Raymond, B. (2013). Information technology, materiality, and organizational change: A professional odyssey. Journal of the Association for Information Systems, 14(7), 379–398. Robey, D., & Boudreau, M.-C. (1999). Accounting for the contradictory organizational consequences of information technology: Theoretical direction and methodological implications. Information Systems Research, 10(2), 167–185. Schultze, U., & Boland, R. J., Jr. (2000). Knowledge management technology and the reproduction of knowledge work practices. Journal of Strategic Information Systems, 9, 193–212. Stoffregen, T. A. (2003). Affordances as properties of the animal-environment system. Ecological Psychology, 15(2), 115–134. Stoffregen, T. A. (2004). Breadth and limits of the affordance concept. Ecological Psychology, 16(1), 79–85. Strong, D. M., Volkoff, O., Johnson, S. A., Pelletier, L. R., Tulu, B., Bar-On, I., ... Garber, L. (2014). A theory of organization-EHR affordance actualization. Journal of the Association for Information Systems, 15(2), 53–85. Sutcliffe, A. G., Gonzalez, V., Binder, J., & Nevarez, G. (2011). Social mediating technologies: Social affordances and functionalities. International Journal of Human Computer Interaction, 27(11), 1037–1065. Turvey, M. T. (1992). Affordances and prospective control: An outline of the ontology. Ecological Psychology, 4(3), 173–187. Turvey, M. T. (2009). On the notion of implications of organism-environment system. Ecological Psychology, 21, 97–111. Vaast, E., & Walsham, G. (2005). Representations and actions: The transformation of work practices with IT use. Information and Organization, 15, 65–89. Volkoff, O., & Strong, D. M. (2013). Critical realism and affordances: Theorizing IT-associated organizational change processes. MIS Quarterly, 37(3), 819–834. Warren, W. H. (1984). Perceiving affordances: Visual guidance of stair climbing. Journal of Experimental Psychology: Human Perception and Performance, 10(5), 683–703. Yin, R. K. (2003). Case study research: Design and methods(3rd ed.). Beverly Hills, CA: Sage Publications Inc. Zammuto, R. F., Griffith, T. L., Majchrzak, A., Dougherty, D. J., & Faraj, S. (2007). Information technology and the changing fabric of organization. Organization Science, 18(5), 749–762. 115