Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Impact of information technology on health care professionals and patient care : a multiple case study… Hebert, Marilynne Arlayne 1996

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1996-147576.pdf [ 20.02MB ]
Metadata
JSON: 831-1.0087831.json
JSON-LD: 831-1.0087831-ld.json
RDF/XML (Pretty): 831-1.0087831-rdf.xml
RDF/JSON: 831-1.0087831-rdf.json
Turtle: 831-1.0087831-turtle.txt
N-Triples: 831-1.0087831-rdf-ntriples.txt
Original Record: 831-1.0087831-source.json
Full Text
831-1.0087831-fulltext.txt
Citation
831-1.0087831.ris

Full Text

IMPACT OF INFORMATION T E C H N O L O G Y ON H E A L T H C A R E PROFESSIONALS A N D PATIENT C A R E : A MULTIPLE CASE STUDY IN C O M M U N I T Y HOSPITALS by M A R I L Y N N E A R L A Y N E HEBERT B.Sc.N., The University of Alberta, 1980 M.Ed., The University of Alberta, 1986 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES Department of Health Care and Epidemiology We accept this thesis as conforming to/|fhe required star^Jar^ THE UNIVERSITY OF BRITISH C O L U M B I A September 1996 ©Marilynne Arlayne Hebert, 1996 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia Vancouver, Canada Date DE-6 (2/88) ABSTRACT The purpose of this study was to investigate two questions: 1) What, i f any, is the perceived difference in impact of Patient Care Information Systems (PCIS) on health care professional groups in hospitals? 2) What factors explain such perceived differences in impact among these groups? A multiple-case methodology was used as this approach permitted factors within the organizational context to be considered. Four professional groups participated in the study, including laboratory technologists, pharmacists, nurses and physicians. They were located in five community hospitals in British Columbia that had achieved various levels of PCIS implementation. From the literature review an analytic framework based on successful information technology (IT) in other industries was developed to investigate impact. To more closely reflect the goals of IT in the health care setting, the framework was adapted by using Donabedian's (1988) three measures of quality (structure, process, outcome). For each participant group these measures were examined at Grusec's (1986) three levels of impact: substitution, proceduralization and new capabilities. Three types of data were collected: interviews, document review and observation. Eighty-five semi-structured interviews were conducted and a selected number of participants were observed using the PCIS. Written documentation and archival material relevant to the adoption and use of the PCIS were reviewed for each site. These included IT proposals, strategic IT plans, task force and steering committee minutes, internal and external correspondence. The data were analyzed manually and with textbase management software ii called FolioVIEWS® 3.1 for Windows. This software facilitated the searching of interview transcripts in ways defined by the investigator and building hypertext linkages in the data. The study findings include differences across hospitals and groups. The hospitals are community based and differ on one important variable: the extent of PCIS implementation. Differences are expected to be related to this variable. Generally this is true, as Hospitals 1 and 2 demonstrate more evidence of impact in structure at levels 1 and 2 for all groups. As expected, Hospitals 4 and 5 do not demonstrate this level of impact. However, unexpectedly these two hospitals are able to identify specific evidence of linkage between the measures of structure, process and outcome. Five themes emerge that illustrate the perceived differences among the professional groups with respect to the impact of PCIS use. The first theme relates to increased efficiency and productivity that result from automating clerical tasks, particularly sending and receiving orders. This theme relates more to pharmacists and laboratory technologists as changes in structure take place when their tasks are automated. The second theme relates to role and responsibility changes, particularly those experienced by nurses and physicians. This is partly due to the expected changes in process (e.g. decision-making) arising from changes in structure (e.g. faster lab results). Role changes also occur when technology can be used by one group to accomplish tasks formerly completed manually by another group. For example, historically physicians write orders on paper charts and unit secretaries transcribe those orders onto paper requisitions, which are then sent to the appropriate department. Nurses are responsible for ensuring this is done correctly and the results forwarded to physicians as needed. When physicians or nurses enter orders directly into the computer, all three roles change. The participants are unable to identify how they expect these changes to unfold. "Visible" accountability is a change experienced by all four groups. For laboratory technologists and pharmacists this change occurs primarily in measures of structure. Automating tasks such as reporting results or medication profiles creates an electronic audit trail that documents the volume and accuracy of work accomplished (e.g., the number of orders processed, number of errors). For nurses and physicians, this change occurs in the process measures as decisions with respect to care are recorded and immediately available for inspection by other professional groups. All four groups experience unexpected consequences of electronic communication. The levels of electronic mail (e-mail) use varies by system availability in each hospital and demonstrates changes in structure. This medium replaces a paper-based system as well as supports new opportunities for committee and group work through multiple access to single documents. Participants predict other uses in use of on-line clinical practice guidelines and interdisciplinary documentation of patient care. The fifth theme, training to use technology versus learning to use information, suggests that users must be able to do more than simply "use the technology" to achieve the potential benefits. This affects pharmacists, nurses and physicians in particular as they attempt to determine linkages between structure, process and outcome. The study contributes to understanding the impact of IT in health care by identifying where differences between professional groups in community hospitals occur. The study concludes that linkages between structure and process or process and outcome must be iv determined before users can expect the computer system to have the intended effect. One example of this is when changes in the work of laboratory technologists and pharmacists (structure) is expected to change the work of nurses and physicians (process). Use of IT also creates role and responsibility changes that contribute to its impact. TABLE OF CONTENTS Abstract ii Table of Contents vi List of Tables ix List of Figures x Acknowledgements xii Chapter 1. INTRODUCTION 1.1 Why is the Study of Impact Important? 1 1.2 Objectives of the Study 5 1.3 Organization of the Thesis 6 Chapter 2. THEORETICAL FRAMEWORK 2.0 Introduction 8 2.1 The Role of IT in Organizational Change 9 2.2 Technological Imperative in IT and Change 14 2.3 Organizational Imperative in IT and Change 17 2.4 Socio-Technical Perspective in IT and Change 27 2.5 Analytic Framework 40 Chapter 3. RESEARCH METHODOLOGY 3.0 Introduction 44 3.1 The Case Study Approach 44 3.2 Case Selection 45 3.3 Unit of Analysis 49 3.4 Data Collection 49 3.5 Data Analysis 55 3.6 Issues in Reliability and Validity 57 3.7 Potential Contribution of the Research Methodology .. 61 Chapter 4. FINDINGS: Comparison of Impact Across Groups Within Hospital 1 4.0 Introduction 63 4.1 Impact on Laboratory Technologists 65 4.2 Impact on Nurses 80 4.3 Impact on Pharmacists 92 4.4 Impact on Physicians 106 4.5 Summary 118 Chapter 5. FINDINGS: Comparison of Impact Across Groups Within Hospital 2 5.0 Introduction 128 5.1 Impact on Laboratory Technologists 129 5.2 Impact on Nurses 138 5.3 Impact on Pharmacists 148 5.4 Impact on Physicians 156 5.5 Summary 164 Chapter 6. FINDINGS: Comparison of Impact Across Groups Within Hospital 3 6.0 Introduction 173 6.1 Impact on Laboratory Technologists 174 6.2 Impact on Nurses 189 6.3 Impact on Pharmacists 200 6.4 Impact on Physicians 216 6.5 Summary 228 Chapter 7. FINDINGS: Comparison of Impact Across Groups Within Hospital 4 7.0 Introduction 235 7.1 Impact on Laboratory Technologists 236 7.2 Impact on Nurses 248 7.3 Impact on Pharmacists 262 7.4 Impact on Physicians 274 7.5 Summary 283 Chapter 8. FINDINGS: Comparison of Impact Across Groups Within Hospital 5 8.0 Introduction 289 8.1 Impact on Laboratory Technologists 292 8.2 Impact on Nurses 298 8.3 Impact on Pharmacists 308 8.4 Impact on Physicians 318 8.5 Summary 322 Chapter 9. DISCUSSION: DIFFERENCES IN IMPACT ACROSS GROUPS 9.0 Introduction 329 9.1 Increased Efficiency and Productivity 333 9.2 Role and Responsibility Changes 343 9.3 "Visible" Accountability 354 9.4 Unexpected Consequences of Electronic 356 Communication 9.5 Training to Use Technology versus Learning to Use 358 Information 9.6 Conclusion 360 vii Chapter 10. CONCLUSIONS 10.0 Introduction 365 10.1 Research Questions 365 10.2 Implications of the Study Themes 366 10.3 Revisiting the Theoretical Model 371 10.4 Implications for Further Research 373 10.5 Implications for Policy and Practice 373 10.6 Limitations and Contributions of the Research Study .. 375 REFERENCES 381 APPENDICES A . Hospitals in British Columbia Over 120 Beds With PCIS 404 B. Correspondence With Sites 405 B. 1 Covering Letter to CEO Inviting Participation in the Study B.2 Organizational Consent Form B.3 Introductory Letter to Individual Participants B.4 Individual Consent Form C. Description of Stakeholder Experts 411 D. Sample Interview Guide 413 E. Site Interview Schedule and Participants by Category 416 F. Documentation Reviewed 417 G. Description of Information Systems Implemented 425 at Each Hospital H. Glossary 432 I. Literature Summary Based on DeLone & McLean's (1992) 435 Success Variables LIST O F T A B L E S Table 2.1 - Theoretical Perspectives of IT and Change 10 Table 3.1 - Choice of Research Strategy 45 Table 3.2 - Summary - IT Implementation in the Study Hospitals by Year 51 Table 3.3 - Numbers of Participants by Hospital and Group 52 Table 3.4 - Keywords Used to Search Transcripts With FolioVIEWS® 58 Table 9.1 - Summary of IT Across Hospitals 330 Table 9.2 - Summary of Impact by Group and Theme 332 ix LIST OF FIGURES Figure 2.1 - Technological Imperative and IT 15 Figure 2.2 - Organizational Imperative in Using IT 18 Figure 2.3 - Factors Influencing Use of Information Technology 19 Figure 2.4 - Variables Related to IT Effectiveness 20 Figure 2.5 - DeLone & McLean's Framework Applied to Health Care IT 23 Figure 2.6 - Socio-Technical Perspective of IT 28 Figure 2.7 - Donabedian's Approach to Quality 35 Figure 2.8 - Elements of Structure - Process - Outcome in Diagnosis 36 and Treatment Figure 2.9 - Theoretical Framework 41 Figure 2.10 - Analytic Framework 42 Figure 3.1 - Hospital Information System 48 Figure 4.1- Manual System for Ordering Lab Tests and Reporting Results 69 Figure 4.2 - Automated System for Ordering Lab Tests and Reporting Results .. 70 Figure 4.3 - Impact of PCIS on Laboratory Technologists at Hospital 1 79 Figure 4.4 - Relationship of Information Systems in Nursing 86 Figure 4.5 - Impact of PCIS on Nurses at Hospital 1 92 Figure 4.6 - Medication Order and Distribution Process 97 Figure 4.7 - Impact of PCIS on Pharmacists at Hospital 1 104 Figure 4.8 - Impact of PCIS on Physicians at Hospital 1 116 Figure 4.9 - Impact of PCIS on A l l Groups at Hospital 1 125 Figure 5.1 - Impact of PCIS on Laboratory Technologists at Hospital 2 138 Figure 5.2 - Impact of PCIS on Nurses at Hospital 2 147 Figure 5.3 - Impact of PCIS on Pharmacists at Hospital 2 155 Figure 5.4 - Impact of PCIS on Physicians at Hospital 2 163 Figure 5.5 - Impact of PCIS on A l l Groups at Hospital 2 172 Figure 6.1 - Automated Lab System : 178 Figure 6.2 - Impact of PCIS on Laboratory Technologists at Hospital 3 188 Figure 6.3 - Impact of PCIS on Nurses at Hospital 3 199 Figure 6.4 - Manual System of Processing Medication Orders 203 Figure 6.5 - Automated System for Processing Medication Orders 204 Figure 6.6 - Impact of PCIS on Pharmacists at Hospital 3 213 Figure 6.7 - Impact of PCIS on Physicians at Hospital 3 227 Figure 6.8 - Comparison of Three Order Entry/Results Reporting Systems .... 231 Figure 6.9 - Impact of PCIS on A l l Groups at Hospital 3 233 Figure 7.1 - Impact of PCIS on Laboratory Technologists at Hospital 4 247 Figure 7.2 - Flow of Information in OR Booking and Surgical Admissions 256 Program Figure 7.3 - Impact of PCIS on Nurses at Hospital 4 259 Figure 7.4 - Multiple Linkages Between Structure, Process and Outcome 262 Figure 7.5 - Impact of PCIS on Pharmacists at Hospital 4 271 Figure 7.6 - Comparison of Medication Order Entry Systems 273 Figure 7.7 - Impact of PCIS on Physicians at Hospital 4 281 Figure 7.8 - Impact of PCIS on A l l Groups at Hospital 4 286 Figure 8.1 - Program versus Hospital Departmental Focus 290 Figure 8.2 - Impact of PCIS on Laboratory Technologists at Hospital 5 297 Figure 8.3 - Impact of PCIS on Nurses at Hospital 5 307 Figure 8.4 - Impact of PCIS on Pharmacists at Hospital 5 317 Figure 8.5 - Impact of PCIS on Physicians at Hospital 5 322 Figure 8.6 - Impact of PCIS on A l l Groups at Hospital 5 325 Figure 9.1 - Efficiency and Productivity Changes 334 Figure 9.2 - Decision Tree for Proceduralizing Abnormal Lab Test Checks 337 Figure 9.3 - Structure - Process - Outcome Linkages Related to Pharmacy 349 Figure 9.4 - Use of PCIS in Documentation 350 Figure 10.1 - Revised Analytic Framework 372 A C K N O W L E D G M E N T S With the assistance of many people I was able to pursue a research project in an area that has long been of interest to me during my career in health care. As my thesis supervisor, Dr. Arminee Kazanjian's insightful questions and thoughtful probing for more information provided guidance in a non-directive way. Within our new Ph.D. Program Dr. Kazanjian bravely ventured into uncharted waters with her support for this project. My committee members also provided guidance in their own unique ways. Dr. Izak Benbasat's quick access to eclectic resources and his clear, concise thinking were invaluable in continuing to stay focused. Dr. Joseph Tan's questions always provided a new perspective to the problem. The project would not have been possible without the assistance of the five hospitals. I am indebted to the staff and many participants who were generous with their time and energy in arranging and participating in interviews, providing tours and retrieving archival documents. Their collective wisdom helped propel the study forward and I trust they will benefit from our increased understanding of this topic. I am very fortunate to enjoy the support of family, friends and colleagues who not only listened to details of the project, but offered their support and advice. I am particularly indebted to my immediate family for their long-standing patience and understanding as the homework seemed to "take forever." To my sons, Dustin (10) and Michael (7), for the many times they missed me on family outings and to my husband Marcel, for his continuing support and encouragement, I say a heart felt thank you for helping me reach my dream. We all look forward to returning to a "balanced" life that includes renewed energy and a sense of humor! Thank you all Impact of Information Technology on Health Care Professionals and Patient Care: A Multiple Case Study in Community Hospitals Chapter 1 - Introduction 1.1 Why is the Study of Impact Important? Throughout the world, health care is becoming more complex, and high costs precipitate calls for reform (Lorenzi, et a l , 1995; Vayda & Deber, 1994; Deber & Thompson, 1992, Evans, 1984). Health care institutions are under mounting pressure to create a cost effective system by controlling operating costs while maintaining quality of care and service (Anderson, et al., 1994). Information technology (IT)1 has the potential to increase timeliness, accuracy and accessibility of information. For this reason it has been expected to have a measurable impact on the provision of health services and to provide solutions for some of the major management problems in the health care system (Anderson, et al., 1994; Anderson & Jay, 1987; Mahajan, 1979). However, unlike other information-intensive industries such as airlines and banking, health care is characterized by slow adoption of IT (Melvin & McLoone, 1991; Kaplan, 1987). Despite the often disappointing or limited success of IT (Conklin, et al., 1988), investment continues to climb (Weill, 1992) with the service sector having little in productivity increases to show for its spending (Roach, 1988). Difficulties in realizing and measuring benefits occur for a variety of reasons related to the technology and context in which it is used. These include: 1 • limited availability of technology (Blum, 1989) as well as underutilization (Gardner, 1990); • antecedents, such as funding mechanisms, organizational structure and the accepted role of management (Stoelwinder & Abernathy, 1989); • the nature of providing health care which is both art and science (Caceres, 1984) and therefore the applicability of IT; • professional characteristics of the potential users of the technology. Professionals in the field are well educated, operate with a high degree of autonomy and have strong, ingrained, procedural traditions that serve as barriers to change (Minard, 1991b; Jay & Anderson, 1987); • differences in goals and values between both health policy makers and developers of these systems, and the intended users (Kaplan, 1987); • users and producers of information are often different groups. For example, clinicians are primarily interested in patient procedures, while managers are interested in the stability of patient costs and improving procedures to reduce those costs. However, managers cannot generate correct information for patient costs unless they have the data for patient procedures and clinicians may not benefit from providing this information (Baugh, et a l , 1995); and • changing characteristics of the organization in moving from hospital to community and the technology moving from mainframe to personal computers (Scherrer, 1988). Context is important in determining the success of any innovation and therefore measures of its impact (Rogers, 1995). Prescriptives for IT are often derived from small 2 samples in the private sector (Bretschneider, 1990). However, public and private organizations differ in many respects, such as the organization of work, personnel and financial management (Pickett & Hanlon, 1990). The public hospital environment is complex, with a variety of major stakeholder groups, including the board, administrators, physicians, nurses and other clinical or professional groups (Kim & Michelman, 1990; Friedman, 1985). Given their respective roles and tasks, these groups may also have different expectations about the nature and function of information systems (Ferrand, et al., 1993). Potential for conflict among the groups also arises out of the dual hospital structure with its medical and administrative hierarchies, the relative independence and practice of physicians and the role of information in maintaining autonomy of each professional group. In many cases hospitals anticipate better care and more efficient administration will result from their costly infusion of IT. To realize these benefits, they must also commit to major and continuous change. This includes investing energy to develop and sustain a new and flexible organizational climate that can focus on the relationship between information and outcome goals for patients. Other issues arise when contemporary, frequently parochial, information systems in Finance, Medical Records or the Laboratory become constituent components of integrated hospital information systems. Formerly they were stand-alone, or separate, systems under the control of their respective functional departments (Counte & Kjerluff, 1988). The information produced by this new generation of technology may be more powerful than the people who are expected to use them are prepared to handle (Barone & Chickadonz, 1991). 3 Paradoxically, while information technology is "revolutionizing how businesses operate" (Walton, 1989), this has not been the case in the health services sector as a result of individual, organizational and system wide reluctance to change. A well-known example is the patient record. It has essentially remained unchanged since the early part of the twentieth century (Bronzino, et al., 1990). This dormancy is in spite of more than thirty years of exploratory work and millions of dollars in research and implementation of computer systems in health care institutions2 (Institute of Medicine, 1991). To address this concern the Institute of Medicine (IOM) conducted a study on the health care environment, needs of those who use the patient record, technology and barriers to computer-based patient record (CPR) development. They concluded the CPR was an essential technology and recommended its widespread implementation within a decade. Although health care was in desperate need of CPR's, they noted technology was not the limiting factor and a concerted effort could make them a reality. This change has not been easily accomplished. At a midpoint in the decade for implementation predicted by IOM, Detmer & Steen (1995, p. 55) note: "Although a broader understanding of CPR's has been achieved and more leadership for CPR development exists today, substantial work remains to be accomplished." Incremental progress is evident in almost all areas identified in the earlier IOM report. However, they describe the current state of CPR development as still having "only pockets of excellence rather than full market saturation." Many hospitals have implemented systems related to various aspects of patient care, collectively called Patient Care Information Systems (PCIS), which form the basis for a computer-based patient record. 4 As IOM (1991) and others have suggested, successful management of technological change is dependent on humanistic as well as technological factors (Sankar, 1991; Zuboff, 1988; Mumford, 1981). Many researchers and practitioners in the field are beginning to realize that harnessing the technology will depend on understanding and managing the complex relationship between the technology, the organization and its stakeholders. Little is known about what these changes are, the definition of success and how to manage the technology and its users to produce the desired impact. An automated record may make it easier for health care professionals to access information. However, the effect it will have on the delivery of health care, on aspects of worklife3 for health care professionals (such as interaction between groups or decision making) and patient outcomes remains unclear. To date the study of impact has been limited to more quantitative measures of the effects intended by developers and implementors. This lacks an understanding of how the technology becomes integrated into the organization and into professional practice. 1.2 Objectives of the Study This study was undertaken to explore the type and degree of impact of Patient Care Information Systems (PCIS) on health care professionals in community hospitals. Defining impact from the users' perspectives as well as understanding how and why it occurs are important factors in the future success of information technology in health services generally, both from investment and implementation perspectives. This study set out to identify areas of impact in this context and the contributing factors, as well as to develop a framework to facilitate this and future research efforts. 5 The specific research objectives were to: • develop a theoretical framework for evaluating the impact of PCIS on health care professionals in community hospitals; • use this framework to describe how health care professionals in several organizations perceive the impact of using PCIS on their work; • identify differences in impact among professional groups and possible explanations for those differences. 1.3 Organization of the Thesis The remainder of this document is organized as follows. In Chapter 2 the literature review summarizes practice-based and empirical evidence of studies of impact. Three different approaches to IT and change in the organization provide the framework for discussion. Chapter 3 describes the research design, a detailed data collection protocol and strategies for analysis of the data. The findings for each of the hospitals are presented in Chapters 4 - 8 . Each chapter outlines findings for the individual professional groups within that site and ends with a summary section. Chapter 9 discusses the differences in impact across professional groups and hospitals. In Chapter 10, implications of the study for research and practice are discussed. A glossary in Appendix H provides the reader with definitions of terminology that are commonly used in the health care context. 6 Endnotes Information technology (IT) and information systems (IS) are frequently used interchangeably in the literature. In some cases they are used in distinctly different ways to represent the hardware and use of the technology, respectively. In an organizational context, use of the software and hardware on which it resides have no purpose one without the other. In this study, only IT will be used as a broad concept which includes: "both IS applications and the IT platform (the hardware, software and communications networks) used to enable the IS applications to function" (Lay & Ferrand, 1995, p. 5). 2 Classic examples of these systems which have been developed on-site at large teaching centres include COSTAR (Barnett, 1984), PROMIS (Fischer, et al., 1987) and HELP (Gardner & Lundsgaarde, 1994). 3 "Worklife" is used in the context of describing aspects of work such as decision-making and control (Kraemer and Danziger, 1990), and not in the sense of human resource planning where it is used to represent number of productive years of work. 7 Chapter 2 - Theoretical Framework 2.0 Introduction As discussed in Chapter 1, a prime objective in health care is improving or maintaining cost-effective patient health outcomes. There is a growing need for the health care industry to better understand the implications of investments in IT and their contribution to patient care. An evaluative strategy must go beyond the technical aspects of IT to understand how well the technology is actually functioning within the organization (Anderson, et al., 1994). This is important in forming a basis for developing specific interventions to enhance system success, as well as to determine strategies for future investment. In health care there has been a long history of high expectations for the role IT will play in change efforts aimed at controlling costs or maintaining quality of care. As the technology and organizations change, it is often unclear what that role is. This may be the case because underlying assumptions about how IT affects organizations imply quite different perspectives on what causes change to occur (Anderson, et al., 1994). The change perspective taken influences the nature of inquiry into "impacts," as well as interpretation of the results. Within a particular perspective, determining the impact of IT relates to three factors: the impact expected to occur, the impact that actually occurs and the reasons for the differences. Three different perspectives of the role of IT in organizational change will be reviewed in Chapter 2. Each of these perspectives is examined in detail in relation to 8 changing technology, organizations and expectations for impact of IT in health care. The chapter concludes with specific questions that direct the research study. 2.1 The Role of IT in Organizational Change The underlying assumptions about what causes change to occur provide the impetus for researchers to select study questions and practitioners to choose intervention strategies. Four different, yet similar, perspectives will be discussed. They each focus on factors thought to be responsible for change: the technology, the people in the organization, and a process involving both people and technology (summarized in Table 2.1). Each of these perspectives is briefly described, followed by a more detailed discussion in relation to IT, specifically in health care. In the earliest of these frameworks, Markus and Robey (1988) suggest three perspectives of IT and change: technological imperative, organizational imperative and emergent perspective. A technological imperative views IT as an external force that constrains or determines behavior of individuals and organizations. IT is responsible for effects such as changes in organizational structure, skill enhancement, deskilling of workers, or change in employment opportunities. The organizational imperative, however, assumes the user has unlimited choice over technological options and consequences. Choices and behaviors of managers and system designers contribute to an IT design that satisfies organizational needs for information. They manage impact of IT by attending to technical and social concerns. These are different from the emergent perspective, where uses and consequences of IT emerge unpredictably from complex social interactions. 9 Focus in Change Perspective Author (year) technology people process Markus & Robey (1988) -technological imperative - IT viewed as cause for organizational change -organizational imperative - motives and actions of designers of IT are cause of organizational change -emergent perspective - uses and consequences of IT emerge unpredictably from complex social interactions Kaplan (1991) -research, development and diffusion models -rational, orderly transition of knowledge -problem solving models - change agents collaborate with client in identifying needs and finding solutions -social-interaction models - stages through which individuals pass in decision to adopt and mechanisms of diffusion DeSanctis & Poole (1994) -decision-making (positivist) approach - characteristics of technology overcome human weaknesses and cause change -institutional (interpretive) approach -technology is flexible and organization directs change -socio-technical (combination of positivist and interpretive) approach - mutual influence of advanced technology and social processes shape each other Anderson, et al. (1994) -computer as external force -brings about change in behavior of individuals and organizations -IT is shaped by organizational needs and change occurs in a rational fashion through efforts of managers, developers and implementors who identify needs and solve problems -complex social interactions within the organization determine use and impact Table 2.1 - Theoretical Perspectives of IT and Change 10 Markus and Robey (1988) note that in this case it is unclear whether interventions are required because it is impossible to predict the indeterminate outcomes. Researchers may advocate 'emancipatory' strategies, such as extensive user participation in the analysis, design and implementation of IT. A central tenet of this perspective is "the social meaning ascribed to IT. This perspective accounts for conflicting research findings about impacts by demonstrating the different meanings that the same technology acquires in different social settings" (Markus and Robey, 1988, p. 595). In developing the second framework, Kaplan (1991) suggests that evaluation research often does not specify underlying assumptions about models of change that may influence both the study questions and accompanying research strategies. Awareness of these models may also enrich information systems research by identifying further areas for study and research strategies for studying changes inherent in the development and use of information systems. Based on early work by Havelock, et al. (1971; cited in Kaplan, 1991), Kaplan introduces three types of change models. In "Research, Development and Diffusion Models" rational, orderly transition of knowledge occurs from research to development, diffusion and adoption. These models focus on the researchers, developers and disseminators. The recipient of the new product is treated as essentially passive and Kaplan suggests many researchers in information systems hold these models. In "Problem Solving Models," change agents diagnose client needs in collaboration with the client. Change occurs in stages where they identify needs, seek solutions and apply them. Lewin's stage theory (with unfreezing, moving, freezing) and force field analysis (Havelock, et al., 1971; cited in Kaplan, 1991) are examples. "Social-Interaction Models" emphasize the diffusion aspect of change. They 11 focus on stages that individuals go through as they decide whether or not to adopt an innovation and the mechanisms of diffusion through the adopting group. Kaplan identifies Rogers' (1983) classic "Diffusion of Innovations" theory as a well-known representative of this model, where knowledge flows back and forth within complex networks and relationships. In the third framework DeSanctis and Poole (1994) examine the role of IT and change with respect to advanced technology. These offer new opportunities for change involving group, rather than individual, use. They describe two existing schools of thought that underlie the study of IT and organizational change and suggest a third new perspective for advanced technologies, particularly Group Decision Support Systems (GDSS). The Decision-Making school of thought "adopts a psychological approach to the study of technology and change," that "emphasizes cognitive processes associated with rational decision-making." Technology has structures (data, decision models) designed to overcome human weaknesses (bounded rationality, process losses). Once applied, technology is expected to bring productivity, efficiency and satisfaction to individuals and organizations. Failure to achieve the desired change reflects failure in technology, its implementation or its delivery to the organization. DeSanctis and Poole indicate that while this perspective has yielded extensive literature on GDSS's and other advanced technologies, it has not produced consensus on how these systems should be designed or how they affect the people or organizations who use them. The second perspective described by DeSanctis and Poole (1994) is the Institutional School, which views technology as an opportunity for change rather than as a causal agent. 12 This perspective focuses less on structures within the technology and more on evolution of social practices within institutions. It considers technology to be interpretively flexible. "Analysis is the process of looking beneath the obvious surface of technology's role in the organizational change to uncover the layers of meaning brought to technology by social systems" (p. 125). DeSanctis and Poole (1994) synthesize these two schools of thought into a third new perspective, Social Technology. This is a more complete view and accounts for the "power of social practices without ignoring the potency of advanced technologies for shaping interaction and thus bringing about organizational change. Technology has structures in its own right, but social practices moderate their effects on behavior." They propose an Adaptive Structuration Theory that explains the dynamic way technology and social structures shape each other over time. Anderson, et al. (1994) describe a fourth framework in a recent book that outlines a range of methods for evaluating health care information systems. Three models of change prevalent in information systems research (based on Markus and Robey, 1988; Kaplan, 1991) are suggested. In their first model they view the computer system as an external force that brings about change in the behavior of individuals and organizational units. Controlled studies in a laboratory setting exemplify this perspective. These studies focus on users' response to the technology, not on how the systems fit into daily work of the organization where they will be used (Benbasat, 1989). In the organizational setting, evaluation focuses on technical performance aspects such as cost, speed and accuracy. Organizational and technological characteristics are assumed constant. In their second model, the information 13 needs of managers and clinicians, as system users, determine system design. They consider IT external to the organization and shaped by organizational needs. Change occurs in a rational fashion through efforts of managers, developers and implementors who identify needs and solve problems. They assume control over the technical aspects of the system and the consequences of its implementation. In Anderson's third model, complex social interactions within the organization are determinants of use and impacts of IT. Organizational change occurs over time and includes dynamic social and political processes as well as characteristics of the individual and organization. While the perspectives in each of these four papers are slightly different, they do present three similar foci in the role of IT and change: technology controls change, people control change through controlling technology, technology and people are interactive and shape each other. Each of these perspectives will be examined with respect to IT in health care. 2.2 Technological Imperative in IT and Change The technological perspective assumes characteristics of the IT will create change. For example, tasks are automated so they can be completed faster and more accurately. The organization applies technology to the task and determines its impact by the number of tasks completed, how fast and at what cost, as illustrated in Figure 2.1. Hospitals first acquired computer systems during an era when they based payment for care on cost reimbursement. Two critical issues were accounting for costs and sending bills out quickly. For this reason financial applications were among the first to be implemented. 14 Manual System Apply Technology Impact *how many? *how fast? computer task r *accuracy? task task *cost? J Figure 2.1 - Technological Imperative and IT In the 1970's and 1980's individual hospital departments began to develop stand-alone systems to meet their own specific needs and generally addressed automation of clerical tasks. In areas with high transaction processing, such as Laboratory and Pharmacy, they expected IT to improve operational efficiency and lower costs. The prime focus was on identifying sufficient benefits to justify the investment, either internally to the organization or to external funding bodies such as the Ministry of Health. From the advent of early systems, organizations assume that automating processes and increasing efficiency, decreases costs. Coffey's (1980) classic description of the assessment of economic impact in one of the first instalments of a hospital information system (the Technicon Medical Information System at El Camino Hospital in Mountain View, California) illustrates this. His study results indicate that the system improved productivity in the medical care departments and caused an overall reduction in patient length-of-stay. However, from a total hospital cost perspective, the results were not definitive because they assumed that increased support department costs were not directly caused by the system.1 IT is usually expected to achieve certain operational efficiencies and personnel savings, and therefore high costs often lead to unrealistically high expectations. Aside from 15 simply replacing manual operations, a number of other issues limit the effectiveness of computers in public organizations: data quality and accessibility, system evolution and interface difficulties, and organizational and political concerns (Tien and McClure, 1986). Studies and reports often focus on indicators of technical performance, such as cost, speed and accuracy. However, despite rapid diffusion of IT into the community hospital sector, there is still limited evidence whether adoption makes financial sense. Glandon and Shapiro (1988) identify three areas of concern in the evaluation IT in health care that are very similar to why impact of innovations is not often evaluated (Rogers, 1995): 1. despite the multibillion dollar investments in IT and annual operations, comprehensive, well-documented, methodologically sound evaluations are virtually non-existent; 2. despite the wide variety of systems evaluated, the numerous study settings, and the diverse methods, most studies conclude that IT was beneficial; 3. barriers to evaluation include an incentive to acquire technology that provides any potential improvements in clinical or administrative practice; methodological and resource constraints to conducting evaluations; and a "water under the bridge" phenomenon. Once the IT is in place, the cost of evaluation usually overwhelms benefits for the individual hospital. Early efforts at evaluating impact of IT focused almost exclusively on cost factors. However, determining the effect of integrated PCIS on hospital operations is more difficult because no single concept or set of concepts in the information or social sciences define the issues involved (Melvin and McLoone, 1991). Review of the literature suggests a diversity of views about the purpose of IT, its organizational implications, costs and benefits. 16 Although outcomes are often described in dollar terms, it is difficult to demonstrate how claimed savings were accomplished and to isolate the role of IT in achieving those savings, particularly in areas such as the patient's length of hospital stay. Assuming that characteristics of IT induce change, organizations expect automating manual tasks to be of financial benefit through improved efficiency and productivity. This reflects the thinking behind early stand-alone systems in health care that focused on automating financial operations and transaction processing. Impact could be determined through counting the number of transactions completed, or measuring the length of time to complete each transaction. As systems became more sophisticated and powerful, a broader perspective on the role that IT played in change emerged. Users became more knowledgeable and began to manage the technology to suit their needs. 2.3 Organizational Imperative in IT and Change An organizational imperative perspective assumes that characteristics of the IT can be shaped to solve problems identified in the organization. They can determine the impact by how well the technology matches organizational need by whether people use the IT, as Figure 2.2 illustrates. This view was prevalent as users became more involved in system selection, development and implementation. The early 1980's saw PCIS providing the first centralized database of patient information by building on a Central Patient Index. This was useful in later applications that crossed departmental boundaries, such as patient location (admission/discharge/transfer systems) and order communication to ancillary departments with 17 results automatically reported back to the Nursing Unit (Worthley and DiSalvio, 1989). As IT became more complex, managers began to try to control some of the factors they thought influenced success. Impact Organizational f Needs/Tasks \ * system used? 1 \ *needs met? identify needs needs met? 1 A Shape J Technology to Fit Needs Figure 2.2 - Organizational Imperative in Using IT However, IT must be utilized for it to be valuable in problem-solving and there has been a lengthy interest in why some innovations succeed and others do not. This has lead to extensive investigations of factors that affect adoption of particular innovations and their diffusion in organizations and societies (Rogers, 1983; 1995). Characteristics of the organization, individual adopter and the innovation itself (in this case, IT) play a role in this process2 as Figure 2.3 illustrates. Identifying factors important for successful adoption and diffusion of IT assists managers and developers to ensure success of the IT. Researchers and practitioners recognize that IT must be used successfully to have an impact. A continuing focus of IT research is the link between the nature of the IT introduced into organizations and its impact on users' work performance. The literature on 18 organizational effectiveness suggests that defining and measuring IT effectiveness via user perceptions is both appropriate and practical (Miller, 1989). This perspective produces a shift away from using characteristics of the individual, organization or technology as predictors of use, toward factors such as satisfaction, perceptions of usefulness and ease of use. These variables have been studied extensively and Figure 2.4 illustrates one conceptualization of their relationship. Organization - culhff e/size/complexiry -urban versus rural -top rnanagement support -stakeholder "champion" -training User Adoption -age and Use of -experience >• Information -attitudes Technology -effect on roles Information Technology M -complexity -ease of use -accessibility Figure 2.3 - Factors Influencing Use of Information Technology 19 Task Characteristics Technology Characteristics Beliefs Attitudes User Information Satisfaction Perceived Usefulness Ease of Use Task/Technology Fit (IT Effectiveness) IT Utilization Performance Impacts Figure 2.4 - Variables Related to IT Effectiveness (modified from Goodhue and Thompson, 1995) Early studies by Raymond (1985) and Srinivasan (1985) investigated relationships between User Information Satisfaction (as a measure of perceived effectiveness) and use that they expected to produce Management Information System success. Goodhue (1988, 1990), Miller (1989) and others felt User Information Satisfaction was too imprecise a measure. They directed their attention to the fit between technology and task, its influence on attitudes and beliefs about using IT, and ultimately through that, performance. Davis (1989) and Davis, et al. (1989) developed a Technology Acceptance Model that focused attention specifically on one aspect of the task-technology fit. They were attempting to explain and predict user acceptance through causal linkages between two key beliefs: 20 perceived usefulness and perceived ease of use. The question of "fit" was also approached from a prospective, rather than retrospective, viewpoint in a number of other studies. They suggested one way to ensure fit was through user participation and involvement in the development process (Barki and Hartwick, 1994; Newman and Noble, 1990). Understanding the gap between actual and expected system characteristics is another predictor of satisfaction with the fit between technology and task (Shirani, et al., 1994). From a review of 180 studies of Management Information System's success, DeLone and McLean (1992) develop a framework that represents an integrated view of this diverse research. Their taxonomy reflects earlier work in communication theory and the role of information in changing behavior. It has six different "dimensions of success" including System Quality, Information Quality, System Use, User Satisfaction, Impact on Individual User and Impact on the Organization. This is a useful tool to organize the multitude of studies in order to gain a sense of the breadth and history of this research. A large proportion of practice literature is based on single case studies and relates IT success to characteristics of users, technology or the organization. Many writers described the experience of their health care facilities with selection, development and implementation of PCIS during the 1980's (for example, Woodend and Cluett, 1992; Kennedy, 1987; Komes, 1987; Hebert, 1985). Limited empirical evaluation and published research on advanced PCIS is available. The difficulty in using the available studies is they come from academic medical centers that have programs in medical informatics, where systems are, for the most part, self-developed (Metzger, 1995). More research is needed on the transferability of successful 21 systems in these sites, to practice environments in community hospitals and physicians' offices. The diffuse nature of IT use in health care, coupled with the lack of common nomenclature or theoretical framework, makes it difficult to consolidate the history of the research and publications in this area. As Figure 2.5 illustrates, the body of IT literature in health care can generally be mapped onto DeLone and McLean's (1992) six dimensions. (See Appendix I for a detailed review of articles.) For many of the descriptive articles, use of the system seems to be the implied goal. This is their ultimate measure of success, with much of the selection, training and implementation activities directed toward this end. What is obviously different between the health care literature and that reviewed by DeLone and McLean is the effect IT is expected to have in supporting better patient outcomes at the same or lower cost. 2 2 -System Selection -System Development* System Quality Information Quality -User Training -User Involvement -Implementation -Usefulness System Use User Satisfaction -Attitudes -Beliefs -Expectations -Productivity -Patient Care Benefits -Decision-M aking -Return on Investment -Reduced Personnel -Organizational < Process Efficiencies Impact on Individual User Impact on Organization **Legend** DeLone & Related McLean's Six concepts of Dimensions of IT success in. IT Success health care Figure 2.5 - DeLone and McLean's (1992) Framework Applied to Health Care IT DeLone and McLean's (1992) six dimensions of IT success and their application to the health care literature are summarized below. An additional category, impact on patient outcomes, has elements of individual and organizational impacts and has been included to reflect this additional interest in health care: 23 1. System Quality - direct access to better quality information, reliability, available wherever decisions are made about care, easy to enter and retrieve 2. Information Quality - more complete, timely, accurate information; quick, value added access to information 3. Use - easy to learn and use. Training and attitude are expected to influence use. In integrated systems that cross departmental and program boundaries, use is generally mandatory because the automated systems replace manual ones. In this case, counting the number of times a system is utilized is not particularly valuable information. The relationship between use and impact has been tested empirically in narrowly focused studies. For example, a randomized control trial was used to assess resource use associated with physician inpatient order writing on microcomputer workstations (Tierney, et al., 1993). The results demonstrated lower patient charges and hospital costs could be achieved but the system required more physician time than the paper charts did. 4 . User Satisfaction - sometimes identified in relation to other "success" variables such as information or system quality, in other words, satisfaction with timeliness of information or accessibility of terminals. Drazen (1995) identifies three common measures of computer acceptance: user attitudes, system use, and user satisfaction. She notes user attitude was the most common measure in early studies and was predictive of diffusion of computers in the health care environment. Since it seems clear that computers will be used in health care settings, assessing general attitudes toward computers is no longer seen as a high priority (Drazen, 1995). However there may be new uses for detailed attitude surveys, such as planning for optimal 24 implementation of new computer systems (Hebert and Benbasat, 1994; Lundsgaarde, et al., 1989). Once computers have been introduced into health care delivery, the most relevant question becomes, "Are these systems useful?" and two common measures of usefulness are voluntary use of the computer and user satisfaction. In situations where use of the system is mandatory Drazen suggests satisfaction is the only way to determine acceptance, although others have developed direct measures of usefulness (Davis, 1989; Moore and Benbasat, 1991). 5. Individual Impact - improved efficiency, more time spent with patients. Individual impact of IT is a function of the people who convert data into the information they require to deal with the complexity of work and decisions for which they are responsible. For example, there are many suggestions for implementing clinical alerts and practice guidelines. However, unless practice behavior changes as a result of the additional information, the IT cannot be considered valuable. This critical link between using the technology and using the information to change behavior has been missing from many early efforts to design successful IT. A recent study demonstrated effective use of clinical workstations in changing physicians' responses to alerts regarding primary care interventions, in reducing admissions (Safran, et al., 1995). Others note that physicians will not be willing to change their minds about using computers in clinical practice until there is evidence of better quality, more time for patient care and more time for themselves (Lunsdon, 1993). These do not necessarily relate to changes in practice on their part. 6. Organizational Impact - cost reduction, profits, return on investment, productivity gains, improvements in process efficiencies. 25 7. Patient Outcome - reduced length of hospital stay, reduced medication errors and other "incidents," reduced re-admissions, increased quality of care (as a result of health care professionals having better access to information), patient satisfaction, conformity of patterns of practice to clinical guidelines. Even though the patient is an important stakeholder, few studies directly examine the impact of IT on clinical practice and patient outcomes due to the time lapse between the intervention and resulting health benefits (Rogers, et al., 1982; Blaschke, 1990). In a number of randomized control studies, computerized record summaries have been associated with improved patient outcomes (Rogers, et al., 1982; Rogers and Haring, 1979). The studies assume timely, accurate reports contribute to the goal of "improved quality of care." A comparison of two interpretations of "timeliness of reports" illustrates the difficulty with these measures of impact (Kropf, 1990). In one facility the Radiology Department standard was a twenty-four hour turn-around for reporting results and transmitting the image back to the ordering physician. The physicians however, often did not return to see their patients again within twenty-four hours, so did not find this feature particularly useful. In a second facility only positive results were telephoned to physicians within twenty-four hours and hard copies of all results sent via mail. Therefore, the increased investment in IT to produce more timely information did not effect the action taken by the physicians. By far the majority of research and practice literature in health care IT assumes an organizational imperative perspective. As mapping to DeLone and McLean's dimensions of success illustrates, organizations expect that attention to system selection and development, user participation and training results in successful system implementation. Research based on this perspective searches for causal relationships between IT, as the independent variable, 26 and organizational change. Survey research is common with resulting "prescriptions" for successful implementation and use of IT. There is an increasing realization that introduction of IT is more than just another technology because information "plays a distinctively social, interpersonal role in organizations" (DeSanctis and Poole, 1994). IT also modifies the environment it was intended to support and changes the way individuals carry out their tasks within the organization. These changes may invalidate some of the earlier assumptions about the role of IT in change (Blum and Orthner, 1989) and therefore requires a new perspective for investigation of its impact. 2.4 Socio-Technical Perspective in IT and Change Information technologies are changing. DeSanctis and Poole (1994, p.l) note that: "The past decade has brought advanced IT such as electronic messaging systems, executive information systems, collaborative systems, group decision support systems and other technologies, that enable multi-party participation in organizational activities through sophisticated information management." The impact of many new information technologies may only become evident as they are integrated into organizations, where their effects are less a function of the technologies themselves than of how people use them. For this reason, actual behavior in the context of advanced technologies frequently differs from the "intended" impacts. Assuming a socio-technical perspective for IT and change moves investigations of impact away from single measures of productivity or usefulness, to the relationship between the users and use of the technology and how they change over time. Leavitt (1965) proposed 27 four factors that were important in organizational change. His paradigm, illustrated in Figure 2.6, is a useful representation of the dynamic relationship found in a socio-technical perspective of IT and change. Organization Technology Task Individual Figure 2.6 - Socio-Technical Perspective of IT (Leavitt, 1965, p. 1145) The current changes in technology are accompanied by new directions in health care delivery. Moving toward an integrated, community-based health care system simultaneously increases the challenges for information and access to that information. It is impossible to provide seamless access to care without also providing seamless access to information, because patients cannot be directed to the most appropriate location of care unless relevant information about the patient is available there (Metzger, 1995). Opportunities for spending continue to increase as resources decline and the contribution of each expenditure to improved patient outcomes must be evaluated. "Opportunity costs" will be determined on a community-wide basis, rather than by individual institutions. These pressures are major forces behind the current interest in patient-care information systems and the development of a computer-based patient record, with direct clinician entry. As the IOM pointed out five years ago, the major limitations for moving in this direction are not in the technology. 28 In the 1990's the role of automated Health Information Systems is "generally believed to have been expanded to provide decision-support information to manage patient care more effectively both from a quality and cost perspective" (Huesing, 1992, p. 167). However, Huesing suggests these changes are more a result of economics related to capping hospital-based acute care funding, the care demanded by a more educated consumer, and the cost advances in medical technology. No economic relief is expected through political means. The only viable alternative is to change the behaviour of those who initiate care (and induce the costs) by providing information to the caregiver at the time of treatment decisions (such as can be done through a CPR), rather than retrospectively (as is the case with the paper chart). Integrated information systems also facilitate the sharing of information across departmental boundaries that can be used to produce patient- and provider-specific costing information.3 This information provides opportunities for closer scrutiny of resource utilization and variations in practice patterns. As well, it introduces dilemmas for health care practitioners, some who argue that their role is not to resolve conflicts between controlling costs and providing services. The impact of computer-based information systems is related to the integration and mutual adaptation of the technological aspects of IT as well as the social aspects of organizations (Walton, 1989; Zuboff, 1988; Iacovou, et al., 1995). In health services this may be pervasive because IT has the potential to affect both social organization and delivery of medical care through the restructuring of work tasks (Aydin, 1994). This imposes new methods and routines on the performance of work (Aydin, 1989) as well as changes in organizational procedures and responsibilities (Gerdin-Jelger and Peterson, 1985; Peterson, 29 1985) creating shifts in power, status and possibly conflict. The delivery of health care requires coordination and cooperation between numerous different occupations and departments. Changes in how these groups perform their work and interact with one another can have important consequences for the organization as a whole. The impact of IT is perceived differently by different professional groups (Kjerulff and Counte, 1988; Fischer, et al., 1987) or different specialties within a single department (Kaplan and Duchon, 1988). Reactions to the same system can range from increased job satisfaction to dislike of the system because it interferes with the job. Research on the use of IT indicates that physicians and nurses do not resist the concept of computers. However, very few studies investigate the type of computer support needed, functions nurses and physicians find most beneficial to automate, and where current systems fall short (Drazen, 1995). For example, one empirical study of physicians and nurses using order entry in a hospital demonstrates their different perceptions of benefits in the same system: physicians valued off-floor accessibility the most and nurses valued legibility and accuracy of orders (Lee, et al., 1996). The interaction of users, task and technology within a certain environment produces social impacts of computing on work, particularly for knowledge workers. This is illustrated in Kaplan's (1995b) study of three clinical systems where she investigated individuals' understanding of their work, as well as the relationship between an information system and work it is intended to support. It is this interaction that ultimately will have the greatest impact on worklife for health care professionals. Kraemer and Danziger (1990) identify six dimensions of worklife that IT use may impact: 30 1. decision making - the capacity to formulate alternatives, estimate effects, and make choices; 2. control - the power relations between different actors; 3. productivity - the ratio of inputs to outputs in the production of goods and services; 4. social interaction - the frequency and quality of interpersonal relationships among co-workers; 5. job enhancement - the skill variety and job domain; and 6. work environment - the affective and evaluative orientations of the worker toward the setting of work. Given the conditions of professional interdependence and a complex environment in health care, a socio-technical perspective is useful in understanding the impact of new, integrated information systems. The computer-based patient record (CPR) is an example of a complex, integrated set of systems which health care organizations currently aspire to and illustrates the appropriateness of this new perspective. A CPR is built around a common patient database and draws information from many areas, including: • clinical systems in the Laboratory, Radiology, and Pharmacy; • transaction processing systems to order tests or drugs and receive results; • documentation systems to record clinical notes, patient observations and treatment interventions; and • communication systems that include remote access capabilities. Anderson, et al. (1995, p. 767) point out these complex systems cross traditional boundaries and have important differences from previous systems: 31 While most of the technological barriers to the development of CPR's have been overcome, patient records are also social systems that use information technology. The implementation of such systems does more than enhance our ability to deliver health care. It also affects practice patterns and professional relations among individuals and groups within the organization. The ultimate success of any system depends upon integrating it into a complex organizational environment and ensuring that it is used effectively by the individuals for whom it was designed. Although the literature suggests computerized records are accepted by physicians, they improve access to information, facilitate patient management and research, and provide educational opportunities,4 it remains unclear what the consequences for other user groups and patients are. Rogers (1983; 1995) points out that most research on the consequences of adopting innovations has a distinct "pro-innovation" bias. Innovations can cause both desirable and undesirable consequences, many of which cannot be separated. An innovation may be more functional for some individuals than for others. Therefore, certain positive consequences may occur for some members of a group at the expense of others, which is likely the case for advanced IT. For example, physician order entry is considered by some as essential to the success of the CPR. This has lead to prescriptive articles and empirical studies related to how this could be facilitated. Kaplan (1994) indicates that physicians have long been ambivalent with respect to the use of CPR's and particularly direct order entry by physicians. She discusses implementation strategies for enhancing benefits and reducing barriers or disincentives to physician order entry. A survey to evaluate users' overall satisfaction with physician order entry also identifies factors associated with satisfaction and dissatisfaction, users' perceptions about frequency of specific features used and usefulness of those features (Lee, et al., 1996). As Anderson, et al. (1995) point out, organizations are social systems that are affected by the introduction of IT. However, little has been written about the effect CPR's will have on 3 2 the organization, interactions between health care professional groups or these professionals and their patients. For example, one report of physician order entry that contributes to efficient drug distribution, benefits both pharmacists and patients (Hubbell, 1994). However, there is no indication how patients perceived the reduced time they spent with the pharmacist. In another setting, nurses experienced changes in their perceived role as integrators of patient information as a result of CPR's (Lorenzi and Riley, 1995). They felt their role in the overall care process had been diminished. In the context of changing IT and health care delivery, the effect of IT on patient outcome is becoming increasingly more important. Unlike DeLone and McLean's (1992) success factors that include individual and organizational impact, the interdependent roles of health care professionals in achieving "quality care" are a more salient focus for this study of impact in health care. Donabedian's (1965, 1980, 1982, 1988) well-known model for assessing quality of patient care provides a framework for evaluating the impact of IT in this way (Hebert, 1995). Donabedian describes three attributes of care in his model: structure, process and outcome. He suggests the most direct way to evaluate the quality of care is to examine the process of that care. Two less direct approaches to use are the assessment of structure and outcome. A more detailed discussion of these concepts and their relationship to one another provides the context for their use as indicators of IT impact in health care. The assessment of "structure" is one indirect approach to evaluating quality of patient care. In this case, the term "structure" is used in a way that differs from other disciplines, such as organizational behavior. Donabedian (1980, p. 81) defines structure to mean "the relatively stable characteristics of the providers of care, of the tools and resources they have 33 at their disposal, and of the physical and organizational settings in which they work." This concept of structure includes the human, physical and financial resources that are necessary to provide medical care. Using structure as an indirect measure of the quality of care depends on the nature of its influence on care. Donabedian suggests structure is relevant to quality in that it increases or decreases the probability of good performance. The usefulness of structure as an indicator of the quality of care may be limited by insufficient knowledge about the relationships between structure and performance. The "process of care" is a set of activities that goes on between practitioners and patients. The quality of that process may be determined by either direct observation or reviewing documentation of the process. Donabedian (1980. p. 79) notes: While "process" is the primary object of assessment, the "basis" for the judgment of quality is what is known about the relationship between the characteristics of the medical care process and their consequences to the health and welfare of individuals and society. Standards for quality of the process of care are normative.. They are derived from the science of medicine as well as the ethics and values of society. The study of "outcomes" is another indirect approach to assessing the quality of care. Donabedian (1980, p.83) uses outcome to mean "a change in a patient's current and future health status that can be attributed to antecedent health care." From his broad definition of health, improvements may be found in the psychological, physical and physiological aspects of patient performance. In an extension of this definition, he includes patient attitudes (including satisfaction), health-related knowledge acquired by the patient, and health-related behavioral change that may be components of current health or contributions to future health. 34 Figure 2.7 illustrates the fundamental relationship among the three elements that is important in understanding this three-fold approach to quality assessment. This relationship suggests the structural characteristics of the settings where care takes place influence the process of care. Similarly, changes in the process of care influence the effect of health care on health status, or outcome. Structure Process Outcome Figure 2.7 - Donabedian's Approach to Quality (Donabedian, 1992) Donabedian (1980) notes that although his formulation of these elements is based on basic relationships of theoretical and operational significance, many ambiguities remain when one tries to classify specific phenomena exclusively under one of the three headings. This is because "the three-part division is a somewhat arbitrary abstraction from what is, in reality, a succession of less clearly differentiated, but causally related, elements in a chain that probably has many branches. In such a chain, each element is, at least to some extent, a cause of the element that follows, while it is itself caused by the elements that precede it" (Donabedian, 1980, p. 84). The "structure-process-outcome" paradigm is a highly simplified presentation of a complex reality, as Figure 2.8 illustrates. There are chains of events and even interrelated chains, so as Donabedian (1992, p. 357) points out, it is to some degree arbitrary where one stops and says, "This is an outcome." Sometimes it is not easy to say, "Here structure ends and process begins or process ends and 35 outcome begins." He suggests asking the key question, "In what way have individuals or groups become different?" to help reduce the ambiguity. Structure D C Characteristics of the diagnostic laboratory Characteristics of the physician Process Tests performed by the laboratory Tests ordered by the physician Results of tests interpreted by physician Outcome Results of the tests Diagnosis: the illness and its characteristics Treatment chosen and executed by physician, other personnel, and patient Change in patient's health Figure 2.8 - Elements of Structure - Process - Outcome in Diagnosis and Treatment (Donabedian, 1992, p. 358) 36 Overall, IT is expected to have an impact on the quality of care, but as discussed earlier in this chapter, it is often implemented with disappointing results. Perhaps not surprisingly, a literature review by van der Loo, et al. (1995) revealed that most of the evaluation studies of automated information systems dealt with structure measures and only 15 of 91 studies investigated effects on the outcome of care process. They describe the effect of IT on each of Donabedian's approaches to quality: a) Structure is the relatively stable characteristics of the settings in which care occurs. The effect of IT on structure may be measured by job satisfaction, user satisfaction, IT performance (that is, previous x-ray films are available), time consumption for personnel and processes. b) Process is the set of activities that go on within and between health care practitioners and patients (for example, what is actually done in giving and receiving care). Effect of an IT on process may be measured through changes in the user's performance (that is, does the system change the physician's diagnosis?), change in volume of services ordered, number of times the IT has been consulted (in other words, database use). c) Outcome is the effects of care on the health status of patients and populations, including behavioural changes and satisfaction with care. Effect of an IT on outcome may be measured by whether use of the system has an effect on mortality, the patient's satisfaction with the accessibility of care and the waiting time of patients. From a socio-technical perspective the evolution of technology and environment produce multi-dimensional impacts. These may occur sequentially or simultaneously because each impact may or may not be a precursor to the other. In Grusec's (1986) evaluation of the 3 7 adoption and diffusion of Office Automation (OA) in government offices, he describes three levels of impact that emerge over several years: 1. Direct substitution - a new way of performing procedures replaces ones previously done in some other way. Traditional efficiency concepts and measures are applicable because the end "products" are essentially the same whether technology is used or not. Most people can easily see substitutive uses in their own work setting. 2. Proceduralization - events previously performed in non-procedural (rule-governed) or semi-procedural ways are transformed into more procedural ones through use of the computer. Almost immediately the "products" may not be equivalent to their manual counterpart and cannot be compared by counting or other simple measures. Simple cost justification models no longer apply. Users have difficulty envisioning proceduralization without hands-on experience in the work setting. It requires examining tasks that require human judgement and action, and recognizing which things may be matched to computer capability. 3. New capabilities - new capabilities may be radically different from the old or may be something that was entirely possible to do before, but was not done because of the inconvenience or high effort required. There are no easily comparable before and after events that make them much harder to see in advance. Some of the new capabilities imply ways of work or goals that were rejected long ago, i f ever contemplated, because they were impossible or extremely difficult to do. In the absence of direct, prolonged and active experience with a computer system, enormous mental, creative effort is needed to try to envisage new capabilities. Rogers' (1995) identifies three levels of technology transfer (replication, innovation and transformation), which are very similar to Grusec's (1986). In contrast, Austin's (1988) 38 suggests levels of impact that reflect three management visions for IT which are roughly equivalent to the levels described by Zuboff (1988): 1. automate - existing manual tasks and procedures; 2. informate - expand and distribute knowledge created by IT through out the organization in order to change behavior; 3. transform - IT used to change the way the organization works and competes (e.g. order entry from physician's office). Austin's (1988) view of IT performance differs from either Grusec's (1986) or Rogers' (1995) because impact depends on management expectations. This precludes the unpredictable effects of integrated systems, as well as the diffuse nature of IT impact used across occupational groups and organizational divisions. Over time it also limits the movement of users from one "vision" to another. For example, an order entry system initially used to automate the ordering of lab tests, may later be used to determine what tests are ordered and to develop individual physician order sets. Researchers in a number of disciplines are investigating the interactive relationship created between the innovation, environment and individual user. These relationships change over time (Walton, 1989), fundamentally changing the organization and what people do (Zuboff, 1988). Unlike earlier innovations, empirical evidence suggests adoption of IT can move beyond diffusion and create a situation of organizational transformation (Orlikowski, 1995). 39 2.5 Analytic Framework The study of IT impact in health care generally has assumed a technological or organizational imperative perspective. The technology developed from early financial systems, to stand-alone departmental systems, and then to integrated systems sharing a common patient database (such as a PCIS) that later became technically feasible and organizationally desirable. From these two perspectives the effects of IT on individual users and the organization are thought to be either a result of the technology itself or its application to a problem by managers, developers and implementors. Topics of interest have included training, implementation, productivity, efficiency and decision-making. A final goal in the evolution of these more sophisticated systems is the CPR, which includes the documentation of patient care. Developers and users of these new sophisticated systems hold high hopes for their potential to change traditional organizational design, intelligence, and decision-making for the better. However, DeSanctis and Poole (1994) point out that a number of questions remain unanswered, such as: "What changes do these systems actually bring to the workplace? What technology impacts should we anticipate, and how can we interpret the changes that we observe?" The impact of complex systems that are integrated across the organization, can no longer be "controlled" through system selection and training. The technology and its use shape both the IT and organizational change. Apart from the direct effect on operations within one's own department, there is very little known about the relationship between changes in one professional group's work and other groups, or patient outcome. Aside from directly measuring quantitative changes such as increased productivity, no theoretical 40 frameworks have developed to provide the basis for determining a broader concept of impact. The investigation of IT impact in health care can benefit from a socio-technical approach that takes into account the interactive nature of development among users, IT and the organization. The impact of IT on patient outcomes can be conceptualized through Donabedian's (1988) model of the measurement of quality patient care. To accommodate the expected administrative and cost component of care, this model is modified to include output as a measure of cost effectiveness (Hebert, 1995). Use of a socio-technical perspective also emphasizes that impact is not a static concept, but rather one that develops over time as the user, technology and context change. As discussed in the previous section, Grusec's (1988) model of the three levels of impact reflects this dynamism. To guide the analysis of this study, the combination of these two concepts in a theoretical framework reflects the temporal changes of IT contributions to quality patient care (Figure 2.9). Information Quality Use of the Information System T3. Impact: Impact: Impact: structure I process / \ outcome output ktput htput Level 3 Level 2 Level 1 User Satisfaction Figure 2.9 - Theoretical Framework (modified from Hebert, 1995, p. 171) 41 There is ample evidence that DeLone and McLean's success factors are relevant to the study of IT in health care. Very little is known about impact, particularly from a socio-technical perspective. This research study sets out to explore the perceptions of impact of a complex information system held by different health care professional groups. Two specific questions were asked. Research Questions: Q l : What, i f any, is the perceived difference in impact of PCIS on health care professional groups in hospitals? Q2: What factors explain such perceived differences in impact between these groups? The focus of this study is on impact of using IT and not necessarily the factors related to successful systems. Therefore, the analytic framework (Figure 2.10) underlying this investigation is only the portion of the theoretical framework (shown in Figure 2.9) that is directly related to impact. Level 2 proceduralization Level 3 new capabilities Level 1 substitution Impact: Impact: Impact: structure structure structure process process outcome output outcome output outcome output Figure 2.10 - Analytic Framework 42 The next chapter describes the research methodology, followed by findings for each of the hospitals studied. Endnotes 1 The study reported: "Variations in three measures of patient volume were studied: number of patients per month, number of patient days per month, and average length of a patient stay in the hospital. The percentage effects of TMIS on the monthly patient census were separated from the influence of other variables, such as changes in demand conditions, trends in treating patients in hospital outpatient settings, inflationary pressures and controls, and patient diagnostic mix." Original results indicated that nursing services showed a decrease in costs while ancillary services had no significant decline. However, support services showed a significant change in cost and in fact the direction of effect was positive, estimated to be as large as 4.5%. Therefore, to arrive at the reported finding of "decreased cost," nursing and ancillary services were combined under one heading of "medical care departments." It should also be noted that the cost of operating TMIS was excluded from the above analysis and actually was an additional cost to the hospital. 2 See for example: Howell and Higgins (1990); Alavi and Joachimsthaler (1992). 3 Patient specific costing is required for Global Dimension reporting in the MIS Guidelines that were developed as a joint initiative between the Federal and Provincial Ministries of Health and Hospitals. The Guidelines are intended to provide a standardized method to report and compare costs within and across health service facilities. 4 See for example: Lee, et al. (1996); Bolley (1994); Burns (1994); Sheps, Rumanek & Noronha (1994); Massaro (1993); Bergman (1993); Michael, et al. (1990); Peterson, (1990); Siegel, et al. (1987); Young (1987). 43 Chapter 3 - Research Methodology 3.0 Introduction The purpose of this study is to investigate how IT affects the worklife of different health care professional groups (including contribution to patient outcomes) and why differences exist. In Chapter 2 a three-level analytical framework is proposed to guide this investigation. Impact is defined in terms of IT's effect on the prime goal of health care, improved patient outcomes. This de-emphasizes expectations of developers and implementors for increased productivity or decreased errors, and focuses on changes in worklife for health care professionals with respect to Donabedian's "quality of care" parameters. A socio-technical perspective is a relatively new approach to IT and change and empirical work is limited. The majority of research and practice based articles report findings related to the implementation process, but do not investigate impact beyond that point. 3.1 The Case Study Approach The case study approach is used to investigate the research questions. As Yin (1989) points out, this approach contributes uniquely to the knowledge of individual, organizational, social and political phenomena. The rationale for choosing this strategy is inherent in Yin's three conditions that determine research strategy: • type of research question; • extent of control an investigator has over actual behavioral events; and 44 • degree of focus on contemporary versus historical events. Each research strategy can be used for all three purposes: exploratory, descriptive and explanatory. Questions related to "how" and "why" are more explanatory and lead to the use of case studies, historical studies or experiments as illustrated in Table 3.1. (Surveys and archival analysis are more suited to questions which answer who, what, where, how many and how much.) Research Strategy Control Over Behavioral Events Focus on Contemporary Events Case Study no yes History no no Experiment yes yes Table 3.1 - Choice of Research Strategy (Modified from Yin, 1989, p. 17) A case study approach is preferred in examining contemporary events when relevant behaviors cannot be manipulated. Of equal importance in choosing a case study approach is that the phenomenon and context cannot be clearly separated. 3.2 Case Selection Researchers often think about "sampling" logic where subjects are randomly selected from a specified population and then results statistically generalized to that population. Multiple case studies use a "replication" logic, where each case is analogous to a single experiment (Yin, 1989). Yin suggests that cases must be carefully selected to either predict similar results (called literal replication), or to produce contrary results, but for predictable 45 reasons (called theoretical replication). The second strategy was chosen for reasons explained below. Given time and funding limitations, case selection was restricted to British Columbia. A survey conducted by the Ministry of Health in 1993 was used to identify eighteen community-based, acute care hospitals with patient care related computer applications (listed in Appendix A). Extended and continuing care facilities were excluded because their information needs, funding and patient-mix differ substantially from acute care facilities. Teaching hospitals were also excluded from the study because they are relatively few in number, and differ substantially from community hospitals in their size, complexity and access to funding. The study identified seven hospitals that had invested in a single vendor, hospital-wide integrated system.1 These hospitals were asked to participate in the study because they provided an opportunity to control for a large proportion of variability that is associated with vendor and system. Five hospitals consented to participate and their IT experience ranged from highly automated, with many modules implemented, to minimal automation. Including all five of these cases reflects Yin's (1989) strategy of "theoretical" replication. The hospitals are comparable on the variables of IT vendor as well as type and size of facility, however differ significantly on the degree of IT implementation. Many definitions and acronyms exist for the types of IT in health care. A Patient Care Information System (PCIS) is generally considered a subset of a Hospital Information System (HIS). A PCIS includes Patient Care Systems and Clinical Systems as shown in Figure 3.1. The broader classification of HIS includes a variety of financial systems, materiels 46 management and payroll/personnel modules. Only their order of implementation with respect to PCIS are of interest in this study. Each hospital started investing in IT at different times, but in all cases financial systems were in place before any decision was made to automate patient-care functions. This does not come as a surprise given that health care systems in Canada have been maintained through funding schemes tied directly to the number of "patient days," leading to financial systems being foremost in IT development and implementation. Payroll systems replace expensive outsourcing contracts and can be justified because they reduce outside agency expenses. They are not far behind the financial systems. Hospitals generally implement Central Patient Index (CPI) and Health Records abstracting systems next. They support the maintenance of accurate patient record information including demographics, admission and discharge dates, diagnosis and procedures. Clinical systems in the Lab, Pharmacy, Radiology and others follow, with order entry, results reporting and patient inquiry implemented last. The hospitals are arranged on a continuum from highest (1) to lowest (5) degree of PCIS implementation. Each hospital is assigned a number to indicate their relative order with respect to implementation. This ordering is not intended to imply their absolute positions. For example, Hospital 2 is not twice as automated as Hospital 4. The hospitals are referred to by their number through out the rest of this document to remind the reader of the context of their level of implementation as well as to protect the identity of the hospital. 47 Admitting Systems -Unique Patient Identifier -Central Patient Index - Adrnissbn/Discharge/ Transfer (A/D/T) Order Entry, Results Reporting, Patient Inquiry Patient Care Systems Health Records -Operations/ Management -Records Abstracting Care Planning Patient Appointments and Scheduling OR Scheduling and Management Administration Systems '-Financial ^ -Materiels Management -Payroll/ Personnel V ) General Tools -Report Writers -Electronic Mail -Executive Support Clinical Systems N Laboratory -Chemistry -Hematology -Microbiology -Anatomic Pathology -Bbod Bank Medical Imaging and Other Clinical Departments Pharmacy -distribution -clinical -inventory Figure 3.1 - Hospital Information System2 Hospital 1 has six years IT experience and the most highly automated PCIS. Hospital 5 is at the other end of the spectrum with the fewest number of integrated systems in place. Table 3.2 summarizes the history of automation in each hospital and generally shows a five year difference between Hospital 1 and Hospital 5. Although Hospitals 2 and 3 began implementing IT before Hospital 1, they introduced X T E C H later. (A more detailed description of the IT implementation history at each site is provided in Appendix G.) 3.3 Unit of Analysis Previous studies focus on a single view of impact from one of the stakeholder groups. This has a narrow scope i f all groups are expected to use, and will be affected by, the integrated system. Four groups of health care professionals who are high users and producers of patient information were selected to participate in the study, including laboratory technologists, pharmacists, nurses and physicians. In each hospital these groups were the unit of analysis, although the context of their experience was also important and analyses take individual hospital circumstances into account. In this way each group was a mini-case study embedded within a larger case, which is the hospital. 3.4 Data Collection In an effort to validate findings, three sources of data were used in the study: multiple interviews with participants, written archival data and observations of IT use. 49 3.4.1 Participant Selection Initially the study included five participant groups: four health care professional groups and managers, with four representatives from each group. The managers represented the management perspective in each professional group. Early into the interviews the interviewer discovered responses from the managers with respect to impact of PCIS were better grouped with their respective professional groups. Participants from each of the four professional groups were selected to represent a range of experience using the PCIS. Criteria for selection in each group included the computer liaison person, a manager and three other people who had a range of experience using the IT (i.e. most, moderate and least). The contact person at each hospital (usually the Manager or Director of the HIS Department) attempted to locate participants who matched these criteria. However, because there was an attempt to schedule all the interviews at each hospital within a single block of time, this meant several potential participants were unavailable. The expectation was that participants had a range of experience, and this was achieved. Appendix E identifies the distribution of participants for each of these categories. In addition, the manager of the HIS Department and/or administrator were included to provide a hospital perspective on the history of PCIS and expectations for impact. Table 3.3 summarizes the numbers of participants in each group. 50 Hospitals Applications/Modules 1 2 3 4 5 Administrative Support . Finance (AR7AP, GL) 1985 1984,1994 1994 199?* 1990 Payroll/Personnel 1987 1994 1994 1990 Payroll/Budgeting 1994 Materiels Management 1991 1984,1994 1984,1994 Executive Support System 1992 Communications Internal E-mail/Office Auto. 1990,1995 1994 1994 199?* 1990 External (community links) 1994 Direct Remote Link 1994 Physician Registry 1986 1986,1994 199?* 1994 Patient Care Systems Admissions/ADT 1985 1985,1992 1984,1994 199?* 1990 Central Patient Index 1985 199?* 1990 Order Entry 1989 1995 1994 Results Reporting/inquiry 1994 1993 Health Records/Abstracting 1989 1986,1992 1984,1994 1990 Medical Dictation 1993 Booking/Scheduling 1992 1986,1992 1993 1990 Clinical Systems Laboratory 1988,1995 1993 1984,1994 1994 Radiology 1991 1993 Pharmacy 1988 1992 1984,1994 1985 1984 Dietary 1986 1984 Documentation/Decision Support nurses' notes care planning patient assessment 1994 physicians' notes workload monitoring 1996 report writer 1987 Table 3.2 - Summary - IT Implementation in the Study Hospitals by Year (* Hospital 4 was unable to confirm the exact date of these implementations) 51 Hospital/ Group 1 2 3 4 5 Total Nursing 8 4 4 5 4 25 Pharmacy 4 3 3 4 3 17 Laboratory 4 3 5 4 4 20 Medical Staff 3 3 3 3 2 14 IS/Others 1 4 3 1 1 10 Total 20 17 18 17 14 86 Table 3.3 - Numbers of Participants by Hospital and Group 3.4.2 Interview Guide Development Semi-structured questions were developed for each of the six success factors identified by DeLone and McLean (1992). Individual impact was expanded to include the six social dimensions related to worklife outlined by Kraemer and Danziger (1990). Further questions were added during the interview to follow up on ideas expressed by the participants. The clarity and face validity of the interview questions were assessed by IT experts in the field from each professional group. These experts were identified through the professional associations for each group and their qualifications are briefly described in Appendix C. With input from these experts the questions were revised and then assessed again through a pilot study with several interviews. A sample interview guide is presented in Appendix D. 3.4.3 Interviews The author of this thesis conducted eighty-six interviews between February and June. of 1995. Each interview lasted approximately one hour and was audio-taped for later transcription. Participants signed a "Consent to Participate" before beginning the interview and were given the option of reviewing the interview transcript. The number of health care 52 professionals in the province is relatively small. It is relatively easy to identify specific individuals given the case selection from community hospitals in British Columbia that have chosen single vendor systems. Participants wanted to be guaranteed their comments would be anonymous within the report for their hospital. For this reason, no specific job titles are used in identifying quotations used in the case descriptions nor are the hospitals referred to by name. 3.4.4 Written Archival Data Over the course of time that scheduled interviews took place, written documentation and archival material relevant to the adoption and use of the PCIS was reviewed. Appendix F lists the specific documents reviewed. Activities during system selection and up to the present time were included in the time frame examined. This represented different lengths of time for each hospital as they varied in the number of applications in use. A review of these documents proved to be instructive in a number of ways: • requests for proposals (RFP's) to IT vendors and internal proposals provided information about benefits that different professional groups were expecting from a PCIS. • minutes of meetings, particularly task forces and steering committees that were struck to manage the PCIS selection and implementation. The membership of these committees represented the initial organizational focus on benefits. • strategic IS plan review and/or development was usually conducted by outside consultants who all recommended a single vendor strategy. • hospital newsletters illustrated the degree of IT integration into hospital operations as represented by information about upcoming changes that were reported. 5 3 • external correspondence, particularly with the Ministry of Health provided insight into provincial funding limitations and priorities that emphasized cost benefits analysis and financial management. One set of documents did not provide any additional information. Job descriptions for the three health care professional groups who were employees of the hospital were sought out. They were intended to provide information on one dimension of social impact that is associated with job enhancement, including skill variety and job domain (Kraemer and Danziger, 1990). In four out of the five hospitals, job descriptions had not been updated since before implementation of their PCIS. Most participants indicated the content of their jobs had not changed, but only the method of doing them. Hospital 5 however, had recently undergone significant changes in their organization of work when they moved from a departmental to a program approach. This had major implications for changes in work that were reflected in new job descriptions. While their PCIS supported some of these changes, the job descriptions did not change due to IT implementation. 3.4.5 Observations Activities and interactions with respect to use of the PCIS were observed over the course of time that interviews were conducted. Notes of each observation were recorded. Observations took place in the departments where the system was being used. This was true for every group except physicians, who were only observed when someone happened to be working at a terminal in the nursing station. In every hospital one participant from each group provided a tour of their work area and made comments about the terminals and their locations. These often related to ease of use and accessibility issues. For example, where 54 terminals had been introduced into an existing work area without ergonomic considerations, a number of problems were created with respect to overcrowding and uncomfortable seating. Overall, IT was difficult to integrate into the workplace due to physical limitations, budget restrictions to make the necessary changes and lack of experience on the part of the users to determine what changes would be needed. On Nursing Units, it was usually relatively easy to install one terminal and printer for the unit secretary to use. As additional system capabilities become available, larger numbers of nurses, physicians and other health care professionals are expected to access terminals in the nursing station. Hand held terminals are being investigated to resolve this to some degree. For the Lab, implementation of PCIS often means having their analyzers on-line. However, in some cases this also means trying to fit an extra terminal on the bench in-between Bunsen burners and other tools. Physicians had the least access in the hospitals, with few terminals and printers (usually in the Doctors' Lounge) that were often poorly maintained. This problem is partly resolved through remote access to the PCIS from their offices and homes. 3.5 Data Analysis 3.5.0 Introduction The data were analyzed using an interpretative approach (Miles and Huberman, 1994; Denzin and Lincoln, 1994; Yin, 1989). Issues and topics for further investigation were identified through iterative analyses of interviews, transcripts, observation notes and documentation (Orlikowski, 1995a). The initial topics were reviewed and aggregated to arrive at a set of recurring themes. (This process was facilitated through use of a software 55 package, FolioVIEWS®, that is described below.) The data were re-examined and an integrated summary for each group, in all five hospitals, is constructed along the delineated themes. Selected quotations serve to illustrate specific points. An introduction and summary for each hospital provides the context for the analysis. The analytic framework introduced in Chapter 2 is then used to analyze the effect of IT for each group and hospital. 3.5.1 Use of FolioVIEWS® A textbase manager called FolioVIEWS® 3.1 for Windows was used initially to assist in analysis of the qualitative data, particularly in the process of data reduction. In order to identify participants' perceptions of each topic, the researcher identified a strategy to search the data. FolioVIEWS® automatically indexes every word in the document and this list is available for review. Variations in spelling (such as turn around or turnaround) and derivations of individual search words (such as train, training, trained) can then easily be included in a search. Keywords in the interview questions were used to develop lists of search words. These lists were expanded to include additional keywords found in participants' responses to the interview questions. For example, when participants were asked whether computers control their work, many equated the concept of control with being better organized. Under the category of Individual Impact, "organize" and "tool" were included in the search for the word "control." Once particular words were located, FolioVIEWS'® hypertext feature for creating jump links between ideas was used to facilitate examination of similarities and differences among transcripts. Segments of interviews containing the search word were then tagged and saved in files for further examination. "Pop-up" windows allowed annotations 56 and memos to be added to the data where needed and document pointers to indicate connections to archival data. Table 3.4 lists the keywords used with Folio VIEWS® software to search the transcripts. 3.6 Issues in Reliability and Validity Criteria for judging the quality of case study research include aspects of research design (internal and external validity), related to generalizability of the study and measurement issues (validity and reliability), related to whether the instruments and their use capture what they are intended to. (Internal validity is a relevant measure only for explanatory or causal designs.) Steps taken to increase the validity and reliability of this study are discussed below. 3.6.1 External Validity External validity relates to the analytical generalizability of case study findings, or generalizing particular results to some broader theory (Yin, 1989). This is in contrast to the statistical generalization expected with survey research. Yin's recommendation for a research design based on replication logic, with multiple-case studies, is used in this study to support external validity. It strengthens the precision, validity and stability of the findings (Miles and Huberman, 1994). A range of users in each group across the five hospitals also supports generalization of the findings to other similar user groups in community hospitals. 57 Information System Quality (themes*) Information Quality Use/Usefulness access, accessible accurate acceptance accessibility depend (on), dependable adoption available, availability quality change (use for change sake vs. being useful) easy to use time savings impact (related to use) simple, simple enough timeliness implement, implementation system selection timely learn (to use the computer) terminal location turn around time orientation, password user friendly satisfaction, satisfied (with the computer system) user participation, input, involvement, interests represented support; train, training;* usage, use, using use (technology vs. information) usefulness Individual Impact Organizational Impact Patient Benefit/ Care accountability* attitude allergies audits communicate bed days benefits (for user) communication, electronic* benefits clinical guidelines cost, costs confidentiality control/organize/tool* culture convenience decisions, decision making* e-mail, electronic mail* community efficiency* fax discharge interaction, social MIS Guidelines errors interpersonal relationships money length of stay (LOS) job content optical disk, health record outcome, outcomes job satisfaction recruitment, retention preferences performance resource use safety, safer productivity* savings security social interaction success (e.g. administration, HIS support) waiting time standards of patient care support standards of practice utilization workload measurement vision Table 3.4 - Keywords Used to Search Transcripts With FolioVIEWS® 58 Three factors may limit the generalizability of findings in this study: • The study participants were in community hospitals. This may limit generalizability of the theoretical framework to types of facilities initially eliminated from the study, such as long term care facilities and teaching hospitals; • The participating hospitals chose a single vendor, integrated IT strategy. Generalizing results of this study to other implementation strategies would be unsupported; • The organizations participated voluntarily and were more likely to share information about successful change efforts. However, this was balanced by a range of successes in their IT implementation efforts. As well, the two organizations that chose not to participate also demonstrated similar variability in their IT efforts. 3.6.2 Construct Validity Construct validity is concerned with establishing correct operational measures for the concepts being studied (Yin, 1989). Yin suggests two tactics in case study design that help ensure construct validity: using multiple sources of evidence in a manner that encourages convergent lines of inquiry and having key informants review draft case study reports. Triangulation was used in this study to find convergence among sources of information and different methods of data collection. Three different data collection methods were used, including interviews, documentation review and observation. As well, three or four participants provided multiple sources of information for each group, at each hospital. A second strategy included eliciting feedback from the participants. They received copies of the original transcripts, as well as draft case study reports. Participants were invited to comment or provide feedback to the researcher by fax, e-mail or telephone. The comments 59 received generally related to inaccuracy of information reported, and these were subsequently corrected. Another strategy used to strengthen construct validity was to have the interview questions reviewed by "content experts." These were representatives from each professional group who were recognized by their peers as having IT expertise in their own professional area. 3.6.3 Reliability Reliability issues relate to the ability of other investigators to replicate the study. Increasing a study's reliability helps ensure that i f later investigators follow the same procedures and conduct the same case study again, they will arrive at the same findings as the original investigator (Miles and Huberman, 1994). Strategies to help ensure this is possible include documenting the procedures used (through a protocol) and data collected (through a database) in the case study (Yin, 1989). An Overall protocol for the multiple-case study design was outlined in the study proposal. A more detailed protocol for data collection was then determined (see section 3.4), and included the development of the interview guide, the process of participant selection and interviews, as well as archival data collection. A database of the documents examined also assists in replicating the study. An index of document "types" was created before the site visits and documents collected were catalogued accordingly. These are summarized in Appendix F. As well, all interview transcripts were coded and filed by participant and hospital, as the table in Appendix E illustrates. 60 Creswell (1994) also suggests that the chances of successfully replicating the study are partly related to biases and values of the researcher. In this case the researcher has over twenty years experience in health care IT, management and teaching. This has an indirect, but positive influence on reliability. 3.7 Potential Contribution of the Research Methodology There are many research approaches that have the potential to shed light on the question of impact of PCIS in health care. The multiple-case study approach is appropriate when the behavioral event cannot be controlled and for new, sophisticated technologies where the impact cannot be predicted. This approach also lends itself to a socio-technical assumption of interactive development between users and technology. A small sample of homogeneous units, coupled with a comprehensive data collection protocol, produced several insights: • Interviewing participants from four different groups about their use of an integrated system presents a unique opportunity to identify and analyze how the perspectives surrounding the impact of a single information system differ. This goes beyond previous studies that focus on individual users or groups, implementation issues, or impact defined by developers and implementors. • The value of a multiple-case study approach was demonstrated. While each participant described his or her own unique "story" with respect to impact, similarities among members of a group, and differences among groups, were delineated through analyses of multiple sites. 61 • Collecting data from several sources including interviews, documentation and observations provided a more complete understanding of impact and the context in which it occurs. Detailed findings from each case are presented in the next five chapters. Each chapter begins with a brief history of implementation at the hospital, followed by a discussion of the findings for each group and concludes with an overall hospital summary. Endnotes 1 Identifying the name of the vendor and their system immediately identifies the hospitals that participated and to some extent, the individual participants. For this reason, the system and vendor are referred to by a pseudonym, X T E C H . 2 This illustration is a composite that draws from a variety of X T E C H documents. 62 Chapter 4 - Comparison of Impact Across Groups Within Hospital 1 4.0 Introduction Twenty interviews were conducted at Hospital 1 from February 22 to 24, 1995 with eight representatives from Nursing, four from Pharmacy, four from the Lab, three from the Medical Staff and one from Hospital Information Services (HIS). They embarked on this integrated information system project in 1985 with a hospital president who was very supportive. Many participants indicate that the vision, knowledge and understanding of the HIS Director enabled them to set a direction and achievable goals for this project. They point to "success after success after success" as important in why they continue to succeed. It has been so contagious that both hospital administration and staff expect IT projects will succeed, both on time and operational. (See Appendix G for more detail on the implementation schedule.) The system is reliable with only occasional unplanned down times. Potential users in clinical areas were unable to be actively involved in the IT selection process in 1985 because they were not very knowledgeable about the technology. However, they did have opportunities to see vendor demonstrations. X T E C H was selected from the limited choices for integrated systems at that time. System selection was initially based on financial needs and finding "a good information system," then later they moved into clinical areas. One benefit of their integrated system is that interpersonal relationships between departments have improved because departments are working on a common project, as this participant explains: I think it's really improved. Personally, I didn't know very many people in other departments, whereas now I know a lot of people in a lot of departments. We have an HIS users' group that is composed of my 63 counterparts in all the other modules. So there's Finance, there's Admitting, there's Radiology, Pharmacy, Medical Records, Materials Management, Lab and Nursing. With respect to the PCIS portion of the project, the Clinical Information Systems (CIS) Coordinator has been instrumental in system development, training and on-going problem solving for the clinical areas such as Nursing, Pharmacy, Laboratory, Radiology and Dietary Services. This role is expanding in a new direction with a project to link community information systems with the hospital system. Physicians also seek out the CIS Coordinator for advice and teaching as there is no official liaison position in Medicine. Many of the X T E C H functions can be performed with four arrow keys, so it is very easy to use and as one user described it: "all you need is one finger and one eye." There is very little typing required and most of the responses can be built in advance so it is simply a matter of selecting the correct response from a list of possible choices. For example, when a patient is being discharged, the user can answer the question, "Discharge To?" by selecting one of the destinations listed in the pop-up window. The fixed choices prevent errors, provide better quality information and all the necessary statistical reports can easily be generated from the data entered. The current messaging system is also easy to use, once the user understands it. However, they are gearing up for significant changes in their registry system and office automation. Notices are prevalent through-out the organization and trainers have completed their own preparation. They feel the new messaging system they are installing soon is so intuitive that "you could sit someone down and they could almost figure it out for themselves, with look-ups, keys and the entire manual on-line." 64 4.1 Impact on Laboratory Technologists - Hospital 1 4.1.1 Introduction and History of Computerization in the Laboratory The Laboratory Information System (LIS) Coordinator was hired in 1988 when they began computerizing the Laboratory. The first lab module included Chemistry and Hematology, two sections of the Lab where the Coordinator had clinical experience. In cooperation with the section heads in these areas, she developed the system dictionaries, which are customer definitions for data, such as patient location, names of tests and ranges for normal results. She was also responsible for training, writing system documentation and developing downtime procedures. Initially training was conducted in a small computer training room set up outside of the Lab, but it was difficult for people to leave their work stations because everyone was busy. Ongoing training is also important in keeping up with the system changes and to ensure X T E C H is being used to its fullest potential. Training is now conducted at work stations in the department during day shift. Evening and night staff come in for a few hours before or after their shifts to learn about new changes and the LIS Coordinator trains them in her office. Training is important in system success. For the Lab in particular, given the volume and complexity of information to be handled, transferring skills from training into practice was also important. In addition, participants point out that experience using the system is essential. However, sometimes the time required to gain this experience is at a premium because new employees are needed on the job quickly. A perception of "limited training" is related to many of these and other issues, as this lab tech describes: 65 / found the training very limited. We had perhaps an hour, and we're expected to he proficient in it after that hour. The set up here is not a good one. We have a test system and a live system. When we get a new module we go into test system and play with it. The test system does not mimic what you're going to have in live, so you're kind of practicing something you're never going to use. You're practicing baseball and you're going to be playing hockey, you know. Initially when the laboratory system was installed, Nursing continued to send manual requisitions to the Lab and these were entered into the computer by Lab personnel. Results were available on-line soon after that. About a year later, order entry became available from the Nursing Units, and the LIS Coordinator worked in conjunction with the CIS Coordinator to train the nurses. Initially the Lab generated "Activity Reports" (one day's results for each patient) and delivered them to the Nursing Units. This quickly proved to be too cumbersome, so the Nursing Units were given the ability to generate these reports themselves, which required additional training. There has been a constant cycle of change since the initial implementation. New versions of their modules arrive annually that require a couple months of testing, plus updates to the documentation and staff training, before they can be implemented. Changes in the Lab such as their controls (used to standardize the analyzers), normal values (used in reporting results), admitting procedures or financial reporting requirements, also require updating in the system. The Lab is planning to implement three new modules this spring and summer: Histopathology, Microbiology and Blood Bank. They expect this to be a large project. The LIS Coordinator has less clinical experience in these areas, so the section heads will take the X T E C H training course and participate more in building the dictionaries and implementation. 66 In the meantime, until the Histopathology module is implemented, other capabilities of X T E C H have been used to routinize the collection and reporting of data in that area.1 4.1.2 Use and Impact of PCIS The laboratory system is easy to use, partly due to local contributions in developing the system. As well, additional information and instructions are available on-line i f needed. Every prompt that requires an answer is accompanied by a lookup that lists all the possible answers to choose from. This contributes to results being reported in a consistent way, as this example illustrates: For most of the tests in Chemistry, if it's a non numeric result like an RA or urinalysis, you can actually put in the results they can choose from and so it's really made consistency in reporting. Urine micro is a good example because where you trained determined how you reported epithelials, white cells, red cells or bacteria. You knew what you meant, but you could get five people to do the same urine and they'd all report it slightly different because of the terminology they were used to using. Now it's all consistent because they only have a choice of so many things, so I think that's a good feature. However, the feeling that there are too many choices and the questions are repetitive can have the opposite effect of annoying the user and producing information overload, as this participant indicates: There are some things that I don't find particularly easy to use. I think there's too many choices sometimes [on] the menu. When you start on your menu and you've got, I don't know how many categories to choose from and then you go into sub-menus and you've got X number of categories. I think they could hone it down a little bit and give people fewer choices. They expected many benefits from the PCIS, particularly in automating manual tasks to increase efficiency and reduce errors (level 1). The Accessioning section of the Lab is responsible for many of these tasks, such as handling specimens and requisitions, and saw implementation of the PCIS introduce many changes into their work. Workload has gone up, 67 rather than down, in one area of Accessioning where additional clerical duties are required to admit outpatients to the PCIS. The information is now more detailed and includes who is responsible for paying the bill. Prior to the Nursing Units sending orders electronically, paper requisitions were sent to the Lab. The Lab sorted them by unit to organize their specimen collections. When analyses were complete, results were transcribed onto the multi-part requisitions by the lab techs. These were sent back to Accessioning to be separated. Copies were filed and sent to the Nursing Units or doctors' offices. The manual system for ordering lab tests and reporting results is illustrated in Figure 4.1. With X T E C H , Accessioning now prints worksheets and labels for their routine collections at 5:00 and 10:00 a.m. and again at 1:00 p.m. (level 2). The computer organizes the lab tech's work so they do not have to worry about forgetting or missing something because it is on their worksheets. Nursing Units telephone stat and urgent requests to the Lab, and the collection labels print immediately in the Lab when they are ordered. After techs collect the specimens, they are "received" by Accessioning and entered into the computer. The specimens are then distributed to the appropriate sections in the Lab for analysis. The automated process for sending orders to the Lab and receiving results on the Nursing Units is illustrated in Figure 4.2. 68 Nursing Unit physician writes order for laboratory test Nurse or Unit Clerk completes paper requisition nurse collects specimen? YES NO write out specimen label collect specimens sort results by patient and file on chart 1 send copy to another Nursing Unit if patient transferred Laboratory Department sort requisitions | by Nursing Unit I write out specimen labels sort specimens by test type and analyze I transcribe results onto multi-part requisitions I distribute sort results by Nursing Unit I file one copy in the Lab Other Departments Finance: transcribe billing data and send Physician's Office send copy if outpatient results Health Records send copy if patient discharged Figure 4.1 - Manual System for Ordering Lab Tests and Reporting Results 69 Nursing Unit physician writes order for laboratory test Nurse or Unit Clerk enters order into PCIS nurse collects specimen? YES NO print specimen label results available in PCIS Laboratory Department print specimen labels and work lists enter sample received into PCIS I analyze specimens & report results in PCIS** T Workload Measurement stats automatically produced Other Depts. Finance: billing info sent automatically Health Records: results if patient discharged Notes: *Stat & urgent requests print immediately. Accessioning checks these lists to determine if other tests were done earlier and the current test can be added on, or duplicate tests were ordered by mistake. **this is done directly if the analyzers are on-line Figure 4.2 - Automated System for Ordering Lab Tests and Reporting Results 70 The Lab expected many efficiencies as they automated tasks. In the manual system multiple copies of each report had to be separated and sent out, resulting in delays in reporting results. The PCIS is more efficient because as soon as stats and urgents are verified, they automatically print out wherever the patient is. A big advantage is that even if patients happen to be in a temporary location, such as in the Operating Room instead of on the Nursing Unit, it will print in both locations. Prior to using the PCIS, the Lab was deluged with calls from the Nursing Units asking for information on the status of lab results. The calls were distracting and interrupted the technologists' work, particularly on week-ends. Nursing now has on-line access to this information, which eliminates the calls and improves efficiency in the Lab.. As well, an inquiry feature in the system allows the user some flexibility in choosing one or more test results to trend, over what time period, and whether or not to graph the results. Split screens are also available to examine two sets of results simultaneously. Accessioning continues to be responsible for keeping track of tests ordered for outpatients, inpatient tests being sent to outside Laboratories for analyses and reporting results to the ordering physician. A disadvantage of their increased efficiency in handling these specimens is the reduced time available for technologists to spend with each patient, as this technologist explains: Because [the computer produces the labels] we don't have to write the names on the tubes, and we have a little bit less time spent with the patient. That's not necessarily good for the patient because as an outpatient, you don't want that patient walking out before you're sure that arm has stopped bleeding. In those few extra moments that it took to write the name on a tube, we'd have a bit of conversation, make that person feel comfortable. 71 Other disadvantages of increasing efficiency through the use of IT is the speed work is expected to take place and the resulting effects on job satisfaction. In other industries this is likened to creating an "intellectual assembly line" that results from increased efficiencies, and therefore increased output. Two participants explain this feeling: ...It's really hard, but some days I know they feel like a machine. It's like an assembly line. You're just pumping out results. You can't control how much is coming in and how fast it has to go out. It has to go out and it has to be right, so there's that pressure. ...you know it is busier, the machines can put out results faster, bar coding has made things faster. It seems the faster [the machines] work, the faster you work. Plus, when the machines work faster, it means you have extra time so you start doing more specialized things. You're always busy... So every time you speed something up, you just get more work. This is related to bar coding specimens that contributes to efficiency and a faster pace of work. Using them may create a distance between users of the IT and the source of their information, the patient. This distance from "reality" is felt when specimens "flip by" and the lab techs lose touch with patients as people, as described here: The only negative thing I've heard about is that before every specimen had a name, whereas now every specimen has a number, so you've sort of lost the personal side of it a little bit. You had John Smith today, and you had John Smith yesterday, and you had John Smith the day before, now you don't even look at names. You're basically interested in the specimen number and the bar code number, so it's sort of dehumanized the system a little bit. Aside from the advantages and disadvantages of increased efficiencies, automating tasks is expected to contribute to decreased errors in a number of ways. Manually reporting results introduces the possibility of transcription errors. These are immediately eliminated when the lab instruments are directly interfaced with the computer system. For those 72 instruments not yet interfaced, the results are printed out and must be manually re-entered into the computer. A false sense of security may be created in thinking that automation eliminates all errors. Errors do not totally disappear, but they expect a decrease in transcription and calculation. There is little data to make comparisons before and after the PCIS, particularly since the types of errors change. Data accuracy remains related to the initial data entry, such as correctly identifying the patient when collecting a specimen, as well as selecting the right patient and the right order from the computer screen. However, some data entry "rules" can be built into the system to ensure the user enters more complete and accurate information. For example, when a blood drug level is ordered the user must enter the time of the last drug dose before the order is sent. (This was also a requirement in the manual system, but users could send the requisition without completing it.) There is no way of knowing whether errors in reporting occur (such as the wrong test ordered) unless the physician picks it up. Other types of errors, such as those occuring through omission, are reduced through automatically flagging abnormal results (level 2) in the PCIS, and reporting the normal reference range (sex and age adjusted) with each result. In the manual system the lab technologist had to know the normal ranges, or compare the results against standard charts. In addition, delta checks are built into the system so that results are automatically compared to previous results to check for significant differences (level 2). To help assess whether the abnormal results are significant, lab techs have on-line access to results from other lab tests, medication profiles, demographic and diagnostic 7 3 information. This saves both time and money in the investigation when tests are not repeated unnecessarily. The PCIS also flags critical values that are outside of the reference range (i.e. abnormal), and also potentially life threatening. The Lab has established a protocol for actions to be taken with these results. It includes repeating the test and phoning the results to the physician (for outpatients), or to the Nursing Unit (for inpatients) and documenting these steps in the computer. The number of such occurrences can be checked through an "Exception Report," which indicates how many results in the last twenty four hours were outside of the normal values and what action was taken, creating a new "visible" accountability for lab techs (level 2). "Visible" accountability which is created by using the PCIS can also be "short-circuited," by working around the computer requirements. These actions may negate some of the accountability as well as safety features of the system, as this comment illustrates: The computer can be erroneous in that it will say the specimen was collected at such and such a time and received at such and such a time. But sometimes things are pre-received before they're collected, which should be a no-no, but it isn't. ... You have to go by what's in there and that's why it's very important to make sure you put what you actually did rather than what you wished it would be. There are a number of examples of "visible" accountability which extend to other professional groups as a result of the changing role of lab technologists. As the computer takes over some of the checking functions which were described earlier, accountability shifts to the originator of the order for ensuring the right tests are ordered, and ordered correctly, as this tech explains: I think that [manual checking] has been the role of the Lab throughout, even before we had the computer system. It was always the Lab who seemed to do 74 the troubleshooting, to catch the [cases of], "we don't have to do this test because it's already been done," or "I think this looks different, we should talk to somebody about it. " I think that the computer system has taken some of that and put it back to the person who originates the order. At least now you can find out who put the order in and actually track that person down. Whereas before they would say there was an order and then an order never came, or "Well, it's in here so you lost it," so it's our fault. But now if it's not in the computer, you didn't put it in. So I think some of that has gotten pushed back to the nursing people, they have a responsibility. It doesn't always just fall to the Lab to pick up the slack or figure out problems. "Visible" accountability extends to physicians as well. In the manual system the Lab recorded critical values in a book and situations out of the ordinary were not always easy to see. With the PCIS, lab techs are more aware of clinical decisions because of the availability of information, and therefore physicians' practices become more "visible," as this tech describes: Another benefit I found is that you can look up certain individuals and ask why they are having all these tests done, day after day after day. We're bleeding them dry. With the computer system you can just look and say, "Maybe we'll get a pathologist. " Not that we have that kind of clout, but you can get a pathologist involved and pull up the data on that patient and say, "Maybe you can look at this " or "His glucose has been running abnormal for seven days and nothing seems to be [happening] in his treatment. Could you maybe get involved in this? " and I think sometimes we can be beneficial to [thepatient's] treatment. Aside from "visible" accountability, implementing an integrated PCIS has proven to be both beneficial and detrimental for the Lab. The relationship between Nursing and the Lab has historically not been very positive with each blaming the other for problems. Integration has provided an opportunity to improve relationships between these two departments, with the audit trail providing specific information on where problems occur so these can be addressed directly with the individuals involved. 75 An integrated system is complex and support from the X T E C H vendor has been good. However, major changes take a long time and the company representative seems to be very busy with many clients. Until recently there were some inconsistencies between modules in X T E C H , such as keys on the keyboard being defined differently in different modules, which caused problems in the Lab when they were accessing information from other areas such as Pharmacy and Radiology where they were still using an "old programming language" from an earlier version of X T E C H . Another related disadvantage of an integrated system is that the combination of old and updated modules may not be the same in any two given hospitals. This means new modules are not tested under all possible conditions, sometimes creating unanticipated problems during implementation. Integration is also related to issues of confidentiality which are important as more users have access to more information. In the manual system, confidentiality relied on professional ethics and physical evidence of inappropriate access, such as looking through a patient chart. With the PCIS, confidentiality becomes more formally linked to accountability because inappropriate access to information can be monitored electronically through an audit trail. These issues have lead Hospital 1 to recently re-examine their confidentiality policies and the consequences for breaching them. Issues of efficiency, productivity and integration are related to workload and its measurement. The Laboratory uses a national workload system with a long standing history. When new IT and instrument automation is introduced, the national unit values change (i.e. the time allocated to each task), but these may not always be in synchronization with individual hospital changes. Overall, staff are more productive because they are able to 76 complete additional work with the same number of people. For example, over the last ten years with automation and computerization, there have been no increases in Chemistry and Hematology staff. This has been in spite of the fact that they estimate the number of specimens they are processing has probably doubled, or more, in that time. Workload statistics are generated monthly per bench, or work station. A n average workload per technologist is calculated, but information on individual techs is not available. Individual workload is monitored in a general way through the "outstanding specimen list," which is generated at the end of the day (level 2) to determine what work has not been completed. Questions can then be asked i f particular technologists routinely have incomplete work (resulting in another "visible" accountability). Another activity affecting workload is the Lab technologists' responsibility for monitoring the quality of their testing through ensuring the reliability of their analyzers. Quality Control (QC) measures are established to ensure all the instruments, analyzers and manual methods are working within specified limits. For example, some analyzers are set up to do the necessary calculations (level 2), but initially when they start using those instruments, the calculations are all manually double checked. Abnormal quality control results are flagged for the user who must document in the system if corrective action was taken (creating another new "visible" accountability). Collecting the data and generating QC graphs has some value in keeping the user closer to the data. In the manual system, QC results were plotted by hand on graphs posted on a big bulletin board, where they could see the results "at a glance." QC was more an integral part of the Lab when everyone was reminded of quality scores, however computer 77 generated reports remove these visual cues and sense of ownership from the environment. The QC reports and graphs are now filed in a binder, making them less accessible, but the lab techs still look up specific information when they work on a particular bench and the QC summaries are posted. 4.1.3 Use and Impact of Electronic Communication Lab techs had few comments about the use of electronic communication. In situations where communication was "not at it's best," e-mail reduced it even more. They find it is so much easier to just type something in the PCIS, then it is to try and get hold of people by phone. However, the responsibility for reading and responding to the message rests with its receiver, as this participant indicates: And then again, it's there. You let them know. Whether or not they read it is their concern. So that's a weakness of the system, I think. You can use the system too much rather than having some interpersonal skills, [laugh] 4.1.4 Summary The impact of PCIS on laboratory technologists at Hospital 1 is summarized in Figure 4.3. It was difficult for lab technologists to have input into the initial decision to select X T E C H . However, sections that are implementing their systems now (like Bacteriology and Histopathology) have the benefit of the collective experience of other areas when making decisions. As well, because they are using an integrated system, there are many similarities between the modules. 78 Level 1-substitution Level 2 - proceduralization Structure a^utomate manual a^utomate procedures tasks in the Lab >checking abnormal 1 results (Exception Report); A Outstanding specimens; 1 efficiencies (less staff) >delta checks (against t data quality (less errors) previous test results); i turnaround for results >specimen collection lists ^ expected Process + better/faster decision-making by nurses and physicians ^ expected Patient '+ printed instructions Outcome • (better specimen, ! patient understands) Figure 4.3 - Impact of PCIS on Laboratory Technologists at Hospital 1 Lab techs benefit primarily through automation of manual clerical tasks which result in increased efficiencies (they can do more with less staff), improved data quality through less transcription errors and reduced turnaround of results. The Lab also benefits from the automatic generation of many reports and flagging of abnormal results rather than depending on individual techs identifying the abnormalities. Increased efficiencies can also have negative effects for the Lab such as the increasing pace of work plus creating a distance between the system user and patient. 79 Lab techs expect physicians, nurses and patients to benefit in a number of ways. Patients should receive better care when physicians can get accurate results the minute they are verified without having to go through a number of phone calls. As well, patients benefit through having instructions print out when certain tests are ordered. These provide details of the test, what to expect and instructions to follow, which help prepare the patient and ensure a better specimen is collected. Lab Manuals which provided information on specimen collection have always been available on the Nursing Units, but were' not necessarily easy to access. In the PCIS, all the information is available on-line and can be accessed before preparations are made to collect the specimen, which benefits the patients, nurses and lab techs. Prior to the PCIS being introduced, the Lab estimated that twenty-five percent of the requisitions were erroneous in some way. Since that time they have reduced the error rate to two percent. 4.2 Impact on Nurses - Hospital 1 4.2.1 Introduction and History of Computerization in Nursing Implementation of X T E C H in Nursing has occurred in stages over the last six years. It began with the Admitting module and order entry to the Lab. Electronic communication of orders expanded to other departments such as Dietary, Radiology and Human Resources as they came on-line. A l l orders, except medications, are entered into the computer. Medication orders continue to be faxed to pharmacy, but they are aware nurses enter orders to pharmacy in a similar hospital using X T E C H . Nurses at Hospital 1 have not wanted to start doing this primarily because ward clerks already enter the other orders. 80 As noted in the hospital introduction, the Clinical Systems Coordinator (CSC) was responsible for PCIS implementation in a number of clinical areas including Nursing. In conjunction with an implementation committee, she organized training and provided active support which was important in the system's success in Nursing. Additional training was provided for key people on the Nursing Units such as head nurses, assistant head nurses and ward clerks, who became the trainers for their own areas. From a Nursing perspective, the PCIS project has been supported by the HIS Department as well as all levels in Nursing. Allocation of money for training and commitment to having staff leave the Nursing Unit for training sessions is indicative of this kind of support from Nursing Administration. Training began as a specialized task for the Clinical Systems Coordinator, who is located in the HIS Department, but as use of the computer became integrated into operations of the organization, its use became an expectation for employment. Using the computer has become incorporated into hospital orientation and new users receive two four-hour blocks of training time. They come away with a checklist to help them identify where they need additional computer time. Training has evolved from simply learning to use the computer, to integrating it into the workplace, as this comment illustrates: We put a lot of effort [into training] and we still have special sessions for our ward clerks. We bring them all together [and ask]: "What kinds of things bug you? What kinds of things are you doing that take a lot of time? " because maybe they didn't know there was a different way to do it. We have meetings and on-line minutes are shared with all of them in a [computer] cabinet. I think it's just giving them a little sense of respect... and they love to being able to criticize and say, "I don't like this, and what can we do to make this better? " Having very open and direct lines for communication [is important]. 81 A "train the trainer" concept has always been used, but with the move toward more specific Nursing applications, more of this responsibility will be taken on by a new Nursing informatics position, assisted by the clinical instructors in Nursing. 4.2.2 Use and Impact of PCIS Nursing uses the PCIS extensively, and their dependence on the computer is summed up by this statement: "I would say that if a nurse didn't have a password, the nurse would not be able to work because everything the nurse does for ordering is done by computer. So, all of their orders, all their referrals, if a patient's side rail on the bed need to be fixed, it's put into the computer. They don't phone anyone, so any of the phone calls they used to make, it's all done by computer now. " But not all nurses were keen to use the system at first, as this participant points out: It works a lot better than I thought it would. I was probably just as pessimistic as everybody else when we started. I loved bedside nursing and I was doing general duty when it came in. I sort of resented going to computer school. Like I thought, "Well, the system is fine. What's the problem?" But, now I think it's wonderful and if I went to a hospital that didn 't have a system as good, I think I would really miss it. Several benefits were expected for Nursing, including improved communication between nursing staff working different shifts, between nurses and physicians and generally in keeping in touch with what is happening. They also expected to spend less time on time consuming, tedious, duplication of handwritten requisitions (level 1). This included elimination of daily recopying of medications on the medication profile and diet orders. A third benefit was access to on-line information which facilitates efficiency in retrieving results. Stat test results can be returned quickly, in fifteen to twenty minutes. In the OR for example, a frozen section can be sent down to the Lab during surgery, and the results returned before the surgery is complete. Activities on any Nursing Unit are often 82 influenced by many outside variables, such as the Operating Room schedule, and therefore increasing elements of "control" over their environment increases job satisfaction. For example, rather than waiting for the Lab to call with results, the nurse can check on-line for results and avoid delays in getting the patient to the OR. Another benefit for both patients and nurses is the ability to store allergy information and have it available on subsequent admissions. Eventually a full electronic chart is envisioned as a tool which will provide access to all stored information and eliminate the time spent searching for old charts and information they contain. The electronic chart is close to being a reality now, but it is projected to be another three years before Nursing is computerized and their documentation on-line. They will be able to achieve benefits directly related to nursing care then, as this participant explains: I'm convinced that if we can get through the next three years and get nursing documentation on-line it will be very, very important for the care that we're giving. Right now in Nursing you'll see lots of frustration because everything is still handwritten, whereas if you went into Radiology, they don't pick up a pencil all day long. .... It's all done on-line. Patient checks in, they do what they need to, patient checks out, they've got another patient in there. It does all their finance, billing, statistics, everything. They don't have to do any notes or keep track of anything manually. The time freed up through these efficiencies is generally expected to benefit patients because nurses can spend more time at the bedside. But Patient Care Coordinators point to the fact that nurses cannot be found at their nursing stations now because they are busy with patients. Whether this is related to using X T E C H , or because of the kind of nurses they are, is not clear. Participants suggest that while the system may free up time, whether that time is spent on patient care is really dependent on the nurse, and those who want to spend extra time 83 with their patients already do. Use of IT does not change a "poor nurse" into a "good one" as these two comments indicate: I've often felt that although they've perhaps intended to save time, I'm not convinced that they do....I mean one could be just as productive in less time but in some ways, and particularly in psychiatry, there is a certain dynamic that actually having the terminal in the nursing station kind of defines and makes that being in there legitimate... so sometimes I wonder whether less time is actually spent with the patient, to be honest. ...logically you would think [that computers free up time to spend with patients] but I think there's a whole other factor. When you have staff that are motivated to spend time with patients, it's because of their own beliefs and just freeing up unmotivated staff doesn't change their motivation. Automating clerical tasks through communication of information and requests via computer also creates a new "visible" accountability for completing the request correctly. In the manual system there were many instances of requisitions going astray, being incomplete or in error. It was nearly impossible to follow-up on these problems, resulting in sometimes tense relationships between Nursing and other departments. Now a complete audit trail means the errors can be detected and followed up as this participant describes: / can go back to the days when Nursing was blamed for doing things incorrectly and the whole ward was blamed. For example, "ICU nurses, you're always screwing up when you do things. You're always making mistakes. " Whereas now, if one nurse in ICU doesn't understand how to use the tool and that person is having a problem, we can sit down with that person or send them a little note, "Didyou know when you do this, this is the problem it causes in the Lab?" So, in a way, it's a quality assurance issue. Then that person learns how to do it properly. Often I would even get a little thank you note back from that person saying, "I didn't know that. Thanks for letting me know. " So, it's really building a sense of respect for one another, instead of blaming a whole group of people. This "visible" accountability is also stressed in the training for new staff who are told that the system is complex and there will be follow-up on any of these kinds of issues. 84 The computer is seen as a tool that provides access to information, but making decisions is still the nurse's responsibility. In some ways decision-making may have improved, but a more serious concern was the "data-rich, information-poor" or DRIP syndrome. One nursing manager described her decision-making as "probably better" only because she knew where to find information in the system and was able to retrieve it with the help of reports set up by HIS. Decision-making is also supported through many standardized patient teaching materials which are used to remind nurses of procedures and define standards of care. Participants describe several examples where they have developed systems on X T E C H which print these information sheets automatically in response to orders being entered ' (level 2). The overall lack of computerization in their area is a concern to Nursing. The response from other departments or administrators has been, "Why should we spend $600,000 for a Nursing System?" and IT investment is difficult to sell when returns cannot be measured in dollars saved or revenue generated. This is expected to change with a nursing administrative position now responsible for spearheading IT projects. She is developing proposals to convince senior administrators of the "worth" of these expenditures. The following comment illustrates these issues: Now I find it fascinating that we don't have any qualms about computerizing other departments. Nursing traditionally is always last. Nurses don't even know we're not computerized in the Nursing department. We feed into other modules like Lab and Radiology, but we don't document on the computer system. I think that Nursing is punished usually because it's going to cost more when you're training 900 staff. That's where your big costs are, the training costs. But if you compared that [to other departments]... if it cost you $60,000 to do Radiology and you have 30 staff, what is that per person compared to $600,000 for 900 staff it's probably about the same. So I think that often we come under a more critical eye because of the dollars attached to it. But on the other hand the Nursing Department spends a third of the 85 budget. We have a thirty million dollar budget, so we really should try to gain some efficiency. A 5% efficiency in Nursing is a lot more dollars than a 5% efficiency in a department that has a $100,000 budget... The Nursing Module is the first step in computerizing Nursing and it is expected within a year. Many nurses feel that the computer could be used in ways to support the work that is uniquely Nursing (level 3), as this participant's comment illustrates: But, there's so many things out there right now that's wrong in Nursing, that nurses are quite disgruntled. 1 think one of the things we haven't done in the past, is provide nurses with the right kind of tools or systems to do their work. We haven't valued that or seriously looked at that. We've just sort of said, "Go out there and do it. " Nursing seems very task oriented to me right now. We need to step back and look at the big picture, look at the patient as a whole, what's the most appropriate thing to do for the patient instead of getting our cart and starting down the hallway... There are many decisions to be made with respect to implementing computer programs for Nursing due to the highly integrated nature of the information they use as illustrated in Figure 4.4. Demonstrations of automated versus manual charting have highlighted the expected reduction in errors, which is an appealing feature to these prospective users. Cost of Nursing Care i Workload Measurement Documentation of Patient Care & Interventions Scheduling of Nursing Staff Individual Patient Care Planning & Interventions Figure 4 . 4 - Relationship of Information Systems in Nursing 86 Documenting nursing care using bedside terminals is another consideration in implementing the Nursing Module. As with many technologies, it is easy to focus on where the terminals will be located, their size, convenience and user friendliness without having clear expectations as to how their use will affect patient care or patient outcomes. Bedside terminals are expected to provide nurses the opportunity to enter their documentation at the bedside as activities and tasks are completed, rather than waiting until the end of the shift and documenting care from notes kept on slips of paper. Anecdotal evidence from managers suggests that staff report staying overtime to finish charting, which is expected to be eliminated with bedside charting. However, many factors contribute to documentation being left until the end of the shift, for example, inefficiencies during the shift, staffing levels and unexpected events, which bedside terminals will not affect. In preparation for documenting care on-line, the Nursing Department has started revising their manual documentation system in order to streamline it and facilitate transition to automation. While it is difficult to document changes in time spent on documentation, a "before-and-after" study was done to evaluate how the revisions changed the content of the documentation. They found the information to be more relevant and concise. On-line documentation also opens up other possibilities for multidisciplinary charting. The single, physical location of the paper chart limits the success of this concept in a manual system. For example, a changing program focus in psychiatry from inpatient to outpatient, coupled with no regular ward clerks, provided an incentive for this area to develop an on-line chart. In this case, the patient and nurse establish goals which are recorded on-line and then 87 are accessible to the social workers and physicians. At this time physicians' orders are not recorded directly on this chart. Nursing is about to purchase a Workload Measurement System which will interface with the Nursing Module. The automated scheduling system already in place is a stand alone system, but there is some potential to link workload with scheduling. Both nurses and managers benefit through another related ability to enter human resources information into the computer. Nurses can make requests for leaves, overtime pay or vacation time, which are authorized by the manager. A l l the departments that need to be notified, such as payroll and scheduling, are automatically notified at one time. The nursing managers use the system extensively for writing reports, determining trends in data (such as comparing budgets, supplies and patient census in materials management) and program analysis (comparing patient census, patient populations and diagnosis). Simply automating the recording of data previously kept in log books makes it easier to search for information and identify trends. For example, every surgical case was recorded in the OR log book. Now they are entered once into a database, and the information used multiple times from there. However, they still experience a few problems with the flexibility of X T E C H in being able to collect the statistics to generate the necessary reports. This stems, in part, from the fact that the vendor is American, and the reports required for the provincial government use different patient classifications than those available in X T E C H . 4.2.3 Use and Impact of Electronic Communication Many nurses use electronic mail (e-mail) extensively and check their, messages daily when they are working. Patient Care Coordinators take advantage of this by communicating 88 changes on their units to all staff, as well as routinely forwarding policy and procedure changes. While the computer manuals are on-line, Nursing is just beginning to use the X T E C H capabilities for accessing their many other manuals electronically (level 2). With shift work and the large number of employees each nursing manager is responsible for, sometimes they do not see each other for long periods of time. In spite of that, decision-making can become more participatory because all staff can easily be contacted for their input. In the same way, Patient Care Educators are able to develop more materials quickly and inexpensively using X T E C H , as well as work collaboratively with staff, physicians and managers to develop them. E-mail also makes it easy to follow-up on potential problems, as this manager comments: You can say to somebody, "I understand you had a problem last night on your shift. Can you leave me a message about the details of Mrs. So and So's care? " or whatever instead of trying to wait and find out when you next can connect with them. You can already start some data gathering or whatever you want to be doing on this particular problem. A change noted with electronic communication is that intended receivers become accountable for reading their message in a very direct way because an "acknowledgment" can be built into the message. When messages were written in the communication book, the onus remained on the sender to retrieve the book and check that everyone had read the message. Now the onus is on the receiver to open the message and acknowledge its receipt. The following comment illustrates this point: One of my coordinators doesn't check in and read her messages which is what normally other people do. So yesterday, for example, she didn't come to the portfolio meeting. She didn't come to the meeting because she did not read her message, but the message I sent was a reminder of the schedule that we set up in December for the next six months and so all along there she's missed the communication. 89 There is a concern that e-mail will create a loss of face to face communication, but this is balanced by the efficiency of getting the same information out to such a large group of people. (Problems that need to be addressed on a one-to-one basis can still be handled that way.) Everybody interprets messages differently and this may still happen with e-mail, but at least the message is consistent. This comment extends to positive or negative messages which can be sent out simultaneously to a much wider audience. As one participant points out, the effect of positive comments such as, "you are doing a great job" may not balance the effect of wide spread negative comments. Electronic communication likely reinforces previous communication patterns, as this participant indicates: Well, theoretically [the computer would affect social interaction]. But again, it's only as good as what goes in, so if people aren't communicating assertively one to one, they 're certainly not going to on the computer either. Requests for information or services received via e-mail are seen as higher quality than verbal requests received in the hallway or cafeteria. The added information often provided electronically saves a lot of time in understanding the request. Electronic requests are also seen as superior to voice mail. Although the receiver can listen to the message several times, e-mail offers the opportunity to see it in writing, take a hard copy if necessary and mull it over before responding. Another added advantage of using e-mail is the reduction in filing cabinets needed. As users have become more sophisticated in their use of the system, they no longer need to keep printed copies of everything, "it's just in the computer." 4.2.4 Summary The impact of PCIS on nurses at Hospital 1 is summarized in Figure 4.5. Nursing's coordination role means many of their communication tasks have been computerized in order 90 to facilitate systems in other departments. For example, the clerical tasks of sending orders and receiving results is well developed. Within the last year Nursing has begun to realize how far behind in automation they are with respect to the rest of the hospital. They expect changes in this direction through their plans for implementing a Nursing System and hiring a nursing systems coordinator. With integrated systems, Nursing is at the centre of many systems and must manage expectations in a new way. So many new ways of doing things are possible, but may not be practical or feasible. They have to learn to assess the cost in time and effort for developing new applications and training to use the technology in a new way. This is highlighted in a physician's request to have Nursing print forms for special types of assessments, which would mean 900 nurses would need to learn how to do this - an exercise not practical or cost effective. The shift to working with the community is a new area which was not initially anticipated for development by the HIS Department (level 3). Historically hospital nurses think primarily about care within the walls of the hospital. Although they are aware of the need for public and mental health services, these seem to be completely separate from acute care. Extending the PCIS into the community opens up that window to see "we're only just a little piece of someone's life and they need a whole continuum of care. When somebody drops into the hospital for a week, two weeks, three weeks, a month, there is still an entire continuum out there." Linkages with the community may become easier and perhaps more effective with the extension of their PCIS (one of their current projects), and provide an opportunity to look more globally at coordination of care. 91 Level 1-subsh'tution Level 2-proceduralization Level 3-new capabilities Structure * automate manual tasks in Nursing >on-line orders and results J f efficiencies t data quality >decreased errors * automate procedures >print patient information sheets & procedures a^utomate documentation >nursing care plans >changes to documentation * trend patient/ financial data *predicted >mulh^ciplinary charting ^ expected • expected : expected • Process + free up time to spend with patients *pre dieted • >critical pathways • Coordination of care ! in hospital *predicted Coordination of care at the • corrrnunity level ^ expected ^ expected : expected • Outcome J + better outcomes J \+ better outcomes i I • • \+ better outcomes \ i • Figure 4.5 - Impact of PCIS on Nurses at Hospital 1 4.3 Impact on Pharmacists - Hospital 1 4.3.1 Introduction and History of Computerization in Pharmacy It seemed to pharmacists that the selection of X T E C H was based on decisions for IT in non-clinical areas, as this pharmacist points out: I think the decision was made for what system we would go to based mostly on financials, medical records and admitting. I think the [HIS Director] was part of the group who made that decision, but I don't think the clinical 92 components were as important in making the decision as the financial component, especially at that time. A supervisor in Pharmacy was responsible for implementing the computer system in their department, a process that began in May 1988. Initially they automated the dispensary and that was the focus of attention for two weeks while the "bugs were ironed out." They soon realized i f they did not enter orders correctly, the computer did not handle them correctly. Intravenous solutions (IV 's) were added next and they went a lot faster because there were not as many exceptions. In August of that same year they began implementing the inventory portion of the system. One pharmacist describes X T E C H as "usable, flexible and continuously seeing improvements," although this view is not shared by all participants. The system is easy to use, but pharmacists require a little bit of training in order to enter the orders correctly. New pharmacists receive a four week orientation, most of which is spent learning how the Pharmacy Department operates. This includes a day and a half orientation to using the computer that they spread out over the four weeks. Training is a little bit frustrating for people who come with experience using different systems because they have to re-learn their keystrokes. For all users, the hardest part of learning to use the system is understanding how what pharmacists enter affects administration of medications by nurses. If orders are not entered correctly, they print on the Medication Administration Record (MAR) in a way that nurses have trouble understanding them. Ongoing training continues to be available and staff are notified electronically when these sessions are being offered. X T E C H is used extensively in Pharmacy, with approximately twelve terminals and four printers available. Terminals are conveniently located in the dispensary, an overflow 93 work room, offices (two supervisors, the director and a drug utilization specialist), technicians' work area, IV room and clinical area. Pharmacists can also access the information they need from any terminal located on the Nursing Units. One frustration with X T E C H that pharmacists have identified is the difficulty in getting changes made to the system. Areas Pharmacy identifies as needing improvements take a lot of time and money to change. A majority of the pharmacy module users in hospitals across Canada and the United States must agree to any major changes and then it takes another two or three years before these changes actually become part of the system. Unlike other hospitals that have modified the X T E C H system "beyond recognition," the Pharmacy Department has only made minor changes in order to avoid additional problems created when they implement system upgrades. 4.3.2 Use and Impact of PCIS Pharmacists expected the system to provide benefits in automating many of their manual tasks such as producing labels, resulting in less errors and higher productivity. In addition, they expected to be relieved of tedious tasks such as calculating workload statistics (that requires counting the number of IV's and dispensary items processed), as well as producing reports on drug usage. These changes were expected to result in better quality of medication therapy. Pharmacy could not go back to the manual system because their workload has increased dramatically over the last few years. They have also gotten involved in more programs and services, many of which could not be done manually (level 3). A computer 94 downtime for more than several hours quickly makes them realize how dependent they are on the computer, as this comment about increasing volumes of work indicates: We did IV admixture before we had the computer, but the work on it has increased tremendously. What we could cope with under a manual system, we couldn't cope with now without a computer system. Manually we used to do maybe 100, 150IV's a day. Well right now, we 're in the area of between 400 and 500 IV's a day. So, if you had to sit down and manually start keeping track of that, preparing all the labels, you wouldn't be able to cope. In the manual system Pharmacy primarily kept track of the medication orders they were responsible for dispensing. With introduction of the pharmacy module, they made the decision to enter all medication orders into the system, whether Pharmacy dispensed them or not. Initially this required an increased number of staff. Since then staffing levels have stayed the same, although the computer enables them to operate additional programs, such as IV admixture and unit dose. For example, the IV Admixture Program was also started early in 1988, and Pharmacy staff began to fill intravenous (IV) drug orders based on information supplied by Nursing. Relieved of this duty, nurses were expected to have more time for patients, and pharmacists could focus more on drug therapy. In addition, IV drug orders were fdled under sterile conditions and underwent several checks, eliminating up to seventy-five percent of potential errors. In the October (1992) Hospital Newsletter, the introduction of a unit dose drug distribution system was announced. The Pharmacy Director commented on how the computer supported these types of initiatives that reflected a new era in Pharmacy. They expected many of the same benefits the IV Admixture service demonstrated. Because the computer system was already in place, they were able to identify features of a unit dose system that would be beneficial and then how X T E C H could support these. This increased 95 automation of the dispensing process was also expected to allow pharmacists to increase their presence on the wards in order to consult with physicians, attend patient care rounds and enter orders directly into the computer. Automating tasks is expected to provide benefits through increasing efficiencies and productivity. However, the nature of these increases is not easy to evaluate as a number of changes have occurred through out the distribution process. This process is illustrated in Figure 4.6. In the manual system, technicians typed up labels for drugs to be dispensed. Pharmacists decanted the drugs and checked their labels against the original order. Information typed on the labels was simple and allergy checks were made i f the medication order sheet listed allergies. Their automated system (in combination with unit dose drug distribution) is more time consuming and elaborate, but they also have more information available. This has produced changes in the work of both pharmacists and technicians, as this pharmacist explains: We [pharmacists] do all the order entry. The technicians also use it for their ATC machine. ... They go in the computer and they do all their debiting and crediting and run their lists. So they use it quite often as well, to do labels and... But, you'll see a pharmacist at the computer probably more often, like we just sit there and basically type all day long. One benefit for pharmacists when they enter all the orders is that they automatically receive information back on drug allergies, drug interactions or orders for drugs in the same drug class. Technicians use the computer to run a list of medications due. They use the A T C machine4 to produce strips of medications needed for twenty four hours for most of the orders and collect the remainder by hand. Initially pharmacists check the medications against the original order and on subsequent refills they check them against the medication profile. Any 96 doses not used on the Nursing Units are returned to Pharmacy and technicians enter them back into the computer as "credits. Nursing Unit physician writes medication order , Pharmacy Department F A X * r YES check MAR* ^ against order record medication on M A R medication] given/ N O receive medication onNU* enter order into PCIS* YES problem? N O L 7 1 produce print M A R labels produce medication profile re-order stock update inventory medication returned to pharmacy *Definitions PCIS = Patient Care Information System M A R = Medication Administration Record N U = Nursing Unit F A X = facsimile transmission I enter medication into PCIS Figure 4 . 6 - Medication Order and Distribution Process 97 In this system more information is entered, but to their advantage, the same information can be accurately reproduced in a variety of formats for different users. For example, medication orders are used to produce M A R ' s which are used by nurses; pharmacy technicians use the same information to prepare their unit dose lists; and pharmacists use the information for their patient medication profiles and medication lists for patient teaching purposes. Another benefit for Pharmacy is the increased quality of medication therapy, including appropriateness of drugs ordered. However, it is difficult to make comparsions with the manual system because they cannot determine the number of drug interactions that occurred because the pharmacist was unaware the patient was on a particular medication. With X T E C H , drug interactions are displayed during order entry and the pharmacist decides if they are significant for that patient, then either responds to the warning or overrides it. Pharmacy does not specifically keep track of how often this happens, or the consequences of pharmacists' decisions in this area. Increased quality also includes reduced medication errors that were expected as a result of both unit dose and computer order entry. One year after unit dose was introduced, an article in the Hospital Newsletter (October 1993) noted that errors and drug wastage were substantially reduced. The decrease is also related to the use of an integrated hospital system because Pharmacy has access to the most up-to-date information with respect to patient location and lab results.5 Consistent information also helps to reduce errors, such as entering allergies once and displaying them whenever medications are entered. There are also less transcription errors and interpretation of handwriting when using computer generated labels 5 98 and worksheets. However, Pharmacy still depends on the prompt action of Nursing Units in faxing down the original orders, which may result in errors related to delays in processing. Errors decreased initially, but Nursing has recently expressed a concern that too many errors in order entry are again occurring and translating into administration errors on the Nursing Unit. Pharmacy is currently monitoring the accuracy of order entry and the consequences of errors. Nurses are also expected to take responsibility for accuracy by double checking the M A R ' s at midnight to ensure medications have been entered correctly. However, the "visible" accountability remains with Pharmacy. The quality of medication therapy is related to the clinical role of pharmacists. Although Pharmacy does not have a formalized clinical program at Hospital 1, they are in the process of making changes in the department to free up pharmacists' time so they are able to be on the Nursing Units more, interacting with nurses and physicians. Pharmacists' clinical decision-making is facilitated by having a lot more information at their fingertips and having more confidence in that information. However, quality of the information retrieved is only as good as what is entered, as well as the individual user's ability to organize their material in the computer. If the information is not readily available from the pharmacy module, two report writers in the system provide enough flexibility to retrieve what is needed. Many pharmacists are quick to point out, however, that decision-making is still dependent on the pharmacist's individual skills and knowledge. Reports that facilitate clinical decision-making for pharmacists can now be generated automatically or on demand. For example, one clinical program that is now possible is the "IV to Oral Stepdown Therapy." Pharmacists monitor at least four different drugs that were 99 chosen because of their cost or possible toxicity. After three days of IV therapy with any of these drugs, Pharmacy runs a report to determine whether physicians have made this change. They generate other reports including chemotherapy patients who are not registered with the Cancer Agency and patients taking drugs which could potentially cause problems like renal failure or need dosage adjustments such as Gentamycin. This type of reporting capability is useful in other areas such as drug utilization review. One pharmacist is designated this responsibility and he routinely generates reports that identify the top number of drugs purchased, their costs, comparisons to previous year, and drugs ordered by physician. Outside of making an "educated guess," it was difficult to retrospectively collect this information in the manual system. The new information produced creates a "visible" accountability for physicians that did not exist before. It can be used in a general way to educate them about costs of different therapies as well as to modify individual ordering practices. The can also recommend lower cost alternatives to physicians who are using expensive drugs. Pharmacy is evaluating a new "satellite" approach to providing clinical services by having pharmacists assigned to specific Nursing Units. They make rounds to those units and enter medication orders directly into the system, the label prints in Pharmacy and the drug is sent up to the Nursing Unit. Use of a PCIS makes this concept feasible (level 3), as this pharmacist describes: It's a concept in Pharmacy that we would prefer to see, where the pharmacists are out in the satellites. They have more direct contact with who our customers are - the nursing staff the physicians, the patients, our fellow staff members. I think being right there in one-to-one contact is sort of the intent of moving to the satellites. The computer system in itself happens to be flexible enough to accommodate that. 100 One of the other clinical duties that pharmacists are engaged in is patient teaching. They provide patients with information about their medications while they are in hospital as well as producing customized patient teaching materials to assist patients in safely taking their medications at home (level 1). The very characteristics of computers that provide benefits of increased productivity and data quality, decision-support and access to information are the ones that also contribute to dissatisfaction, particularly for knowledge workers who are used to "relying on their brains." Pharmacists feel they are not required to think about activities and information entered because the computer does that for them by providing drug interaction and allergy checks, as these comments indicate: It's just the volume of work and you can process it twice as quick. But then it's got it's drawbacks too, because that's all you do is just processing. It's just like a cashier at the grocery store. You just push the products right through and sometimes that's all you feel like you're doing. You're not really using, quote/unquote your brains. Although you're still looking at the profile and checking for interactions, but you can't know the whole story by sitting at the computer... I enjoy having interactions with nurses and patients and doctors. You're kind of limited when you're the one person who is designated to sit in front of the computer and answer the phone. But, I think that's just a combination of the volume, the work and... The computer does it all for you so you don't need to go out and do all these things, which is convenient. But at the same time you probably miss out on it a little bit. Productivity increases in the sense that more orders can be processed with fewer staff. This happens partly because efficiency is accomplished by separating and streamlining tasks, thereby reducing the flexibility in each job. In other industries this is called "an intellectual assembly line." It can have a negative effect on job satisfaction as pharmacists see themselves sitting behind a computer terminal all day. The separation of tasks also moves 101 pharmacists away from the reality of patients, as the pharmacist entering orders may identify specific concerns, but another person is responsible for following them up. The effect of computer use on job satisfaction varies depending on the emphasis pharmacists place on the dispensing, distribution or clinical aspects of their jobs. Some pharmacists feel their prime responsibility is in dispensing and distribution functions, in other words, "to get the right drug to the right patient at the right time." Therefore, i f patients do not receive the right drug, there is no point in conducting clinical investigations to determine whether or not their kidney functions will be affected by the drug. However, other pharmacists feel their focus should be on clinical work, and expect technicians to be responsible for distribution. For these pharmacists, use of the computer, and an increased emphasis on order entry, may decrease job satisfaction. Since pharmacists rotate through all the duty areas, such as dispensary, IV preparation and clinical follow-up, job satisfaction on any one day may be more reflective of the match between philosophy and the task at hand, than use of the computer to do the task. Use of an integrated system introduces structure into the workday that is sometimes seen as the computer system "controlling their work." For example, certain tasks must be done at specified times, such as running the refill list and dispensing schedules, as this pharmacist explains: If it's 3:30 in the afternoon you have to be aware of the time and know there's a refill list that has to be run. You have to be a little bit more cognizant of the time of day and the things that the computer needs to have done in order to make the system work... If you don't run that list by 3:30, then you get the next day's workload coming off. ... This causes problems for the people who are working in that area the next day because the list is not up to date. 102 4.3.3 Use and Impact of Electronic Communication The messaging system is used extensively in Pharmacy. Communication has been a key system benefit because staff pharmacists and technicians work extended schedules and may be away for up to five days at a time. They are able to leave messages for someone they will not see for four or five days, which helps provide continuity in their patient follow-up or programs. While it takes away from personal communication, it is a little more reliable than messages passed by word of mouth. They also send messages to the entire group, which is also beneficial when staff work shifts. Pharmacists see benefits in using electronic communication between their department and Nursing Units because IT reduces phone calls that cause distractions and interruptions in their work. 4.3.4 Summary Impact of PCIS on pharmacists at Hospital 1 is summarized in Figure 4.7. Medication orders continue to be faxed from the Nursing Unit to Pharmacy, creating increased legibility problems, a time lag between when orders are written and processed and the potential for misdirected or lost faxes. This method of communication bypasses an accountability loop because there is no "paper trail" with respect to who sent and received the order, time it was sent, received or filled. Their trial of satellite pharmacies may alleviate some of this because pharmacists enter orders directly from the original order on the Nursing Unit. They expect to benefit primarily from the efficiencies gained through automating clerical tasks (label production, M A R ' s and medication profiles) resulting in reduced transcription errors, more legible and complete records. 103 Level 1-substitution Level 2-proceduralization Level 3-new capabilities Structure * automate manual *set parameters / * satellite pharmacy tasks auto generate >kbels reports >medication profile >workbad stats ^automatic trending u i . of drug utilization | transcription errors ^ expected ^ expected ^ expected Process + better, more + effect on physician ? effect on decision-timely decisions by decision-making making nurses & physicians >change drug ordering (not specifically practices defined) + more time for patients / clinical • ^ expected : expected Patient + benefits for patients |? benefits for patients Outcome >reduced medication • (not specifically errors • defined) >patient teaching materials i Figure 4.7 - Impact of PCIS on Pharmacists at Hospital 1 They expect these changes to benefit physicians in their decision-making and nurses in administration of medications. Patients should also benefit through decreased medication errors and complete, accurate patient education material. There are many expectations for reduced medication errors, partly based on the number of potential errors that could occur. Reductions may be more significant if many errors occurred before the system was installed. 104 Other than serious errors, there seems to be little in the way of documentation, but participants speculate that many errors occurred that were never detected. Integration of systems provides pharmacists with more information "at their fingertips." They are able to set parameters in a number of areas (such as drug use versus lab tests) and automatically generate reports that were too labor intensive and less systematic in a manual system. Patients are expected to benefit through avoidance of drug related renal problems, although the extent of this is unknown. Within pharmacy, having access to patient information, drug orders and associated costs provides opportunities to analyze drug use in new ways. A positive effect on physicians' ordering practices is expected to occur. The impact of behavior in one department on other departments is now more evident. The informal processes for dealing with inaccuracies or uncertainty are no longer available. Use of an integrated information system introduces formal interactions between departments that do not occur in a manual system. For example, when a nurse enters "drug allergies," these affect the pharmacist's order processing. When the pharmacist is unusre that an allergy exists and feels the patient's symptoms are an adverse reaction, he or she must stop the order process to verify this or override the warning. In other words, they can no longer ignore "errors" on the part of Nursing users and must come to some agreement on the meaning of the terms "allergy" versus "adverse reaction." The very nature of the expected benefits contributes to dissatisfaction for pharmacists. In many ways electronic communication between Pharmacy and other departments, as well as within the department has many benefits. The reduced phone calls decreases interruptions and distractions. There is increased accuracy and reliability in the information being entered 105 once and used by many different groups. It is produced in easy to read computer print. Efficiencies are gained by dividing up tasks and having one pharmacist assigned to entering orders while others follow up on problems. The downside of these benefits is that entering orders becomes very routinized "like the scanner at the grocery store." The follow up to problems becomes separated from identification of problems creating a "distance from reality." There is less face-to-face communication, although very efficient, is also very isolating. Differences between training and learning are evident in the expectation that quality of information is dependent in part on the individual user's ability to organize what is entered into the computer. "Just in time" training was recommended by pharmacists several years earlier, so computer training is broken up and matched with orientation for specific areas in the Pharmacy. Users discover new ways to use the system as they gain experience with it. 4.4 Impact on Physicians - Hospital 1 4.4.1 Introduction and History of Computerization in Medicine One physician, along with hospital Board Members, Administration and the Director of HIS, were involved in the early deliberations to replace a previous Accounting System, which was "at the end of its useful life." This committee also determined expectations for the replacement system and selected X T E C H . Although it took awhile to achieve their original goals, use of the system has now gone beyond what they envisioned. They expect the next changes will be "quantum leaps," and include introduction of an electronic chart (level 3). Advantages in cost savings and efficiency are anticipated, as well as benefits from 106 a medical/legal point of view. Advice offered to physicians in other hospitals is to become involved early on, as this candid comment illustrates: I pointed this out to colleagues of mine who complain their computer systems aren't helpful to them, and their hospitals don't seem to have any interest in physicians. They have to get involved early in the process. Even though some of these committee meetings are pretty technical, financially oriented and boring, you need a physician in there to remind them physicians have to be involved. There has to be something in it for the physicians if it's going to work. There is no designated physician liaison for the medical staff, although this role is filled to a certain degree by the Clinical Information Systems Coordinator (who has a strong clinical background), as well as other HIS staff who provide training and support. The HIS department initiated training classes and "physicians were invited to come along." The physicians who were interested came to the early classes, and those who were not interested eventually came because they found they were not able to do their work without using the computer. A training program provided by the HIS staff is now available for new physicians coming on staff and provides a one hour introduction to the system. However, this may not be the answer to training as one physician remarked: "You get better use out of the system if somebody sits down with you and shows you some of the finer points of what it can do." Physicians expressed resistance to using the system in a number of ways. Initially some unique reasons were offered for why physicians should not use the system, as this humorous example illustrates: Some physicians are somewhat conservative, [laughing] It was amusing... to get our messages, before we had the computer system, when we came to the hospital there was a little old black telephone that you would dial in your code. My code was 28, I think. So, in the morning when I came in I would dial two, eight. If there was a message waiting for me at switchboard, a little light would flash. I would pick up the phone and say, "It's Dr. Bildoff." They would go through their scraps of papers and say, "Oh yes, you've had 107 somebody admitted." That was fine, but with the new system, you type in your password and it gives you the message. So, some physicians counted the strokes that were needed. In other words, you only had to do 2, 8 and lift before; now you have to do B-I-L-D-Return. That's five, and therefore a backwards move. That was the sort of [laughing] resistance that there was, but that changed gradually. From the physicians' point of view, resistance also relates to the "opportunity costs" of spending money on computers versus clinical programs or equipment, as these two examples illustrate: For as long as I've been in this hospital, there have always been budgetary restraints. It's just a way of life in hospitals. Administration is saying you have to cut back and from our point of view, money always seems to be cut from clinical programs. If you're a cynical physician sitting there, and you see HIS spending another million dollars and another five million dollars on their computer, you might really wonder whether that money [could be] better spent in the best interest of your patients. So, that is probably part of where the resistance comes from. I remember going to an MAC [Medical Advisory Committee] meeting and trying to defend the initial investment in the system. At the time, I think we were trying to get our first CT Body Scanner. That was the question somebody asked: "Should we spend money on the computer or should we get a CT scanner? " So I expected resistance, but no more or no less than we got. 4.4.2 Use and Impact of PCIS Up to this point, physicians have been able to sign in and out, retrieve a current listing of their patients and receive consultation requests without using a password. Major changes in the messaging system are being implemented in the near future, and a new hospital policy goes into effect, which will require use of their passwords for everything, including electronic mail. Physicians use the PCIS inquiry function primarily for accessing patient information such as lab and radiology results (which has always required a password). Several months 108 prior to the interviews, a two week audit trail revealed only about thirty-five to forty percent of physicians were accessing the PCIS daily for the patient information they needed (seventy to eighty out of over two hundred physicians). Results of a new audit received by this author approximately nine months after the interviews show this has increased to fifty-five percent (one hundred and ten out of two hundred) of physicians who are using their passwords routinely to access their own patient information. The effect of the computer on how physicians practice medicine seems to vary depending on the type of practice. It introduced minimal changes for General Practitioners. In areas where patient turnaround time is very quick, such as Short Stay, Same Day Admission and Day Care, physicians were also not as likely to use the system as those whose patients had a longer stay. Specialists began to get their consultation requests through the computer, which was expected to speed up this communication process. The descriptions of how two physicians start their day illustrate the range of systems use. These two participants could be described as a high user (a specialist) and a moderate user (a General Practitioner): We have the XTECH package and it allows us to highlight certain clinical parameters that we want to follow (level 2). My day usually starts by printing the computer generated patient list and then going through each individual patient, looking at their clinical highlights. That's before I make any patient contact in the wards or with any of my consultations, or what have you. I find it the most efficient and fastest way to get right on top of what's going on, at least from a laboratory sense. The computer is quite good at flagging new things that have been done as well, so if my patients have had investigations ordered by other physicians, I'm usually on top of that. In the hospital I use it every morning, as everybody does I think. When I come in I sign on, get all my messages (level 1) and that tells me any patients that have been admitted. Sometimes there are requests from nursing stations about my patients and they want me to do something particular or tell me something about the patient. That's on a daily basis. Frequently, but not necessarily daily, I use it to access lab results, X-ray results. I also use it for E-mail to other physicians and other non-physicians in the hospital. 109 Most of the medical staff