International Journal of Medical Informatics 69 (2003) 235 /250 www.elsevier.com/locate/ijmedinf
Implementing computerized physician order entry: the importance of special people Joan S. Ash *, P. Zoe¨ Stavri, Richard Dykstra, Lara Fournier Division of Medical Informatics and Outcomes Research, School of Medicine, Oregon Health & Science University, 3181 SW Sam Jackson Park Road, Portland, OR 97201-3098, USA
Abstract Objective: To articulate important lessons learned during a study to identify success factors for implementing computerized physician order entry (CPOE) in inpatient and outpatient settings. Design: Qualitative study by a multidisciplinary team using data from observation, focus groups, and both formal and informal interviews. Data were analyzed using a grounded approach to develop a taxonomy of patterns and themes from the transcripts and field notes. Results: The theme we call Special People is explored here in detail. A taxonomy of types of Special People includes administrative leaders, clinical leaders (champions, opinion leaders, and curmudgeons), and bridgers or support staff who interface directly with users. Conclusion: The recognition and nurturing of Special People should be among the highest priorities of those implementing computerized physician order entry. Their education and training must be a goal of teaching programs in health administration and medical informatics. # 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Point-of-care systems; Focus groups; Interviews; Anthropology, cultural; Medical records systems; Computerized
1. Introduction Computerized physician order entry (CPOE) is a process that allows a physician to use a computer to directly enter medical orders. Although CPOE is recommended as one mechanism for reducing medical errors * Corresponding author. Tel.: /1-503-494-4540; fax: /1503-494-4551. E-mail address:
[email protected] (J.S. Ash).
[1], a survey of US hospitals discovered that CPOE reportedly exists in one-third of hospitals, but is minimally used [2]. The overall purpose of this study is to identify perceptions of CPOE held by diverse professionals at sites where CPOE has been successfully implemented. These professional groups include clinicians, administrators, and information technology personnel. Successful implementation is defined as heavy use (over 80% of orders are entered electronically) by a
1386-5056/02/$ - see front matter # 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S 1 3 8 6 - 5 0 5 6 ( 0 2 ) 0 0 1 0 7 - 7
236
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
large number of physician users. We are studying successful sites because they serve as models. Although there are a number of successful sites with ‘homegrown’ systems developed internally, this study focuses on ‘off the shelf’ systems which have the potential for being widely adopted in multiple sites. One especially important result of the study* that special people are important to a successful implementation* will be further explored in this paper. /
/
1.1. Background As part of our research about success factors for implementing CPOE, we have identified a taxonomy of ten major themes [3]. We have continued to collect and analyze more data in an attempt to validate or invalidate those themes and to further develop them. Qualitative research is iterative in nature, and research questions are continuously refined as new findings and additional research literature help to crystallize initial research questions [4]. A recent reading of the late Diana Forsythe’s work reminded us of a powerful challenge she issued to the informatics community in 1996. In particular, the posthumous publication of an anthology of her work [5] instigated our re-reading of her classic paper titled ‘New Bottles, Old Wine: Hidden Cultural Assumptions in a Computerized Explanation System for Migraine Suffers’ [6]. This paper describes ‘the problem of user acceptance’ [6] so common with informatics applications. It points out that we often blame the users for not embracing new systems, yet a system may ‘‘embody perspectives that may not be meaningful to or appropriate for their intended users’’ [6]. We had described the ‘multiple perspectives’ of administrative, clinical, and technical staff
in hospitals regarding CPOE in a paper in 2000 and reached the same conclusion [7]. In our previous work [3], we identified three assumptions concerning POE: . Order entry is a linear process, beginning with the physician entering an order and ending with its being carried out. . Physicians are recalcitrant: they resist behavior change. . Structured input of data for orders is good for two major reasons: first, for legibility, and second, for analysis purposes. These assumptions are the ‘old wine’ put into the ‘new bottle,’ CPOE. Diana Forsythe asked: ‘‘Whose assumptions and whose point of view are inscribed in the design of technical systems? Who will benefit from adoption of a given system, and who stands to lose?’’ ([6], p.118). We had discovered that CPOE is not a linear process: the order communication process is, in fact, exceedingly non-linear, with many people involved in formulating the idea of the order, modifying it throughout the process, and carrying it out and documenting it. We found that physicians are not necessarily recalcitrant: they simply do not wish to spend additional time during the ordering process, time they would prefer spending with patients. Finally, structured input benefits administration and it fits the capabilities of computer systems, but it does not necessarily benefit the physicians, who prefer entering or writing free text. The Forsythe paper ends with the question ‘‘who will monitor the hidden cultural assumptions built into computerized tools for medicine?’’ [6]. The purpose of our research was to identify success factors for implementing CPOE and this is a relevant question, given that CPOE is a computerized tool. Now, it seemed, we had enough data to begin
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
answering the challenge: Who will monitor the hidden cultural assumptions embodied in CPOE?
2. Methods 2.1. Selection of sites Experts in informatics helped us identify successful CPOE sites. Based on geography, type of organization, ownership, and date of installation, we selected: (1) on the East Coast, the University of Virginia, which has a teaching hospital, is state supported, and implemented the Technicon/Eclypsis system (Eclypsis Corp., Delray Beach, FL) in 1989 [8]; (2) on the West Coast, the Veterans Affairs Puget Sound Health Care System campuses in Seattle and American Lake, which were test sites for the VA’s Computerized Patient Record System (CPRS), beginning in December, 1998 [9]; and, also on the West Coast, El Camino Hospital in Mountain View, California which, beginning in 1966, became a development site for a medical information system originally developed by Lockheed and purchased by Technicon (now Eclypsis) [10]. 2.2. Data collection methods We chose to use observation, oral history interviews, and focus groups in an iterative process over a 5-year period. Participant observation, an unobtrusive method which ‘‘produces detailed descriptive accounts of what (is) going on (including verbal interaction)’’ [11], has been used effectively in prior informatics studies [12,13]. Researchers accompanied clinicians in the course of typical daily tasks in the hospital and conducted informal interviews. Informal interviews took place when informants were being ob-
237
served and there were breaks in the normal routine. They generally took 10 15 min and were sometimes tape recorded. Data were also collected using focus groups, which provide an efficient way to gather information from a group with the benefit of synergy among participants [14]. In addition, formal taperecorded oral history interviews of representative informants provided the opportunity to ask more open-ended questions [15]. These were scheduled, taped interviews of approximately 1 h in duration, during which the interviewee was asked to recall and describe past events. /
2.3. The data Fieldwork at the University of Virginia was done in 1998. We shadowed two internresident pairs in the critical care unit and in labor and delivery, a surgical intern, and a medical intern, for a total of six physicians. Because we visited in August, the interns had only 1 month’s experience with CPOE. Informal interviews were held with each of the six physicians plus five nurses. Formal oral history interviews were held with eight individuals: the chief clinical information officer, a clinical systems specialist with a nursing background, two administrators who are physicians, two faculty who used the system as residents, and two other physician faculty. Two focus groups were held, one with about ten pediatrics residents and the other with about 20 general medicine residents. At the VA, in our initial visit, we observed an attending in an outpatient primary care clinic, an attending in a nursing home unit, a psychiatrist at American Lake and a medical student/intern/resident team on a general medicine unit in November of 1998. Informal interviews were held with each of these six individuals plus four nurses and a clinical pharmacist. One focus group was held with a
238
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
ward team including a medical student, an intern, and three residents. During our second and third visits, we shadowed clinicians in inpatient settings representing a wide range of acuity, and continued data gathering in outpatient settings, adding data from 180 h of observation, 23 formal interviews, and three focus groups to the total. At El Camino, during the first visit in August of 1999, we shadowed physicians in the emergency room, medicine, oncology, the ICU, and CCU. Informal interviews were held with nurses, administrators, a care coordinator, clinical pharmacists, and a dietician. We conducted four formal oral history interviews with a physician, two nurses who practiced nursing when the system was implemented and subsequently worked for the information technology unit (now outsourced to IBM), and a person who had helped with the original cost evaluation of the system. We visited a second time in August of 2000 and observed nurses on the post-surgical floor, the emergency room, and the CCU. We shadowed a gastroenterologist, a radiologist, and two hospitalists. We also observed a CPOE training session and we shadowed individuals in the clinical laboratory, pharmacy, and radiology. We did informal but taped interviews with 13 individuals. We did formal taped oral history interviews with three physicians, three administrators, two information technology people, four nurses, and a medical records administrator. In all, we did a total of 340 person-hours of observation and informal interviewing and 59 h of formal interviewing. Table 1 summarizes the method and subjects at each site. The last three columns show the numbers of clinical, information technology, and administrative sources of data, each important for gaining multiple perspectives concerning CPOE. The clinical column aggregates two kinds of numbers: (1) the number of individual clinicians
shadowed for periods of at least 4 h, and (2) the number of units (like nursing stations) observed for at least that period of time (when the observer was watching all activities rather than one informant). 2.4. Data analysis Field notes were transcribed from the handwritten notes by the investigators themselves. Transcripts of interview and focus group sessions were produced from the audiotapes by experienced oral history transcriptionists. When put into a format suitable for analysis, the total number of pages was 1825. A grounded theory approach was used to identify emergent themes. Grounded means ‘based on and connected to the context-dependent observations and perceptions of the social scene’ [16]. Rather than starting with a predetermined a priori list of code words, the informants’ own words guided code development. For the data prior to the second El Camino visit, two researchers (JSA and ML) independently reviewed the field notes and transcripts, annotating important themes in the margins. Qualitative data analysis software (QSR NUD*IST 4, Sage Publications, Berkeley, CA) assisted the researchers in reviewing and indexing these patterns and themes. The researchers worked together to name the patterns and themes, thus creating a taxonomy. Two other researchers (PNG and JL) reviewed the documentation separately and noted major themes. The team of four met three times to discuss the taxonomy. Four high level themes were identified: (1) organizational issues such as power, politics and control, collaboration, pride, and culture; (2) clinical/professional issues such as local development and customization; (3) technical and implementation issues including usability, order sets, training and support, and time; and (4) issues related to the organization of
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
239
Table 1 Methods and participant groups Site
Method
Clinicians or units studied
University of Virginia Observation Informal interview Formal interview Focus group
6 5 4 2
Veterans Administration Observation Informal interview Formal interview Focus group
57 5 6 3
El Camino Observation Informal interview Formal interview
18 13 9
information and knowledge. Since there were a number of unanswered questions and foci identified in that analysis, a second visit to El Camino was planned to concentrate on these issues. This is typical of the iterative nature of qualitative research. In addition to outstanding issues identified by the entire team, individual foci for each of the field investigators were openly identified. Because each person came from a different professional background and we wanted to assure that we took advantage of this diversity, we added special assignments to each person’s observational work. For example, the nurse on the team made a special effort to observe different nursing shifts. The second El Camino visit yielded so much data that it was analyzed separately. Researchers (JSA, PG, ML, LF, and often JC) met 18 times to discuss selected documents they had coded. Paragraph by paragraph each was discussed, with coding of all researchers recorded and discussed. Once a sense of saturation was reached with repetition and validation of prior themes occurring frequently, the meetings gave way to thorough
IT professionals studied
Administrators studied
2
2
2 4
1 7
1 4
5
reading of remaining material. After one researcher (ML) completed the coding using NUD*IST, the full team met to agree on names for themes and sub-themes, with the result that the original four themes expanded to become ten as the El Camino data were blended with prior data. Finally, an objective outsider who was uninvolved in data gathering (ZS) independently coded all of the final El Camino data and reviewed prior data as a final check and validation of the ultimate themes taxonomy. Data from the two final visits to the Puget Sound Seattle hospital and from a second American Lake visit were analyzed using N5, a newer version of the QSR software. These data were blended with prior data. The research team (JSA, PG, LF, RD, and VS) met 33 times to discuss the new data in light of the ten themes. The themes were further validated, and we were able to explore them in greater depth. Finally, our outside participant (ZS), who knew more about the data now but who still had not collected any of it, reviewed the new transcripts and added her insight to help confirm the results.
240
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
3. Results A taxonomy of ten high level themes was developed, including (1) separating CPOE from other processes, (2) terms, concepts, and connotations, (3) context, (4) tradeoffs, (5) conflicts and contradictions, (6) collaboration and trust, (7) leaders and bridgers, (8) the organization of information, (9) the ongoing nature of implementation, and (10) temporal concerns. Theme 7, leaders and bridgers, might more succinctly be called Special People, and it is explored further here. Described below are patterns that were seen across all of the sites. The quotes we have selected are representative of the sub-theme and are especially well articulated. Because individuals were promised confidentiality, an effort has been made to report what was said without identifying the site. 3.1. Special people theme When asked ‘‘what made your implementation successful?’’ many interviewees answered that a certain individual was key and went on to describe that individual’s attributes and actions in detail. Special people were highlevel leaders, non-physician clinicians who assisted with the implementation, or physicians who played a special role during implementation. Their roles spanned disciplines from administration to information technology to the clinical realm. Because they lived in more than one world and knew the vocabulary of each, they could interpret from one to the other. In addition to vocabulary, they could interpret disciplinary culture as well. Fig. 1 depicts how these special people have overlapping, intersecting roles, sometimes holding an administrative role or a paid position in informatics and sometimes practicing another clinical specialty such as pharmacy while providing technology support.
Fig. 1. Overlapping roles of special people (administrative, information technology, and clinical staff).
Each has a view of CPOE that is colored by his or her role or combination of roles. We divided the roles into three major levels: the leadership level, including the chief executive officer (CEO), chief information officer (CIO), and chief medical information officer (CMIO); the clinician level, including champions, opinion leaders, curmudgeons, and the clinical advisory committees; and the bridger/ support level, which includes those who do training and support and interact directly with users (we call these people bridgers because they translate user needs to the higher levels and vice versa). How are these special people alike? Their behavior as interpreters between the technology and clinical worlds is key. Even the CEOs who were at these successful sites during implementation of CPOE exhibited the ability to understand the clinicians and their behavior. All held a vision of the future and the way CPOE fits into the electronic medical record. Many personal attributes were universal, including stability through adversity, steadfastness, initiative, and toughness.
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
3.2. Administrative leadership level We include the CEO, CIO, and CMIO here. The CEO administers the entire organization and information technology and clinical leadership report to him or her. The CIO generally heads information technology and often information services units and reports to the CEO. The CMIO is also an administrator but has clinical information as a focus and reports to one of the former levels and/or a Chief Medical Officer. Administrators and clinicians often overlap roles, as Fig. 1 also indicates. The CEO or the CIO may have a clinical background. The CIO may be part of the administrative team and at the same time head IT. Leaders also include individuals who serve in a Chief Medical Information Officer role. This person usually plays roles in administration and IT, has a clinical background, and may even continue to practice medicine. These people may hold other titles as well, but the role is one that is somewhere between the CEO and the users of CPOE. 3.2.1. Administrative leadership level: the CEO Leadership and top-level support were usually cited among the most important success factors for implementing CPOE. We define the CEO as the individual who heads the entire organization, even if the person has a different but equivalent title. This person is depicted in Fig. 1 as Admin, or Administration. The following sections describe the behavior of CEO’s at successful sites. 3.2.1.1. Provides top level support and vision. ‘‘What made a huge difference for us is that we had the unwavering backing of our director [CEO] and then the people who were on the next level [CMIO] also, you know, [they said] this was going to be done.’’ At another site, an informant said of the CEO: ‘they
241
realized the importance of keeping it going and you know, when they said Oh we want to try [various alternatives], they said ‘Good, don’t, you know, just say tell us how much it’s gonna cost: do what you have to do and give them the help and the support that they need to keep it going.’ ’’ Of course, financial support is key. One CEO said: ‘‘hire really good people, give them the tools, and stay out of their way. I mean, give them support... We had to put a lot of money into the infrastructure. We had the design done, and then I just devoted the dollars that it would take. I took equipment dollars, the construction, the nonrecurring maintenance dollars, and, without hampering patient care, devoted the lion’s share of those for a couple of years into that.’’ 3.2.1.2. Holds steadfast. One informant said ‘‘What kept us going here during the tough times is the fact that our administration... said [to IT], ‘You know, guys, do whatever you can’; they realized the importance of keeping it going, just tell how much it’s gonna cost, and do what you have to do.’ ’’ 3.2.1.3. Connects with the staff. ‘‘He was here over 25 years and politically astute and more of a team builder and got people working together.’’ From field notes, ‘‘I was struck by how much X is like the legend of Ed Hawkins at El Camino. People seem to see him in the same way. He also goes out on to the floors and considers himself one of them.’’ ‘‘He actually knew most of the employees by their first names, and he was very genuine. He would come around to the departments, and he was very open. You could go up and see him.’’ 3.2.1.4. Listens. ‘‘They listened to nursing carefully in developing the system.’’ Participation, not buy-in, was solicited. A bidirectional flow of communication is important. These
242
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
CEO’s really listened and accepted help from users in planning for and implementing the system. ‘‘You get louder because you think people aren’t listening.’’ 3.2.1.5. Champions. Champion is defined here as someone who fights for the cause, e.g. implementing CPOE. The CEO is often champion. One CEO said ‘‘you really have to be a cheerleader and an advocate as well as a decision maker.’’ 3.2.2. Administrative leadership level: the CIO The CIO often reports to the CEO and is usually directly responsible for CPOE implementation. Our informants told us that this is a position that requires political skills and a thick skin. 3.2.2.1. Selects champions. Successful CIO’s know how to select clinical leaders: ‘‘X was a master at this, was finding the leaders and giving them a lot of special attention, helping them understand how it could help them, like with personal order sets, and then this leader does say this saves me time and it makes the whole process more efficient and it saves me getting a phone call.’’ 3.2.2.2. Gains support. ‘‘He did have a lot of political moxie and he was very determined and he grew his department, it got bigger and bigger.’’ 3.2.2.3. Possesses vision. Like the CEO, the CIO has a vision: ‘‘He was pretty goal oriented, he would see a goal and work towards it regardless of what was in the way.’’ 3.2.2.4. Maintains a thick skin. As the person directly responsible for POE, the CIO may also get the blame for shortcomings: ‘‘X was embattled much of the time was my impres-
sion, I don’t know if he’d say that, because he was constantly in the role of defending it.’’ 3.2.3. Administrative leadership level: the CMIO The CMIO is called different things at different hospitals, but each had one or two people in this role. These physicians straddle the line between medicine and administration (see Fig. 1, where they are labeled ‘Clin’). They have many of the same attributes as the CEO and CIO. 3.2.3.1. Interprets. They understand the business of the hospital and try to interpret it for their clinical colleagues, and they attempt to explain the culture of medicine to administrators. 3.2.3.2. Possesses vision. ’’We were very very fortunate to have Dr X who was able to hold the bigger vision of this project, because I’d seen other facilities who haven’t had that luxury or didn’t hire ten [clinical support staff] to bring it up.’’ ‘‘I think, for the leaders, you need to have a firm vision and a team that can help you get there, and to stay the course.’’ 3.2.3.3. Maintains a thick skin. ‘‘Make the corrections that are necessary, but you* part of being a leader is also taking the input that you get and, but not necessarily abandoning when things get tough.’’ ‘‘Dr X was a much hated man, but a nice guy.’’ /
3.2.3.4. Influences peers. About a CMIO from a support staff member who had trouble dealing with a physician: ‘‘It just became clear that I wasn’t going to be enough... so I had [the CMIO] work with her [the physician], initially. And, given his very calming personality and the fact that he was relating to her physician-to-physician, that helped a lot.’’ A CEO said of his CMIO: ‘‘ he has the clinical
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
relevance to the others because he works beside them and that was the kind of person I needed was somebody that had credibility and good strong skills and a real fire to make an electronic record work for us.’’ 3.2.3.5. Supports the clinical support staff. While the CMIO needs the support of the CEO and the CIO, he or she likewise needs to support the critical support staff members who deal directly with the users. ‘‘They [the CMIO and those above] kept telling us that we were important, and whenever we needed to have a work-life balance, they were supportive of that, because they wanted us to feel valued.’’ 3.2.3.6. Champions. As one user noted, the CMIO is always a champion: ‘‘one part of that [CMIO] role is being a champion.’’ 3.3. Clinical leadership level Successful sites put resources into identifying and sometimes even hiring physicians and other health care professionals to socialize the idea of CPOE throughout the organization. We identified four different types of leaders in this category. 3.3.1. Clinical leadership level: champions We define champions as individuals who fight for the cause, who believe in CPOE, its importance as part of the EMR, and in the value of technology for achieving goals [17]. The champion does not necessarily need to be someone with advanced technical or clinical skills. Champions are sometimes looked on skeptically by the majority because they are early adopters and perhaps have such a passion that their view does not appear to be balanced. For example, about one champion, it was said in field notes that it was ‘‘hard to get him to talk about weaknesses [of CPOE].’’
243
3.3.1.1. Need. A leader said, ‘‘certainly I would want to have some clinical champions identified before taking on anything of this size, any project.’’ One CEO, when asked how he would do it differently next time, replied ‘‘I would get the clinical champions in place earlier.’’ ‘‘I think you need a number of clinical champions within the areas, the clinical areas, separate from the clinical informatics group.’’ ‘‘Very important to deal with* at the outset to deal with people who would buy in or are going to have the insight that this is a change for the better, that once you do it you’re not going to regret it, in fact, you’re going to love it. If you get* if you identify these people* we called them champions, and I was one of them, these become the agents that are going to sell this change among their peers.’’ /
/
/
3.3.1.2. Holds steadfast. One informant told of a clinical champion who failed because he had an interest but not a ‘fire in the belly for technology.’ To succeed, the champion needs to be persistent: ‘‘The head of the physician group down there... was very positive and he’s fairly soft spoken but he doesn’t go away, he just keeps saying ‘we need to do this.’ ’’ 3.3.1.3. Influences peers. Sometimes the champions were recruited to help with training on a one-on-one basis: ‘‘if you were unsure of what to do, a doctor would sit next to you and you weren’t made to feel that you were incompetent. There was a lot of one-on-one.’’ 3.3.1.4. Understands other physicians. One of the champions expressed the feelings of his peers about response time: ‘‘did somebody talk to you about response time, how critical it is? That’s in my experience the key in getting doctors to use anything, because we’re all Type A... there were slowdowns and God, it was terrible, but doctors are very impatient,
244
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
and they can screw it, they’ll say the h___with this thing.’’ 3.3.2. Clinical leadership level: opinion leaders An opinion leader is an influential physician who may be in favor of the new system or work against it. Opinion leaders are respected by peers, usually for both professional and social skills [18]. 3.3.2.1. Provides a balanced view. ‘‘They were just people with common sense, I guess. You know there’re a lot of different personalities and I would guess if I were to try to pigeonhole them, I would consider them for the most part to be more levelheaded, even tempered, you know’’. 3.3.2.2. Influences peers. ‘‘And using peers to influence peers to go through the change, rather than giving a directive [that] everybody should do it’’ is a good idea. ‘‘Dr X worked closely with [the CIO]... he liked computers, but he kept his practice and did this also. He had a large contribution, not to the initial buy-in but later. He was very active, he was a surgeon, a lot of activity happens around surgery lounges.’’ ‘‘The hospital paid him a salary, he was interested, quality assurance guy for awhile, he had that ability to lead and get doctors together, but he had help from the hospital to do paperwork, etc. very respected on staff, he was a neutral guy, kind of in the middle.’’ 3.3.3. Clinical leadership level: curmudgeons We define a curmudgeon as a skeptic who is usually quite vocal in his or her disdain of the system. Our informants felt that convincing the curmudgeon is a key to system implementation success because the loudest skeptics may turn into the staunchest supporters.
3.3.3.1. Provides feedback. ‘‘I mean, the person who screamed the loudest was a X fellow. She was a nightmare, frankly, but she also made the biggest changes and helped us learn what we needed to do to make the order sets work.’’ 3.3.3.2. Furnishes leadership. ‘‘You can become a physician leader by being an irritant, screaming and yelling every time anything happens, you get acknowledged and catered to, you can be a skilled practitioner who’s respected by peers and be charismatic. You can also have political skills where the others who don’t will come to you... they [the CMIO and support staff] targeted them [the curmudgeons].’’ 3.3.4. Clinical leadership level: the clinical systems committees We include within this category the committee of physicians and other clinicians charged with oversight of CPOE. It is a hospital committee of Special People and its work was often described as an important success factor by our informants. This committee has different names at each hospital, but it is generally an advisory committee that includes clinician members focused on CPOE alone or on clinical systems in general. At each of these successful sites, this committee was given serious responsibility for making recommendations to be followed by the leadership. 3.3.4.1. Solves problems. The Committee serves an important problem solving role: ‘‘if there is a problem, then it’s up to the [clinical systems committee] to deal with it.’’ ‘‘Probably 20 people got involved to try to work towards solving the problem.’’ 3.3.4.2. Connects units. Another role the committee plays is that of fostering teamwork.
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
Generally, different professions are represented, including nursing, dietary, medical records, and others, and, as one informant said about this multidisciplinary committee, ‘‘we’d get them together in a room...where decisions were made.’’ A committee member said about a CIO: ‘‘he wanted a group of clinicians to serve as an oversight and he didn’t want it to be senior management because he knew management fairly rarely practice and so they sort of miss something, but they don’t have a lot of credibility n the trenches unfortunately and they also don’t understand a lot of the issues. So we, we got a bunch of people sorta at my level in their careers who are still practicing and spend a lot of time on the wards.’’ An information systems committee member and an opinion leader at a different site said: ‘‘I am on that committee, and [the] other [committee I am on] is the organizational committee which is much more of a business type of approach. I basically sit and listen or represent the viewpoint of perhaps a fair number of physicians that have moderate... a modest functional use of any type of computer.’’ 3.4. Bridgers/support staff level At each successful site, there was a cadre of special people* clinicians bridging the gap between the information technology group and other clinicians. These are ‘bridgers’ or ‘translators.’ One site calls them ‘clinical application coordinators.’ Working at a grassroots level, they do more than champion the system: they work for results. These are people who understand the work, are paid to train and assist physicians in using POE, and make changes in the system. Although their personalities differ, they share a respect for physicians, a belief in the benefits of CPOE, and a value system in which the patients come first and the users next. Most often, those who /
245
train and support CPOE are non-physician clinicians. Informants in this category came from nursing, pharmacy, and physical therapy backgrounds. These are the people who ‘bend over backwards’ to help, and in so doing make the difference between success and failure. 3.4.1. Need One CEO said: ‘‘we made a decision that we needed to get enough resources into the support activities and the training activities to make this a success.’’ ‘‘It made no sense to put the organization through the pain of going from a paper record to an electronic record if we didn’t have the support in place to make it a success. I had seen that when we even wanted to take a baby step and hadn’t put the resources there, it failed.’’ 3.4.2. They provide help at the elbow All of the successful sites had help available during implementation all day every day. All continue to have valued assistance easily available. ‘‘[Betty] was the key person to help with* she went to other units to facilitate implementation.’’ ‘‘I think [Betty] was the difference.’’ ‘‘We had to track and we knew exactly who these people were that weren’t using the system and we sat down and helped them make their own personal order sets and we just sat down and held their hand.’’ These were also people who were known in the organization: ‘‘Another thing that I think helped with implementing the system was that X and her staff were here so long and already respected so that when they went out selling this thing, when people had problems and called, they knew who they were talking to.’’ /
3.4.3. They make changes Bridgers generally interface not only with the users, but also with the vendor. ‘‘The
246
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
vendor did all the coding and our job was to describe problems to the vendor and work with them and beat them over the head. Gradually we took over our own coding.’’ There are always tweaks that need to be made, even at the site with the world’s longest CPOE use: ‘‘We have a very good staff, they’re very technical, and we do routine changes every week* we try to do it on Wednesdays* and we use the committees that we have for approvals.’’ Changes are suggested only if users feel free to request them: ‘‘If you get your users involved, they won’t be hesitant to call you and say, you know, do you think this would be a good idea, or, if there’s a problem, to fix it...the user out there that uses it all the time says, Well, why didn’t you kind of do it a little bit differently; so we strongly encourage people to tell us.’’ /
/
3.4.4. They provide training These bridgers are also the people responsible for most training, either in formal sessions or informal one-on-one appointments. At one site, after the system was somewhat established, ‘‘we were sort of onsite all over the place doing you know, kind of point-of-care training, instead of classroom training.’’ ‘‘When we brought the inpatient up, we timed it so we would get a new crop of residents so that we would do that 1-h training with that new crop of residents.’’ ‘‘So we’re all expected to be facile in resident training.’’ ‘‘I can tell you why I became a (support person). I like to teach, and there’s a tremendous amount of teaching that goes along with the position.’’ 3.4.5. They test the systems ‘‘When any new change comes in, it’s put into the test account, and the (support staff) and others who are designated testers hammer on it in the test account.’’
3.5. Skills and training 3.5.1. They possess clinical backgrounds Most informants felt strongly that bridgers needed to have clinical backgrounds. The position ‘‘does require quite a bit of knowledge of the medical world. You have to know what you’re working with.’’ From another clinical support person: ‘‘I can’t understand how anybody can do this kind of thing without having a medical background. We do have some very highly technical people here that don’t have a medical background, and we always have to explain to them, you know, this is urgent, and this will never work this way because that’s not the way they think.’’ These bridgers often need to ‘‘go in and reenter orders,’’ so they need to be licensed to do so. ‘‘Some of the examples would be the difference between having a clinician have to completely rewrite a set of orders because of some problem and having a clinical (support person) who understands nuances being able to resurrect them or rescue the orders and avoid having to rewrite it.’’ One said ‘‘in this job you have to have been a user, you have to have a medical background of some sort.’’ 3.5.1.1. They gain skills on the job. None of the bridgers we interviewed said that they had previous formal training when they were hired into these positions. ‘‘We’re making it happen as we go along, we really are... I remember when I first started here I kept asking What kind of courses can I take? There are no courses that you can take for medical information systems, there really aren’t... sometimes what they (the users) want is not necessarily what they need, so you need to know what it is, to have been there, you know, to know what it is they’re talking about.’’ Another said ‘‘the learning curve was* oh, it was the steepest learning curve I’ve ever /
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
experienced.’’ ‘‘I had sort of grown into this local expert role on the computer... X really took me under her wing and trained me... when I interviewed, I said I don’t have the computer skills; I have a very good working knowledge of the system, I know the staff, I know ordering, I know patient care, I know presentation communication, I have all those skills.’’ Another said: ‘‘the main question that X asked me when I interviewed was how did I diffuse anger, because people get very angry. I don’t know that I fully understand why, except that it might have something to do with the loss of control, and people who are used to sort of being captains of the ship being in situations where they don’t really know or understand, that’ very frightening to them.’’ In fact, one observer noted ‘‘I was struck by the calm and patient demeanor of the (support staff). The nurse practitioner was clearly at the end of her fuse, expressing her frustration at every opportunity.’’ 3.5.1.2. They show patience, tenacity, and assertiveness. Patience, assertiveness and tenacity were high on the list of attributes cited as necessary: One of the support staff said ‘‘And once I snag physicians here* and sometimes they’re resistant to this, and I say, look, give me an hour. Then they want to know just a little bit more. And I said, it’s a good investment of your time. Trust me.’’ Initiative was cited numerous time as a needed attribute. We were told that one staff member who lacked initiative did not survive: ‘‘X really didn’t take a lot of that initial initiative that one had to do to learn how to be a [clinical support person]. He left shortly thereafter.’’ Unfortunately, burnout is also a factor among these staff members, although many were also long timers: ‘‘that’s a very big service, very diverse and difficult for one person to handle alone. So I think there was a burnout factor there.’’ /
247
4. Discussion Three categories of Special People have been identified as important and necessary if a CPOE implementation process is to be successful. The sites studied are successful in that CPOE is routinely used by most physicians. The kinds of people described above were present at each site and the quotes are representative of many strong statements made by interviewees and numerous descriptions in field notes. Lorenzi and Riley provide general insight into why Special People are needed: they refer to the ‘the cast of characters [19]. They reason that people in different stakeholder categories play different roles in the change management process. They warn that identification of individuals and their roles is important so that a variety of people can be involved from the beginning. Patel and Kaufman offer a reason based on cognitive science: they state ‘there is a need for bridging disciplines to enable clinicians to benefit from rapid technologic advances’ [20]. They point out that any discipline like informatics, which is made up of people from different backgrounds, needs to find an effective way to communicate, since ‘many of us are not native speakers of medical informatics’ [20]. Until the discipline matures enough to have its own language, people who span or bridge disciplines can serve as interpreters. Each of these categories of Special People has been described to some extent in the literature. Top-level support has been recognized as necessary by some, for example, but results of empirical studies have been mixed. Cooper and Zmud found that senior level support was needed for information technology implementation success [21]. Weir found in a Delphi study that top-level support and commitment are important for CPOE implementation [22], but in a later survey study it
248
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
was not found to be significant [23]. Ash [17] found that top-level support was not a significant predictor of success for computerbased patient record implementation. The need for champions for any change has been discussed in the literature [24], and Ash found they were significant for successful infusion of some health care information technologies [17]. Weir also found that interdisciplinary implementation groups are important in addition to intensive support [22]. Aydin and Forsythe, in their ethnographic study in ambulatory care, found that ‘tutors available on-site to answer questions in the clinical setting’ [25], the people we are calling bridgers, are indicated for successful implementation of the electronic medical record. The present study provides greater insight into the three categories of Special People than has been described previously, however. A number of attributes are shared among the three groups of Special People. They are all excellent communicators. The leadership level individuals connect with one another and with those in the clinical realm. Those in the clinical realm, even the curmudgeons, are connected with the leadership level and with the bridgers. The bridgers themselves are connected with the clinical realm and the CMIO. Bridgers share an interest in interpreting the language and culture of the different groups, thereby acting as interpreters needed by each group. All levels share a vision, a goal, and a commitment to make CPOE and clinical systems work on behalf of patient care: even the curmudgeon can be motivated by an increased ability to provide better patient care. Each category of people follows through with its passion for reaching the goals by utilizing behavior that encourages others: the leadership gives support, the clinical realm provides encouragement to peers, and the bridgers provide help at the elbow for users.
Another attribute they share is toughness. Complaints registered by users at the highest levels of the organization and frustrations with the system are often taken out on the individuals who provide support. The data raise a number of concerns about these Special People. All of these positions are extremely difficult. All of the individuals wear more than one hat and different roles often pull them in different directions. Time commitments are extraordinary. Those who continue clinical work while taking on responsibilities related to clinical systems are especially vulnerable to overwork. Burnout was overtly mentioned in relation to bridgers, but undoubtedly it is a factor in the turnover of staff in other areas as well. The methodology used for this study was labor intensive, but the complexity of the overarching research question demands multiple perspectives and methods. The use of observation can verify what one is told in interviews, and group interviews/focus groups offer the benefits of synergy and the building of ideas. The analysis process is time consuming, but to gain a true picture, a mix of researchers with different clinical and research backgrounds must become immersed in the data for a period of time. Studying one site rather than four for a shorter period of time with fewer researchers would have produced less data to analyze, but might have produced less transferable results.
5. Conclusions and recommendations: new bottles, new wine The results of this study have implications for informatics education and training. We were told that preparation for the Special People roles is inadequate at this time. At the CEO level, clinical knowledge is often lacking, as well as information systems expertise. The
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250
CIO may have limited administrative and clinical knowledge. The CMIO seems to need it all. The clinical realm, with the possible exception of the curmudgeons, includes many individuals desiring more information systems expertise. The bridgers are the most extraordinary group: they hold clinical expertise and they must gain their technical knowledge on the job. We would recommend that those involved in graduate programs, including programs in health administration and medical informatics, learn more about the roles of these Special People so that they can prepare students for them. The results also have budgetary implications. There must be general recognition that these Special People are important and their presence takes resources, both to hire them and to retain them. Most hospitals have CIO’s, and many larger hospitals are hiring CMIO’s. Special People in the clinical realm need to be identified as well, however. Also, a reward system that can motivate the champions, opinion leaders, and even the curmudgeons can increase involvement and, therefore, improve chances of success. It is imperative that bridgers be identified, trained, and rewarded in adequate numbers so that users have help at the elbow when and where they need it. These are the Special People who will interpret the hidden cultural assumptions in CPOE and in clinical systems. They must be heralded as the heroes of any successful implementation.
Acknowledgements This work was supported initially by Paul Mongerson and grant DE-FG03-94ER61918 from the US Department of Energy and later by grant LM06942-01 from the US National Library of Medicine. Special thanks go to
249
Tom Massaro and Gerri Frantz at the University of Virginia, Tom Payne at the VA Puget Sound Health Care System, and Bart Lally and Lynda Winterberg at El Camino Hospital for their assistance with the study. Mary Lavelle, Veena Seshadri, Paul Gorman, Jason Lyman, Jim Carpenter and Paul Nichol, helped to gather data. Special thanks to William F. Polsgrove for graphics.
References [1] Institute of Medicine Committee on Quality of Health Care in America, To Err is Human: Building a Safer Health System, National Academy Press, Washington, DC, 1999. [2] J.S. Ash, P.N. Gorman, W.R. Hersh, Physician order entry in U.S. hospitals, Proceedings of AMIA Annual Fall Symposium (1998) 235 /239. [3] J.S. Ash, P.N. Gorman, M. Lavelle, J. Lyman, L. Fournier, J. Carpenter, P.Z. Stavri, Perceptions of physician order entry: results of a cross-site qualitative study, Methods Information in Medicine, in press. [4] B.L. Berg, Qualitative Research Methods for the Social Sciences, fourth ed., Allyn & Bacon, Boston, 2001. [5] D.E. Forsythe, Studying Those Who Study Us: An Anthropologist in the World of Artificial Intelligence, Stanford University Press, Stanford, CA, 2001. [6] D.E. Forsythe, New bottles, old wine: hidden cultural assumptions in a computerized explanation system for migraine sufferers, Medical Anthropology Quarterly 10 (1996) 551 /574. [7] J.S. Ash, P.N. Gorman, M. Lavelle, J. Lyman, Multiple perspectives on physician order entry, Proceedings of AMIA Annual Fall Symposium (2000) 27 /31. [8] (a) T.A. Massaro, Introducing physician order entry at a major academic medical center: I. Impact on organizational culture and behavior, Academic Medicine 68 (1993) 20 /25; (b) T.A. Massaro, Introducing physician order entry at a major academic medical center: II. Impact on medical education, Academic Medicine 68 (1993) 25 /30. [9] T.H. Payne, The transition to automated practitioner order entry in a teaching hospital: the VA Puget Sound experience, Proceedings of AMIA Annual Fall Symposium (1999) 589 /593. [10] (a) J.P. Barrett, R.A. Barnum, B.B. Gordon, R.N. Pesut, Evaluation of a medical information system in a community hospital, Battelle Columbus Laboratories, Columbus, OH, Report No.: NTIS PB 248 340, 1975.; (b) J.P. Barrett, P.L. Hersch, R.J. Caswell, Evaluation of the impact of the implementation of the Technicon
250
[11]
[12]
[13]
[14] [15]
[16] [17]
[18]
J.S. Ash et al. / International Journal of Medical Informatics 69 (2003) 235 /250 Medical Information System at El Camino Hospital. Part II. Economic trend analysis, Final rept. 27 November 72 / 31 July 79, 1979. J.G. Anderson, C.E. Aydin, S.J. Jay, Evaluating Health Care Information Systems: Methods And Applications, Sage, Thousand Oaks, 1994. B.K. Kaplan, D. Duchon, Combining qualitative and quantitative methods in information systems research: a case study, MIS Quarterly 12 (1988) 570 /586. D.E. Forsythe, Using ethnography to build a working system: rethinking basic design assumptions, Proceedings of AMIA Annual Fall Symposium (1992) 505 /509. D. Morgan, The Focus Group Guidebook, Sage, Thousand Oaks, 1998. L.W. Brunet, C.Y. Morrissey, G.A. Gorry, Oral history and information technology: human voices of assessment, Journal of Organizational Computing 1 (1991) 251 /274. B.F. Crabtree, W.L. Miller, Doing Qualitative Research, Sage, Newbury Park, 1992. J.S. Ash, Organizational factors that influence information technology diffusion in academic health sciences centers, Journal of the American Medical Informatics Association 4 (1997) 102 /111. E.M. Rogers, The Diffusion of Innovations, fourth ed., Free Press, New York, 1995.
[19] N.M. Lorenzi, R.T. Riley, Managing change: an overview, Journal of the American Medical Informatics Association 7 (2000) 116 /124. [20] V.L. Patel, D.R. Kaufman, Medical informatics and the science of cognition, Journal of the American Medical Informatics Association 5 (1998) 493 /502. [21] R.B. Cooper, R.W. Zmud, Information technology implementation research: a technological diffusion approach, Management Science 36 (1990) 123 /139. [22] C. Weir, M. Lincoln, D. Roscoe, C. Turner, G. Moreshead, Dimensions associated with successful implementation of a hospital based integrated order entry system, Proceedings of AMIA Annual Fall Symposium (1994) 653 /657. [23] C. Weir, M. Lincoln, D. Roscoe, G. Moreshead, Successful implementation of an integrated physician order entry application: A systems perspective, Proceedings of AMIA Annual Fall Symposium (1995) 790 /794. [24] J.M. Howell, C.A. Higgins, Champions of change: Identifying, understanding, and supporting champions of technological innovations, Organizational Dynamics 19 (1990) 40 /55. [25] C.E. Aydin, D.E. Forsythe, Implementing computers in ambulatory care: Implications of physician practice patterns for system design, Proceedings of AMIA Annual Fall Symposium (1997) 677 /681.
