Electronic communication and collaboration in a health care practice1

Electronic communication and collaboration in a health care practice1

Artificial Intelligence in Medicine 12 (1998) 137 – 151 Electronic communication and collaboration in a health care practice1 Charles Safran a,b,*, P...

94KB Sizes 0 Downloads 32 Views

Artificial Intelligence in Medicine 12 (1998) 137 – 151

Electronic communication and collaboration in a health care practice1 Charles Safran a,b,*, Peter C. Jones b, David Rind a,b, Booker Bush a, Kayla N. Cytryn c, Vimla L. Patel c a

Department of Medicine, Beth Israel Deaconess Medical Center, 330 Brookline A6e., Boston, MA 02215, USA b Center for Clinical Computing, Har6ard Medical School, 350 Longwood A6e., Boston, MA 02115, USA c Cogniti6e Studies in Medicine: Centre for Medical Education, McGill Uni6ersity, 1110 Pine A6e W., Montreal, Qc H3A 1A3, Canada Received 31 December 1996; received in revised form 31 July 1997; accepted 1 September 1997

Abstract Using cognitive evaluation techniques, this study examines the effects of an electronic patient record and electronic mail on the interactions of health care providers. We find that the least structured communication methods are also the most heavily used: face-to-face, telephone, and electronic mail. Positive benefits of electronically-mediated interactions include improving communication, collaboration, and access to information to support decision-making. Negative factors include the potential for overloading clinicians with unwanted or unnecessary communications. © 1998 Elsevier Science B.V. Keywords: Collaboration; Electronic mail; Electronic patient records; Decision support; Evaluation

* Corresponding author. Tel.: + 1 617 6671596; fax: + 1 617 6671002; e-mail: [email protected] 1 The study was in part supported by a cooperative agreement with the Agency for Health Care Policy and Research and the National Library of Medicine (U01HS08749), and in part by the Social Sciences and Humanities Research Council of Canada (No. 410-95-1206). 0933-3657/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S 0 9 3 3 - 3 6 5 7 ( 9 7 ) 0 0 0 4 7 - X

138

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

1. Introduction The electronic patient record that was developed and is now deployed at the Beth Israel Deaconess Medical Center has fundamentally changed the practice of medicine in ways that its developers never foresaw [39]. This type of highly interactive and work flow enabled program is creating new collaborative roles for computers in complex organizations. The system is able to supervise and monitor care [38,40], enabling computers to perform many care coordination and documentation functions, freeing people to concentrate more fully on interpersonal interactions and providing of health care services. This article reports preliminary results from one of the first projects to apply cognitive evaluation techniques to the assessment of the effectiveness of clinical systems deployed in a health care practice. A variety of theoretical articles have described models for using computers to support group activities, including linguistic models [49], form-based models [13], process models [20], and intelligent agent models [30]. Further articles have discussed evaluation of collaborative systems in laboratory environments [24,26], showing that group interactions are enhanced by incorporating work flow into application systems. Examination of the effect of computerization on social processes in real world settings has been lacking, however. [17,19]. Evaluations of deployed collaborative systems have tended to focus on the ability of these systems to replace conventional communication media without describing the effects on interpersonal and organizational communication [6,7,15]. Observations of the organizational impact of implementing hospital-based computer systems have indicated that the impact on individual roles and communication patterns can be both significant and unexpected [1–3]. This study applied a global evaluation strategy to the validation of the effectiveness and usability of computer systems in an outpatient practice, together with an assessment of the effects on practice and group interactions within the health care team [16,21,22,33]. The effects of computerization on the interactions of health care providers with patients and families whereas also examined, as well as patients’ attitudes toward the computer system. Medicine is characterized by the need for experts in many roles (physician, nurse, etc.) to work collaboratively to perform complex tasks. A patient normally interacts with individuals in many roles during the course of a visit, each requiring a high degree of specialized training, and having clearly defined responsibilities. Patient care is thus an ideal domain for investigating the effects of computer-assisted collaborative systems on complex real-world environments. Given today’s explosion of knowledge, decision making has become a process of distributed cognition in which team members bring their expertise and domain knowledge to the issue at hand, collaborating to arrive at the best possible solutions [29,31,35,42]. Technology has contributed to these processes by providing improved information management, just-in-time decision support, and new means of communication such as electronic mail (e-mail). In a clinical practice, providers in nearly every clinical role must communicate with those in every other role to some degree to meet patient needs. This interdependence of roles favors a highly egalitarian

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

139

pattern of communication. Consistent with this egalitarian pattern, we found that the least structured modalities of communication are also the most heavily used: face-to-face, phone, and electronic mail. Electronic mail was observed to be the preferred mode of communication for brief messages that are not immediately urgent. Previous studies have documented the economic and interactive benefits of electronic mail in co-operation with other forms of communication [8,10,34,45,47]. This is one of the first studies to document the emergence of electronic mail as the dominant and preferred communication method for a wide variety of interactions. The ability to reach anyone at any time without going through administrative ‘red tape’ can be seen as one of the chief benefits of electronic mail. This openness has the undesired consequence however, of generating large amounts of e-mail. We found that some clinicians are becoming overloaded with more electronic mail than they can handle. It is not unusual in the practice studied for a provider to receive more than 50 electronic mail messages a day. Some clinicians reported receiving as many as 100 messages on heavy days. Using cognitive evaluation techniques, this study examines the effects of the electronic patient record and e-mail on the interactions of health care providers, evaluating the changes generated by the introduction of computer capabilities into the health care team.

2. Background Boston’s Beth Israel Deaconess Medical Center is served by a highly integrated clinical computing system introduced in the late 1970s [5]. It functions in the coordination of the flow of information between clinicians, who keep their medical records online, as well as between clinicians and their support environment. The computer system at the Beth Israel Hospital includes approximately 5000 computer terminals throughout every department of the hospital. Each department communicates via these terminals (as well as through other methods such as telephones and face-to-face contact), so that information is entered directly into the computer system and is then instantly available to anyone who requires it. E-mail is also available and heavily utilized (16 000 messages/week), and has become a key method of communication among the hospital staff [43,46]. Additional features which provide support to the health care professionals include decision support programs in areas such as electrolyte and acid-base program, information on medications [46], support in the care of patients with HIV [40], computer-based alerts to changes in condition [38], and assistance with determination of discharge medications [44]. A MEDLINE literature search program, PaperChase, is also available and is used frequently, with 1600 searches in a 1 week period. Analyses of components of the systems have been carried out since their inception and the results have been encouraging in establishing the usefulness of the individual technologies introduced. The manner in which technology has been incorporated into everyday practice has been evaluated through studies of fre-

140

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

quency of utilization [5,39,43] as well as its effects on clinician’s patientcare decisions and on patient outcomes [38,40,46,41]. The study reported here broadens the scope of these examinations of technology in health care by investigating the impact of computerization on collaborative processes among team members.

3. Methods

3.1. Study site The Beth Israel Deaconess Medical Center in Boston is a 478 bed hospital affiliated with Harvard University with approximately 30 000 admissions and 200 000 out-patient visits annually. These patients are cared for by 3500 health care professionals. Healthcare Associates (HCA) is an outpatient unit delivering primary care. It occupies the sixth floor of the outpatient wing of the Beth Israel Hospital. It is divided into three suites, or groupings, designated south, central, and north. Each suite has its own door, registration desk and waiting area. They lead to interconnected areas consisting of examining/interview rooms, conference rooms, work rooms, and practice assistant pods. Each of the three suites includes a conference room, a work room, examining/interview rooms, two pods, and several hallway stations with terminals. Internal waiting areas provide places for patients to wait that are more accessible to the health care providers. The conference and work rooms contain four to six terminals, the pods contain two terminals, and each examining room contains one terminal. The practice has had an electronic patient record since 1991, and clinicians are able to care for patients without traditional paper records [39]. The terminals are either dedicated terminals or consist of a stand-alone personal computer with terminal capabilities.

3.2. E6aluation methods Analysis of work patterns within the practice was conducted to identify interactions between team members, examining social dynamics of the decision-making process, group strategies and processes through which decisions are made, peer influences, and organizational (hospital) influences on decision-making within the unit [32]. The methodologies employed were developed at the Centre for Medical Education, McGill University, and are aimed at assessing the group processes involved in complex medical decision making. Techniques have also been developed for the analysis of the use of advanced computer systems in clinical settings [23,33]. The outcome of this analysis is a model of the interactions within the unit, detailing roles and interaction patterns. This interactive

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

141

model also provides information as to the roles of the tools provided by the computer system, validating their effectiveness in terms of whether team members employing them achieve their goals. The methodologies implemented are based on an ethnographic approach, with the aim of observing the environment and the roles within it while not disturbing the processes under investigation [9]. The study began with a work domain analysis [36,37], in which the objectives, functions, activities, and resources of the team were determined. Goals and limitations were identified and priorities determined. General functions and activities were then described. The setting in which these activities take place and the material resources available provide the physical context for the work domain, and are therefore significant components of the analysis. Data was collected over two periods of 2 weeks each, during which key practitioners were accompanied as they went about their daily activities. This generated a picture of a typical day in the lives of individuals in the unit. Each subject was accompanied for one half day as he or she went about delivering patient care. Practice assistants (who assist providers with administrative tasks, direct patients and book appointments, assist with procedures, draw blood samples, answer telephones, etc.) and secretaries at the reception desk were also observed in a less structured manner. The purpose and methods of the study were explained to each subject. When patients were involved, the provider explained the study to them, including their right to refuse with no repercussions. Detailed field observations and audio recordings of all interactions were made. Interactions through voice mail and e-mail were included as well [9]. Patient and provider consent were obtained before audio or video recording. Patient-provider interactions were videotaped [1,22]. These tapes were then analyzed for the setting, i.e. HCA office or telephone, the diagnoses of each patient and the health care providers involved in their care, and the amount of time spent by the provider doing several activities. Semi-structured interviews were conducted and audiotaped exploring the attitudes of the patients toward the computer and their providers’ use of it [11]. By combining a multiplicity of data collection techniques and sources, we attempted to generate a broad picture of the functioning of the unit [50] providing a model of interactions and group decision-making processes within the unit, and an evaluation of the role of the computer system within those processes. Data was collected at the Beth Israel Hospital and analyzed at McGill University. An activity model was developed based on these observations. The analysis was conducted with particular attention to the role of technology in supporting practice in the unit. The work domain, including the physical setting, the individuals, and the patterns of work activity were analyzed, and an activity analysis initiated. Results of the analyses were compared with characterizations of traditional primary care units to assess changes generated by the implementation of the computer system.

142

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

4. Results

4.1. Communication among clinicians Interactions among clinicians were analyzed by defining clinical roles, specifying goals, tasks, and responsibilities accomplished within each role, and observing the types of interactions that occurred among people filling each role. The variety and complexity of issues presented by the patient population of HCA require a broad range of expertise. HCA personnel therefore includes experts in the medical, nursing, mental health, and community resource domains. Within each domain, the level of expertise ranges from novice (e.g.: Intern) to specific expert (e.g.: Faculty Physician, Psychiatrist, Nurse Practitioner, Social Worker). The administrative hierarchy is based on lines of accountability, reflecting level of domain expertise, education, training, and experience required of the various roles. The provider roles described do not function in isolation but rather as a team. Orasanu and Salas [29] describe a team as not only a group of people, but as a group with members who are interdependent and coordinated with defined roles and responsibilities. They receive information from many sources, manage internal resources adaptively, and possess expertise relevant to the tasks involved in achieving the goals of the group. The actions and decisions of the individuals within the team occur in the context of the team, the social environment, and the physical environment [16,18,31,35,42]. Communication is a fundamental pathway through which individual roles develop into a functioning team [29] and therefore was examined further. A content analysis was performed on observed communication between providers. No single topic category was seen to account for more than 25% of communication, and six categories each counted for more than 10% (care plan, patient status, team activity, intervention, diagnosis, and administration). This diversity of topics suggests that standard forms such as mail templates will probably be of limited use. We feel that text templates are best applied to information that is already structured, such as patient assessments, care plans, and discharge plans. A very small percentage of communication was devoted to evaluation (0.5%). This finding suggests that computer systems for the automatic evaluation of care, while often discussed as a potential application of artificial intelligence technology, may have limited application in an outpatient setting. It is clear that evaluation in the practice studied is occurring continuously as part of the delivery of care. The pattern of interactions showed a constant flow of communication, consistent with the team approach of the unit. This flow is not based on any visible hierarchy, though there is more communication within provider areas. The most striking observation about the interactions between the providers of HCA is the egalitarian pattern. Communication centers on the goal of the interaction, with the players, content, and direction being determined by the nature of the expertise required and the caregiver who possesses it, be that Community Resource Specialist or Practice Director. Interactions among providers are based on individual levels and domains

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

143

of expertise rather than on hierarchical positions, creating this egalitarian pattern. The communication pattern observed within the team is consistent with earlier studies that observed egalitarian tendencies in computer-assisted communication [17,48], and with cognitive research demonstrating that efficient communication is a hallmark of experts and expert teams [12,28].

4.2. Integration with clinical practice Of the sample of 12 providers interviewed, 11 reported finding the computer a useful addition to their practice and an improvement over previous tools. The 12th provider later modified initially negative comments, stating that the computer’s usefulness outweighed its drawbacks. A major advantage of the computer system observed during the delivery of patient care is the easy accessibility to information from any terminal. Terminals themselves are easily accessible, located at key locations such as examining/interview rooms, hallways, conference rooms, work rooms, and secretaries’ desks. This allows providers access to all of it’s the system’s features without tying them to specific locations. This is particularly useful for roles involving coordination of patient flow, such as Clinical Triage Nurses, Community Resource Specialists, Practice Assistants, and Secretaries, in that they have quick and easy access to the patients’ records and can assess issues based on knowledge of the patients’ histories. Results of tests and consultation referrals are available, and medication orders and progress notes are easily accessed and verified. Unscheduled issues can be assessed quickly and efficiently, even during a phone call, through access to the online medical record system (OMR). Appointments can be coordinated since everyone has access to appointment schedules of both patients and other providers. Several providers report finding the output of the system to be useful. A popular feature of the OMR is the prescription reordering and printing, in which prescriptions do not have to be re-entered and re-written. Both steps can involve as little as a few key strokes, which is seen as a major advantage over paper methods. It also facilitates the maintenance of an updated list of medications in the medical record. A printout of portions of the OMR is also provided to providers prior to the patients’ appointments. They were seen to rely on it heavily, referring to it during visits and making notes on it. Many clinical areas outside of the primary care practice do not utilize the on-line medical record for complete charting. This affects the inclusiveness of OMR and creates frustration for the primary care providers when looking for records of departments that are not online. All providers at the hospital have computer access to results of tests ordered by other departments.

4.3. Effects on patient-pro6ider interaction The computer system facilitates several components of the patient-provider interaction. Easier access to information releases providers to attend to the patient and increases the resources and information available to do so. Examples include

144

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

ordering medications, which no longer requires a hand-written prescription, the readily available Physician’s Desk Reference, a well-known reference source for physicians’ practice, and the Formulary, a list of all medications provided by the Beth Israel Hospital pharmacy. Their accessibility facilitates providing patients with information and clarifying unfamiliar medications that patients ask about, as well as maintaining an updated medication list, checking other appointments that the patient has booked or that a provider of interest has available, sending referrals and consults, and maintaining a readily available problem list. The system also provides teaching tools, such as the ability to print out information describing any medication in the Formulary, while the patient is available and without having to look through files, shelves, and cabinets. Whenever any provider comes into contact with any patient, they have access to the medical record as well as planned appointments. This assures that the providers are up-to-date with the patients’ care and gives patients a sense of being known, as any provider will have an awareness of the issues of importance to them. This access is extended to providers who are outside the hospital, allowing them to access the system from home, other offices, and other institutions. This allows providers to prepare for appointments ahead of time, to follow-up on expected results quickly, and to remain accessible and in communication while away from the unit. Initial observations show that the presence of the computer does affect the provider/patient interactions in subtle ways. The necessity of ‘interacting’ with the terminal is distracting, taking attention away from the patient. Providers were observed to have compensated for this by reviewing patient material prior to the appointments and by using the OMR printout provided rather than the OMR itself while actually with the patient. The extent of the influence on the interaction relates very strongly to usability. The more interaction required by the system, the more distracting it is and the more likely providers are to resort to more traditional, familiar modes of functioning, i.e. paper records. This might be remedied by having the patients looking at the screen as well as the providers, creating a three-way interaction that might be more comfortable for the providers as well as for the patients. Patient interviews showed no relationship between the amount of time providers spent with the computer and patient satisfaction. In fact, patients were uniformly positive about the system, though one patient reported that she had initially been leery of it. This finding is consistent with a recent study showing a high degree of acceptance of patient-provider electronic mail [27], which is used at HCA by hospital employees who are also patients. Features noted by patients included improved access and ease of prescription writing for the provider. There were no reports by patients of the computer interfering with patient-provider interactions. The major concern expressed was that of confidentiality. Voice mail provides similar capabilities as electronic mail, but listening is more time-consuming than scanning a written note and is more difficult if not impossible to screen. E-mail is also more likely to be used than voice mail because of

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

145

the awareness, as reported in interviews, that providers will be alerted to the presence of e-mail messages each time they log onto their terminals and can quickly flip through these messages. Voice mail is reported to be less noticeable and picking up messages was observed to require more thought and attention. Providers spend proportionally more time actually talking with the patient during a phone call than during an office visit, with MDs using the OMR as backup during a phone call more than NPs. Providers use the computer more when on the phone, and use the printout more during an office visit.

4.4. Alerts and reminders Alerts and reminders are a feature of OMR which prompts providers that a particular intervention might be indicated for a particular patient. They appear automatically and are a major advantage of on-line medical records in that they provide just-in-time triggers, again reducing the cognitive load required of the provider. Reminders differ from alerts only in that the messages in reminders are not important enough to interrupt workflow, whereas the information in an alert message may require urgent action. A reminder might indicate that the clinician has not signed a progress note after a certain amount of time, while an alert might suggest that a provider should immediately change the dose of an antibiotic because of a change in a patient’s renal function [38]. Of the providers observed, most appreciated the presence of alerts and reminders and checked them regularly. Another perspective, seen more in expert practitioners, saw them as annoying in that they were unnecessary. It is possible that the expert has a sufficient mastery of the area that such reminders are not required and therefore serve only as distractions, while the provider at an intermediate level of expertise would appreciate the support. However, this expert perception is dissonant with observed behavior in a controlled clinical trial of alerts and reminders [40,41], in that providers at all levels of expertise found great benefit in computer generated suggestions. The availability of just-in-time reminders and alerts together with the other decision support systems online represent a significant difference from more traditional methods of the delivery of care. Although at times perceived as distracting, these tools support the reasoning of the expert provider as well as reducing the memory load. Without reminders, the provider must remember and be alert to all the information related to each patient. This is a significant demand on attention and on the memory of a traditional provider. Allowing the user to choose the level of alerts and reminders, adapting to their own level of expertise, might reduce the interference in expert reasoning processes. Access to reference materials during the patient visit also supports decision making. In a traditional unit it is necessary to locate references. Common ones are often available in a central location, but this does not match the efficiency of on-line availability.

146

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

4.5. Collaboration Collaboration is one of the tenets of good patient care in a practice like HCA. The level of expertise of the providers is such that they practice independently, with most communication being either to inform or to consult. This reflects an efficient use of interactions, with the expertise of each provider being at a level such that independent practice can be supported. Roles and responsibilities are defined sufficiently clearly that each provider provides care required by the patient and can communicate that care process using the computer system, so that comprehensive care is still provided. Consultation was observed to be most frequent between specific experts and generic experts. Nurse practitioners consult with preceptors and primary care physicians (PCP) regarding medical decisions. Clinical nurses consult with nurse practitioners and PCP regarding decisions requested by patients asking for advice or information. Residents consult with preceptors and physicians frequently. The nature of the expertise necessary to deliver health care is beyond the ability of one individual to master. Decision making is therefore distributed among the members of the team. In a study of an ICU setting [31], it was found that each member of the health care team had specific responsibilities, and that each complemented the other. This is observed in this sample as well. The involvement of the computer in this process is facilitated by its ready availability throughout HCA as well as from outside the hospital. When the team is discussing an issue, the computer is at the location and is included as a resource.

5. Discussion One of the challenges in the design of computer systems to assist health care providers is how to support collaboration while not requiring that people meet face-to-face. Our evaluation has indicated that the hospital’s electronic patient record and e-mail system represent successful applications of technology for this purpose. However, we have found that the introduction of these new technologies changes the clinical environment in both positive and negative ways. Positive benefits include improving communication, collaboration, and access to information to support decision-making. Negative factors include the potential for overloading clinicians with unwanted or unnecessary communications and perceived isolation of providers. A major difference observed between the HCA and more traditional primary care units is the accessibility of the patient record. This is in contrast with a traditional unit in which all information is in a paper chart which is usually stored in a medical records department. If access to the chart is required, it must be requested and retrieved from this central location. This can take as little as a few minutes or as long as a day. Because clinical records are so accessible in HCA, they are always consulted when discussing an issue with a patient or provider, with no consideration of time or effort required to gain access to a paper chart. Every time a patient

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

147

calls with a question, the OMR is accessed and an informed reply provided. This accessibility extends to locations outside the hospital, allowing providers to monitor their patients even when not physically in the building. Without computers, this requires telephone contact, which can involve delays. The nature of data entry in a computer system allows for more efficient updating of records, such as problem lists and prescription lists. They do not have to be rewritten, or crossed off when no longer applicable. Also, computers can simply print out the contents of the record. This is particularly useful in the area of medication re-ordering. With a computer system, producing a paper prescription for the patient to take to the pharmacy requires a few keystrokes. In addition, the prescription is legible, reducing the risk of error. In a traditional system, each medication must be re-written each time the prescription must be renewed. For this reason, this feature is one of the most popular of the OMR. In addition to the traditional telephone, voice mail, pager, and meeting, clinicians communicate using e-mail and the electronic patient record. The electronic patient record has become a means of communication in that the observations, plans, and interventions of other providers are readily available. If an entry is urgent or otherwise important, it is sent via e-mail rather than the electronic patient record. The efficiency of e-mail and its ability to convey detailed information in a rapid manner has facilitated the exchange of information between providers. E-mail was both observed and reported by providers to be a major method of communication. Voice mail provides similar capabilities as electronic mail, but listening is more time-consuming than scanning a written note and is more difficult if not impossible to screen. E-mail is also more likely to be used than voice mail because of the awareness, as reported in interviews, that providers will be alerted to the presence of e-mail messages each time they log onto their terminals and can quickly flip through e-mail messages. Voice mail is reported to be less noticeable and picking up messages was observed to require more thought and attention. The large volumes of e-mail observed (up to 100 messages per provider per day) might appear to some degree a sign of a successful application of technology. People use the e-mail system extensively because they find it superior to other modes of communication for certain types of messages. Individuals might exchange several electronic messages in a day even though they are located in offices a few yards from each other. Using e-mail rather than the phone or a face-to-face meeting in the hall allows each party to attend to the communication when it is most convenient, which can significantly improve efficiency. We feel that the difficulties providers face is not the volume of mail but rather the appropriateness and relevance of the content of messages. We believe most providers would welcome a system where the messages are brief, to the point, delivered to the appropriate individuals, and contain links to the additional information needed for the recipient to take whatever actions are required. We are therefore focusing on improving the content of communication by incorporating multimedia objects, providing Internet access, and allowing documents and data to be attached. These capabilities are allowing computers to assume a new role in complex health organizations. Earlier studies [14] have described the impact that technology has

148

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

had in changing human roles from an active system manipulator to a system supervisor or monitor. Modern data processing systems, however, need very little human supervision and are able to take on roles within a project team. With the reduced need to supervise and monitor equipment, people are able to concentrate more on interpersonal interactions and providing of services. An example of this is the use of e-mail, which now allows a computer system to replace the mail trucks and runners that in an earlier day were indispensable to the functioning of a large complex organization. We may conceptualize this change by saying that a computer has replaced people in performing the ‘letter carrier’ role within the organization. When the computer takes on the letter carrier role, the process changes, and with it, the levels of expectation. Messages are now delivered instantaneously and extremely reliably. People have begun to expect a response within the same day, often sooner, and are annoyed if they do not receive it. E-mail is often the fastest medium for getting a response, faster than scheduling a face-to-face meeting and faster than leaving a phone message. By virtue of technology, mail has changed from a means of communicating over distances and creating a written record to the favored way of communicating brief messages that require a quick response. E-mail as it is currently provided, however, uses only a small fraction of the capabilities of a computer. As long as the computer system’s functions are limited to the delivery of mail in the form that it is received, the computer is performing an administrative support role, and can not truly be described as collaborating in the work process. Computers for administrative support must achieve high standards of quality, often higher than would be expected of a person in the same role. Computers are effective as providers of e-mail, for example, because they rarely lose a message and deliver mail much faster than the post office. Collaboration places the additional requirement on a computer system that it functions in a way that enhances the capabilities of the people on the work team, just as would be expected of a human collaborator. This places enormous demands on the user interface, because it must not only be easy and effective to use, but must also anticipate the kinds of tasks that users are likely to want to perform [4,25]. The bridge to collaboration is crossed, we feel, when the computer begins to process the contents of the communication. A key capability is seen in automatic alert and reminder programs — programs that send a message to a clinician about some event or workflow issue that needs the clinicians attention. These programs cross an implied social barrier by initiating an event based on information that the recipient of the reminder or alert may not be aware of. They carry the implication that the determination of the decision to display the alert is sufficiently authoritative to demand the recipient’s attention. Electronic patient records and health information systems that include e-mail can change the practice of medicine. Although such systems can improve collaborative care, they can also contribute to information overload. In the future it will be important to design clinical systems that incorporate these work-flow considerations.

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

149

References [1] E. Arborelius, T. Timpka, In what way may videotapes be used to get significant information about the patient-physician relationship?, Med. Teaching 12 (1990) 197 – 208. [2] C.E. Aydin, Occupational adaptation to computerized medical information systems, J. Health Soc. Behav. 30 (1989) 163–179. [3] C.E. Aydin, R.E. Rice, Bringing social worlds together: Computers as catalysts for new interactions in health care organizations, J. Health Soc. Behav. 33 (1992) 165 – 185. [4] N. Bjorn-Andersen, Are ‘Human Factors’ human?, Comput. J. 31 (1988) 386 – 390. [5] H.L. Bleich, R.F. Beckley, G. Horowitz, J. Jackson, E. Moody, C. Franklin, S.R. Goodman, M.W. McKay, R.A. Pope, T. Walden, S.A. Bloom, W.V. Slack, Clinical computing in a teaching hospital, N. Engl. J. Med. 312 (1985) 756–764. [6] P.J. Branger, J.S. Duisterhout, Communication in health care, Methods Inform. Med. 34 (1995) 244–253. [7] P.J. Branger, J.C. van der Wouden, B.R. Schudel, E. Verboog, J.S. Duisterhout, J. van der Lei, J.H. Van Bemmel, Electronic communication between providers of primary and secondary care, Br. Med. J. 305 (6861) (1992) 1068–1070. [8] N.S. Contractor, E.M. Eisenberg, Communication networks and new media in organizations, in: J. Fulk, C. Steinfeld (Eds.), Organizations and Communication Technology, Sage, Newbury Park, CA, 1990, pp. 143–172. [9] G. Cooper, C. Hine, J. Rachel, S. Woolgar, Ethnography and human-computer interaction, in: P.J. Thomas (Ed.), The Social Interactional Dimensions of Human-Computer Interfaces, Cambridge University Press, New York, 1995, pp. 11 – 36. [10] A.B. Crawford, Corporate electronic mail — A communication-intensive application of information technology, MIS Quart. 6 (1982) 1– 14. [11] K.A. Ericsson, H.A. Simon, Protocol Analysis: Verbal Reports as Data, MIT Press, Cambridge, MA, 1993. [12] D. Gaba, Dynamic decision-making in anesthesiology: Cognitive models and training approaches, in: D.A. Evans, V.L. Patel (Eds.), Advanced Models of Cognition for Medical Training and Practice, Springer-Verlag, Heidelberg, Germany, 1992, pp. 123 – 148. [13] H. Hammainen, E. Eloranta, J. Alasuvanto, Distributed form management, ACM Trans. Inform. Sys. 8 (1) (1990) 50–76. [14] P.A. Hancock, M.H. Chignell, Adaptive control in human-machine systems, in: P.A. Hancock, Peter A. (Ed.), Human Factors Psychology. Advances in Psychology c 47, North-Holland, Amsterdam, 1987. [15] P. Hart, D. Estrin, Inter-organization networks, computer integration, and shifts in independence: The case of the semiconductor industry, ACM Trans. Inform. Sys. 9 (4) (1991) 370 – 398. [16] D.R. Kaufman, V.L. Patel, J.F. Yale, R. Bouchard, A.W. Kushniruk, Physician’s knowledge and practices in diagnosing and treating hypercholesterolemia and the effects of a CD-I program on learning and behavioral change, Technical Report, Cognitive Studies in Medicine, Centre for Medical Education, McGill University, Montreal, Canada Qc, 1994. [17] S. Kiesler, J. Siegel, T.W. McGuire, Social psychological aspects of computer-mediated communication, Am. Psychol. 39 (10) (1984) 1123 – 1134. [18] G.A. Klein, J. Orasanu, R. Calderwood, C.E. Zsambok, Decision Making in Action: Models and Methods, Ablex Publishing, Norwood, NJ, 1993. [19] R. Kling, Cooperation, coordination and control in computer-supported work, Comm. ACM 34 (1991) 83–88. [20] H. Krasner, J. McInroy, D.B. Walz, Groupware research and technology issues with application to software process management, IEEE Trans. Sys. Man Cybernet. 21 (1991) 704 – 712. [21] A.W. Kushniruk, D.R. Kaufman, V.L. Patel, Y. Levesque, P. Lottin, Assessment of a computerized patient record system: A cognitive approach to evaluating an emerging medical technology, MD Comput. 13 (1996) 406–415.

150

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

[22] A.W. Kushniruk, V.L. Patel, Cognitive computer-based video analysis: Its application in assessing the usability of medical systems, in: R. Greenes, H. Peterson, D. Protti (Eds.), MEDINFO 95. Proceedings of the Eighth World Congress on Medical Informatics, North-Holland, New York, 1995, 1566–1569. [23] A.W. Kushniruk, V.L. Patel, J.J. Cimino, R.A. Barrows, Cognitive evaluation of the user interface and vocabulary of an outpatient information system, in: J.J. Cimino, (Ed.), Proceedings of the 1996 American Medical Informatics Association Annual Fall Symposium, Hanley and Belfus, PA, 1996, 22–26. [24] J. Lee, T.W. Malone, Partially shared views: A scheme for communicating among groups that use different type hierarchies, ACM Trans. Inform. Sys. 8 (1990) 1 – 26. [25] T.W. Malone, Computer support for organizations: Toward an organizational science, in: J.M. Carroll (Ed.), Interfacing thought: Cognitive aspects of human-computer interaction, MIT Press, Cambridge, MA, 1987. [26] T.W. Malone, K.Y. Lai, C. Fry, Experiments with oval: A radically tailorable tool for cooperative work, ACM Trans. Inform. Sys. 13 (1995) 177 – 205. [27] R.A. Neill, A.G. Mainous, J.R. Clark, M.D. Hagen, The utility of electronic mail as a medium for patient-physician communication, Arch. Family Med. 13 (1995) 268 – 271. [28] J. Orasanu, Shared mental models and crew decision making, Technical Report Number 46, Cognitive Sciences Laboratory, Princeton University, Princeton, NJ, 1990. [29] J. Orasanu, E. Salas, Team decision making in complex environments, in: G.A., Klein, J., Orasanu, R., Calderwood, C.E. Zsambok (Eds.), Decision Making in Action: Models and Methods, Ablex Publishing, Norwood, NJ, 1993, pp. 327 – 345. [30] J.Y.C. Pan, J.M. Tenenbaum, An intelligent agent framework for enterprise integration, IEEE Trans. Sys. Man Cybernet. 21 (1991) 1391 – 1407. [31] V.L. Patel, D.R. Kaufman, S.A. Magder, The acquisition of medical expertise in complex dynamic environments, in: A. Ericsson (Ed.), The Road to Expert Performance: Empirical Evidence from the Arts and Sciences, Sports and Games, Lawrence Erlbaum, Hillsdale, NJ, 1996, pp. 127 – 165. [32] V.L. Patel, A.W. Kushniruk, J.F. Arocha, Professional performance improvement in the education sector: Report to the World Bank, Technical Report, Centre for Medical Education, McGill University, Montreal, Que., 1996. [33] V.L. Patel, A.W. Kushniruk, J.F. Arocha, D.R. Kaufman, Cognitive evaluation of emerging health information technologies: HEALNet Canadian Network of Centres of Excellence, Technical Report, Centre for Medical Education, McGill University, Montreal, Que., 1995. [34] C. Perin, Electronic social fields in bureaucracies, Comm. ACM 34 (1991) 75 – 82. [35] D.N. Perkins, Person-plus: A distributed view of thinking and learning, in: G. Salomon (Ed.), Distributed Cognition: Psychological and Educational Considerations, Cambridge University Press, New York, 1993, pp. 88–110. [36] J. Rasmussen, Deciding and doing: Decision making in natural contexts, in: G.A. Klein, J. Orasanu R., Calderwood, C.E. Zsambok, (Eds.), Decision making in action: Models and methods, Ablex Publishing, Norwood, NJ, 1993, pp. 158 – 171. [37] J. Rasmussen, A.M. Pejtersen, L.P. Goodstein, Cognitive systems engineering, Wiley, Toronto, Ont., 1994. [38] D.M. Rind, C. Safran, R.S. Phillips, Q. Wang, D.R. Calkins, T.L. Delbanco, H.L. Bleich, W.V. Slack, Effect of computer-based alerts on the treatment and outcomes of hospitalized patients, Arch. Intern. Med. 154 (1994) 1511– 1517. [39] C. Safran, H.L. Bleich, W.V. Slack, Role of computing in patient care in two hospitals, MD Comput. 6 (1989) 141–148. [40] C. Safran, D.M. Rind, R.B. Davis, D. Ives, D.Z. Sands, J. Currier, E. Caraballo, K. Rippel, Q. Wang, C. Rury, W.V. Slack, H.J. Makadon, D.J. Cotton, Guidelines for the management of HIV infection in a computer-based medical record, Lancet 346 (1995) 341 – 346. [41] C. Safran, D.M. Rind, D.Z. Sands, R.B. Davis, D. Ives, J. Currier, W.V. Slack, D.J. Cotton, H.J. Makadon, Effects of a knowledge-based electronic patient record on adherence to practice guidelines, MD Comput. 13 (1996) 55 – 63.

C. Safran et al. / Artificial Intelligence in Medicine 12 (1998) 137–151

151

[42] G. Salomon, No distribution without individuals’ cognition: A dynamic interactional view, in: G. Salomon, (Ed.), Distributed cognition: Psychological and educational considerations, Cambridge University Press, New York, 1993, pp. 111 – 138. [43] D.Z. Sands, C. Safran, W.V. Slack, H.L. Bleich, Use of electronic mail in a teaching hospital, Proc. Annu. Symp. Comput. Appl. Med. Care (1993) 306 – 310. [44] D.Z. Sands, C. Safran, Closing the loop of patient care-a clinical trial of a computerized discharge medication program, Proc. Annu. Symp. Comput. Appl. Med. Care (1994) 841 – 845. [45] T. Singarella, J. Baxter, R.R. Sandefur, C.C. Emery, The effects of electronic mail on communication in two health science institutions, J. Med. Syst. 17 (1993) 69 – 86. [46] W.V. Slack, C. Safran, H.L. Bleich. Computerization in hospital-based delivery systems, in: T. Trabin, M.A. Freeman, (Eds.), The Computerization of Behavioral Healthcare, The Jossey-Bass Managed Behavioral Healthcare Library, Francisco, Jossey-Bass, San Francisco, CA, 1995, pp. 151–171. [47] L. Sproull, S. Kiesler, Reducing social context cues: Electronic mail in organizational communication, Manage. Sci. 32 (1986) 1492–1512. [48] E. Williams, Experimental comparisons of face-to-face and mediated communication. A review, Psych. Bull. 84 (1977) 963–976. [49] T. Winograd, A language/action perspective on the design of cooperative work, Human Comput. Interact. 3 (1988) 3–30. [50] D.D. Woods, Process-tracing methods for the study of cognition outside of the experimental psychology laboratory, in: G.A., Klein, J., Orasanu, R., Calderwood, C.E. Zsambok, (Eds.), Decision Making in Action: Models and Methods, Ablex Publishing, Norwood, NJ, 1993, pp. 228–251.

.