The Camera as an Actor: Design-in-Use of Telemedicine Infrastructure in Surgery

Transcription

1 Computer Supported Cooperative Work 12: 1 20, Kluwer Academic Publishers. Printed in the Netherlands. 1 The Camera as an Actor: Design-in-Use of Telemedicine Infrastructure in Surgery MARGUNN AANESTAD Deptartment of Informatics, P.O. Box 1080, Blindern N-0316 Oslo, Norway ( margunn@ifi.uio.no; Phone: (+47) ; Fax: (+47) ) Abstract. This paper describes the evolving interrelationship between a pre-established work practice and a new technology, with an emphasis on how the technology itself participates in the process and introduces changes, while at the same time being changed itself. The case study concerns the introduction of multimedia communication technology into a surgical operating theatre. Concepts from Actor-network theory are found to provide a useful perspective on the description and analysis of the case. The technology and the work practice are viewed as a new heterogeneous actor-network, whose configuration changed continuously. These changes are conceptualised as alignment attempts where the different actants interests are translated and inscribed into e.g. artefacts, rules or routines. The alignment of this heterogeneous network was achieved through a continuous process of design, test and redesign of different configurations of people, practices and artefacts. The relevance of the findings is discussed, related to how we may think about design of open and generic technologies. Viewing design as design of configurations; the creation of a well-working mix of people, practices and artefacts, may be a helpful and relevant design metaphor. Key words: actor-network theory, alignment, configuration, design-in-use, health care, inscriptions, telemedicine, translations, video-mediated communication, work practice 1. Introduction The interrelation between work practice and technology has always been the central research focus within CSCW, with its aim to design technologies to support human interaction in cooperative work arrangements. Ethnographic studies of technologies-in-use have described how successful technologies were inseparable from the situated activities in which they existed, and user-centered and situated design was advocated (Suchmann et al., 1999; Greenbaum and Kyng, 1991). Such situated, worker-oriented design might involve continuous further development, design-in-use or tailoring (Bjerknes et al., 1991; Henderson and Kyng, 1991; Mørch, 1995). The insight that the end-result of such processes may be unexpected has been suggested by the notion of drifting (Ciborra, 1996), which points to the deviations from the original goals and objectives and the unintended consequences of introducing information technologies. At least part of the responsibility for the drifting resides with the technology; due to its openness and plasticity the users can appropriate it differently.

2 2 MARGUNN AANESTAD Despite this, the role of technology in design-in-use has been little theorised, as the focus of the CSCW field has been on human cooperative work, its characteristics and its requirements to design, more than on the technology as such. Some empirical studies have focused on the technology s role in work practices and have described a variety of roles and tasks for technology (see e.g. Suchman and Trigg, 1991; Luff and Heath, 1998; Berg, 1997; Berg, 1999; Lundberg and Sandahl, 2000). However, there is still a lack of empirical studies that investigate the dynamics of the process where technology and work practice evolve together. The importance of understanding this dynamic interplay may become more evident if we consider the change within information and communication technology in recent years. From the earlier company-specific and locally designed information systems, more and more networked, standardised and off-the-shelf products are used. Where communication technologies are concerned (as in this paper), the technology is rather open and generic; the proper way to use it is not specified and it can be appropriated in different ways. Using network technologies necessarily involve other actors, over whose actions and choices one has limited control. In this context design will happen more as design-in-use and less in defined design projects. Design will also have a broader scope than software development or tailoring, as the products are more or less packaged. We may envision that an important part of design will be to create an interrelationship between the technology and the work practice (or other human activity), in other words, to design technology use. The aim of this paper is to contribute to the understanding of these broader design-in-use practices by focusing on how technology may take part in such processes. The paper presents a case study from telemedicine, where multimedia communication technology (cameras, microphones and loudspeakers) was installed in a surgical operating theatre. This facilitated communication with external viewers, and the intended use was to demonstrate medical procedures to others, either to guests at the department or to remotely located surgeons in training. However, the technology was not implemented in one batch, according to definite plans. It was rather characterised by piecemeal growth, changes and adjustments of both the technical set-up and of rules and procedures. Based on a description and analysis of this process, a view of design as design of configurations is proposed. The paper describes how such design of configurations may play out in very practical terms. 2. Case study and method 2.1. ABOUT THE CASE STUDY AND THE RESEARCH SITE The Interventional Centre was established in 1996 at the National Hospital in Oslo, Norway, to do research and development in image-guided and minimalinvasive therapies in medicine. Minimal-invasive or keyhole surgery is different from ordinary (also called open ) surgery in that it minimises the invasiveness of

3 TELEMEDICINE INFRASTRUCTURE IN SURGERY 3 the procedure. Instead of a large cut to facilitate the surgeon s direct vision and manipulation of organs, surgical instruments and optics for a small video camera are entered through small incisions that may be 5 10 mm wide. The surgeons focus is on video monitors that provide the image of the patient s organs as well as the instruments actions. Thus the surgeon s visual and tactile information, as well as the actions on the organs, is indirect or mediated by technology, i.e. the monitors and the surgical instruments (see Figure 1). Usually an assisting surgeon holds and moves the video camera according to directions from the main surgeon, and both surgeons may have their own monitor to watch, as they often are positioned on either side of the operation table. The scrub nurse hands over instruments to the surgeon, either on explicit directions or based on anticipation of the surgeon s needs. Thus the mediated images are important also for the nurses in the room as the images provide awareness of general progress and facilitate anticipation of specific tasks. The presence of a mediating technology makes minimal-invasive surgery a likely candidate for telemedicine. The video image is regarded as the main information source for the surgeons during the procedure, and it is relatively easy technically speaking to transmit this video signal. Minimal-invasive surgery is a work practice that is heavily integrated with a lot of technology in continuous development and change, so the introduction of new devices (e.g. communication technology) does not signify a change from a non-technical to a technically oriented work practice. Rather, it is just one more addition to the array of tools. This new technology is not a necessary tool for the primary object of work; it is rather an additional, non-critical, and voluntary technology, which may facilitate new forms of mediated communication not previously possible. A lot of visitors to the Interventional Centre were anticipated when it was planned. To avoid disturbing the operating theatre team more than necessary, as well as let the guests avoid changing clothes when entering the sterile zone, a local analog transmission facility was set up between the operating theatres and an external room. Images from overview cameras in the operating theatres were available; in the interventional suite (which is the theatre in focus in this paper, see also Figure 1) there were one wall-mounted camera as well as one camera mounted in the roof directly above the operation table. In addition, images from several other sources, including x-ray, gastroscope, laparoscope, and ultrasound equipment etc. could be transmitted and displayed in the external room. It was possible to transmit and display images on 16 small monitors, and one of these images could then be selected and displayed on a large monitor (Figure 2). Twoway audio through microphones and loudspeakers facilitated conversation between the operating theatre team and the guests located in the external room. This room was also the hospital s studio for videoconferencing and the videoconferencing equipment was connected to the local network, so that image and sound from the local facilities could be transmitted to any ISDN-connected receiver. This possibility was intermittently used, but the external transmissions became more frequent and regular during a large telemedical project, where image and sound was trans-

4 4 MARGUNN AANESTAD Figure 1. A minimal-invasive surgical procedure in the operating theatre. The two surgeons to the left, and scrub nurses in the middle and to the right. Note the two monitors with the internal video image, and the microphones hanging from the roof. mitted across an ATM network to receivers at other hospitals, mainly focused on training sessions on minimal-invasive surgery. The department employed several community workers, who served 14 months service as an alternative to army service (conscientious objectors). They were not regarded or paid as ordinary employees. The workers that were recruited were mainly engineers with different kinds of computer science background. Due to lack of space in the department, they were assigned this room as their workplace. Managing telemedicine transmissions was not intended to be their main task, but it became an important and large part of their duties as the activities expanded. It is these persons who are called technicians in the paper. 3. Method Participant observation was the main method used for gathering data. Around 50% of the author s time during one and a half years was spent at the Interventional Centre, assisting in the planning, execution, and evaluation of most telemedical transmission that took place during that period. My practical role in the execution of transmissions was that of a technician, assisting with equipment set-up, camera control etc. My impression is that I was perceived as an ordinary employee who was pursuing a Ph.D. like several others among the engineering and medical staff, rather than as an external researcher observing the local work practice. Some of

5 TELEMEDICINE INFRASTRUCTURE IN SURGERY 5 Figure 2. The technicians and the equipment in the external video room. Note the rack with 16 small monitors. the staff was employed part-time (still working in their mother department), so my part-time presence was not unusual. Also the hospital was a teaching hospital, which meant that there were very often guests and students present in the operating theatres. At this particular department visitors were even more common, so the operation team were used to strangers observing their work. Participant observation may imply, among other problems, a lack of distance and a bias towards partial representations. My role and identification with some parts of the work practice is partly a result of active choice, partly a result of contingencies. From the empirical material a focus on nurses and technicians work is evident, as well as a corresponding lack of focus on the medical-professional aspects of the surgical work. The bias towards nurses and technicians work should not be taken as a negation of the importance of the medical-professional parts of the work practice. The integration of telemedicine into clinical practice is crucially dependent on acceptance from the medical doctors, i.e. the surgeons in this case. Clinical quality and safety are crucial to surgeons, and this is a topic that is complex and important enough to require a separate study. With the limited focus here (integration into the work practices in the operating theatre) these issues are outside the scope of this paper. Surgeons have a central role and position within the surgical work practice, but their temporal and spatial presence (within the operating theatre) is limited, and their responsibilities and tasks are clearly demarcated from those of other groups. My conviction is that this in some ways shields them from much of the alignment work that goes on inside the operating theatre, which is the topic

6 6 MARGUNN AANESTAD of this paper. This work is most clearly to be seen in the general preparation work of the nurses, as well as in the facilitation work of the technicians. These considerations are the basis for my pre-occupation with these groups work tasks, as well as for the choice of participant observation as the data generative method. This method provides the best way to get access to instances, problems and discussions through a phase of learning and re-configuration. The issues and problems that emerge in such an explorative phase may be thought of (by the informants) as too trivial and mundane to be captured by formal interviews. Examples of such tasks may be the technicians searching for cables of appropriate length, or the nurses relocating the operation lamp so as not to block the camera view. In addition to participant observation, a log was kept either by myself or another technician over each transmission (date, time, duration, content, topic, support time needed for preparation and execution, technical and other problems encountered, free text etc.). This log provided documentation about when the issues discussed below arose, which problems were encountered and how they were solved. In addition, personal diary notes were filled in on details, and textual documents including project documents, documents from planning of transmissions, technical cabling plans etc. have been available to me. Discussions between the technicians occurred after each transmission and were focused on discovering causes and remedies for technical or logistic problems. Informal discussions with the medical staff provided information on how the transmissions were perceived. Also a couple of more formal discussion meetings were carried out to bring out perceived problems and suggested solutions. 4. Theoretical perspective The field work was started without a pre-defined analytic framework, but the observations soon indicated the central role of the nurses and technicians work with the technical artefacts in order to achieve the wanted benefits from the technology. A theoretical framework should allow this aspect of the deployment process to stay central. Although there are several candidate theories to account for the sociotechnical aspects, the wish to investigate specifically the technology s contribution required a framework where the technology would be acknowledged on line with other actors. This is what has been perceived as actor-network theory s main benefit, and is also what motivated my choice of it (Berg, 1998, 1999; Monteiro and Hanseth, 1995; Walsham, 1997). I use it as a perspective that throw new light on the observations, rather than as a rigid framework to explain them. Thus the benefits of the theory may be seen both in the structuring of the empirical data, and even more so through the leverage it allows in the interpretation of the case. Actor-network theory (ANT) is in practice not a unified body of concepts, and an eclectic approach to it is widespread. This necessitates a brief overview over the ANT variety used in this paper. An actor-network is a heterogeneous network of human and nonhuman actors or actants (actant is used by some as

7 TELEMEDICINE INFRASTRUCTURE IN SURGERY 7 a more neutral term than the human-centred actor) where the relations between them are important, rather than their essential or inherent features (Latour, 1987; Callon, 1986, 1991). The theory argues that it is analytically fruitful to reject any a priori distinction between elements in the network, as e.g. the distinction between humans and non-humans. Differences are instead viewed as effects (i.e. as achieved or constructed) rather than as pre-given (Law, 1992). This quest for symmetry has created a lot of controversy around ANT, both in general and within the IS community (see e.g. the chicken debate (Collins and Yearly, 1992; Callon and Latour, 1992)). The critique (at least within the IS community) has often focused on the concept of material agency. Attempts to avoid the strong symmetry assumption of ANT, e.g. (Jones, 1999) are evidence of uneasiness about granting non-humans agency on line with humans. Jones suggests the possibility of viewing material agency as capacity, lacking the intentionality component that is a distinctive feature of human agency. However, to respect the epistemological foundations of ANT we also need to keep in mind that agency is an emergent and not an essential or inherent property of the actors. It is the network that provides the actors with opportunities to establish and use agency (Law, 1992). In this paper the words capacity for action or capacity for influence is used when analysing the agency of the camera (i.e. the communication technology). A pragmatic approach to this debate might be to argue that the network elements are not interesting per se (their agency as isolated entities); rather the relations and connections between them in real life are what we are interested in. When it comes to network dynamics, ANT claims that the actors in the actornetwork have different and possibly incompatible interests, and that stability (or order, agreement, success, goal achievement) is obtained when the network is aligned (Latour, 1987). The alignment of the network occurs through a process where the actors interests are translated (i.e. reformulated, modified, or changed) into more generally agreeable expressions, so that several actors may support the resulting translation (Callon, 1991; Law, 1992; Latour, 1991). The translation may be inscribed into a medium, e.g. the characteristics of an artefact, a rule, a procedure or a standard (Latour, 1991). The inscription attempts to define a framework for possible action (a program of action), and it may be more or less strong (Akrich, 1992). The inscription s strength is not just a static entity, once defined by the designer and inscribed into the object. It is also a dynamic and relational feature that emerges in the actual network when actors are delegated roles (Hanseth and Monteiro, 1997). Delegation is when an actor stands in for and represents other actors and acts on behalf of them. As people may be delegated roles in an organisation, actors (human or non-human) may be delegated a role in the actor-network, with associated power, competence and responsibilities. The actors in the network may follow, twist or oppose the inscriptions program of action, depending on the inscription s strength (Akrich, 1992). If the actors are supporting (more or less voluntarily) a given translation and its inscription, they have been enrolled and are cooperating (still more or less voluntarily) towards a

8 8 MARGUNN AANESTAD common goal (Latour, 1987). If sufficiently many or strong actors are enrolled in the network, the opposing interests (the non-cooperating actants) may be forced to yield. A network may evolve towards a stable state with relatively irreversible and unchangeable inscriptions (Callon, 1991), it may be in a state of flux and instability, or it may be only temporary stable and aligned. I also use the concept of configuration in addition to these mainstream ANT concepts. Configuring was used by Steve Woolgar (1992) to describe how users were ascribed identity (e.g. capacities and character) and how the users range of actions were defined and delineated by computer designers. In this paper the word configuration is used to denote a particular set-up or mix of entities, as a kind of instantaneous picture of the actor-network. Both which elements that are included, and how they are connected to each other is important aspects of a configuration, as this defines their roles and relations and thus also their power and capacities. This resonates with actor-network theory s claim that it is in the relations that capacity for action lie: Action is a property of associated entities... Action is simply not a property of humans but of an association of actants... Provisional actorial roles may be attributed to actants only because actants are in a process of exchanging competences, offering one another new possibilities, new goals, new functions. (Latour, 1999, p. 182). I attempt to use these concepts to describe how the interrelation between the work practice and the technology unfolds. A new network emerged when several actors with different interests were brought together, and the description of the following alignment process is focused on the technology s role and contribution. 5. The camera meets the surgical work practice The following report of empirical data from the case study is separated into two main parts, corresponding to a time-based structure. First, some of the effects and consequences of the installation and use of the local transmission network is described. This created some changes and disturbances to the pre-established work practice, which spurred a process of modifications of both the technology and the work practice. Second comes the description of the issues that arose when the local network was expanded and connected to an external telemedical network THE LOCAL TRANSMISSION FACILITY The local transmission facility was fulfilling a wish of both the operation team (being relieved of visitors in the operating theatre during work) and of the guest (who would avoid changing clothes). The room overview cameras were remote controlled, which further relieved the operation team of the task of camera operation (i.e. change the viewing angle, focus, or zoom), which would have

9 TELEMEDICINE INFRASTRUCTURE IN SURGERY 9 been irrelevant tasks to their primary activity, the surgical procedure. However, this particular solution had some unexpected effects and consequences as well. Through three examples from the integration process we will investigate how the matching of the different interests and demands was carried out Yielding to the demands from the technology All the information that one wanted to transmit across the communication channel had to be captured and made electronic. In order for the surgeon s voice to be audible to the remote listeners, there had to be a microphone in a position where it captured the sound waves of the speech (and preferably not too much other sound). In order to provide a view of the room or the operation table, a camera had to be present, turned on and connected to the network. The medical imaging equipment had to provide a copy of the image signal to be captured and transmitted via cables. We could say that some of the communication devices required feeding of input material. These inherent inscriptions in the technology required changes in the work practice by introducing new works tasks, mostly for nurses and technicians. The nurses had to connect the imaging equipment to the network, turn on the cameras, and ensure adequate positioning of microphones. These tasks became included in the nurses general preparation work of surgical procedures when a transmission was to take place. The technicians had to provide cables or other artefacts, devise solutions to technical problems, and assist in connecting and managing the equipment. Another consequence of the introduction of the technology was changes in the conditions of the work within the operation theatre. Several disturbances were introduced, e.g. noise and sound through the audio lines and a generally messy situation with cables on the floor and presence of technicians in the operating theatre during the operation Inscriptions activated in the network configuration The presence of cameras and microphones in the operating theatre that were linked to screens and loudspeakers in the external room created uncertainty over surveillance issues. The new technology provided a possibility of invisible spectators, as technically knowledgeable personnel (and others present in the room) could watch and listen to the activities in the operation theatre without themselves being seen or heard. This new possibility contributed to a shifting of control over the work situation, and the department had to decide on routines and rules for usage. After a while a key was installed beside the wall-mounted camera in the operation theatre. This key had to be turned on by the operation theatre personnel in order for the transmission of video signals to be possible, and when the key was turned a red on air lamp was lighted. The sound transmission lines were not included in this system, so it was still possible to listen to the conversation without the red lamp on. This was remedied through a voluntary and explicit commitment by the technicians

10 10 MARGUNN AANESTAD to never listen (i.e. actively send the sound signals to a loudspeaker) unless the operating theatre personnel had activated the video transmission control Inscriptions created in the network configuration The chosen technology and the installation of it implied a projection of expected receivers and use areas. Both the department s research activities and the specific telemedicine activities had a strong focus on medical doctors (especially surgeons and interventional radiologists), as opposed to other groups. The part of a surgeon s work that occurred within the operation theatre had a defined localisation in time and space, it was performed after the patient and equipment was prepared and occurred mainly around the operation table. One of the cameras was mounted in the roof above the operation table, and also the microphones and loudspeaker were located here. Thus the placement of the equipment in the room reflects this focus on surgery. As most of the actual transmissions were performed for medical doctors, the installation of the equipment was adequate. When the use of the network deviated from the projected usage, this became problematic. The operation or anaesthesia nurses work was distributed over much larger space, from reception of the patient at the entrance, and across the whole of the operating theatre. Nurses often worked in parallel with different tasks in different parts of the room, and their work was stretched out in time with less uniform intensity than the surgeon s work. In order to document nurses work for an offline video production, an additional mobile video camera was enrolled to be able to capture the work that was distributed in space. The limitations were also related to the audio part, as the microphones above the operation table was unable to capture the verbal explanations when the workers were located elsewhere in the room. Audio (explanations and comments) had to be laid on in the editing process. The inscriptions evident in the installation of the technology were rather strong; still it was possible to observe several instances of unintended use. For example the technicians might use the camera to monitor who was present in the operating theatre. Especially if one of the technicians was on the way to the operating theatre to correct something, the camera might be directed towards the door to see when he entered. Then the control room technician could phone or speak through the loudspeaker to communicate with him. At some instances the cameras would also be focused on cables and connections to try to detect the source of lacking images or other problems. Sometimes when there were no transmissions or guests, the audio lines would be used for transmitting music from the video room to the team working in the operating theatre EXPANDING THE NETWORK New challenges and changes were introduced when the local transmission facility was hooked up to a telemedicine network. The old ways of handling the issues were insufficient, and new alignment activities were necessary. These changes also

11 TELEMEDICINE INFRASTRUCTURE IN SURGERY 11 reverberated backwards into the previously aligned network, and changed the established and tested configurations Even more changes in work practice becomes necessary The expansion of the network required enrolling more equipment, and as a result a more complex material infrastructure emerged in the video room. Audio mixers, several loudspeakers and headsets were purchased to be able to monitor and adjust sound to and from the different participants, and more cameras and monitors also became necessary to manage more image sources and destinations. In addition several video recorders were included in the set-up, to be able to record transmissions. The resulting increase in complexity led to more dependence on technicians. Earlier the departments medical doctors had been able to go into the video room and initiate a local transmission for their guests, or to carry out an ordinary videoconference. With the new complex set-up they became dependent on the assistance of the technicians in order to do this. The support work changed in content, character and importance, and the technicians acquired a more intermeshed and crucial role in the network (Johansen et al., 1999). In addition to managing the local equipment, new tasks for the support technicians included establishing the connection to the other side, verifying proper image transfer both ways and doing sound checking (which in its nature was very interactive). Then the local network had to be activated and connected to the external (telemedicine) network, with corresponding image and sound checks, typically between nurses and technicians. The required images from the operating theatre had to be selected, the sound quality adjusted, and the transmission monitored during its whole extent. Planning and coordination work, both internal (with local participants/the work team) and external (with the other site s technicians or receivers) became a larger part of the technicians work than before. Other use areas outside the operating theatre (e.g. transmission of lectures and meetings, off-line video recording and editing) were also important experiences, as they provided the technicians with opportunities to learn to handle the technology and exploit possibilities (Aanestad and Hanseth, 2000) Problematic network configurations The surgeon s main focus during a laparoscopic procedure was on the video image from the inside of the abdomen. However, during some procedures also other image sources were used, e.g. ultrasound or X-ray images, which were displayed on separate screens. Sometimes the surgeon switched focus between the different images frequently and rapidly. This dynamic image use was not a problem when students were present in the operating theatre, as the surgeon s direction of vision and other small bodily or verbal cues were easily detected. However, it turned out to be difficult for the technicians in the control room to detect or interpret the surgeon s change of focus and consequently switch to the correct image source

12 12 MARGUNN AANESTAD to transmit. This demanded constant and close attention to the verbal interaction and all the available images from the operating theatre (on the 16 small cm, black and white monitors in the control room), a task that the technicians felt tiring and demanding. They also lacked the medical knowledge necessary to determine which image was in focus at which time. To be sure that the transmission would be successful, the surgeon had to comment explicitly on the image use, a task that was not necessary when the students were present in the room. At one instance, a students facilitator was present with the operating surgeon, explaining and commenting on the procedure. The explicit comments on image use greatly helped the technicians to decide and switch images correctly. In other words, a new work task needed to be performed by the surgeon, or a new role as a medical producer may be called for. Other alternatives may be envisioned: The technology may perform this task, e.g. via intelligent image switching, based on automatic detection of surgeon s direction of view. Or the task could be delegated to the receivers if all available images were transmitted and displayed at the same time, so they could choose for themselves which image to watch. Teaching students, no matter whether they were co-present or distant, required explanations and demonstrations. The form of these explanations and demonstrations changed when they were mediated through a communication technology. The co-present students would have to be directed and positioned in the operating theatre in order to provide them with an adequate view of the activity. When teaching distant students, the challenge was to provide them with the adequate view in the virtual space and ensure that they would receive the intended information. To achieve this the surgeon might have to request certain camera views from the technicians or to pause action and minimise movements of instruments in order for the transmitted image quality to be optimal Feedback and reciprocity The discussions in the early phases about surveillance issues led to the enrolment of a control key and a red warning light. This usually worked well, as the guests present in the external room often were expected and known in advance, or they were presented to the staff in the operating theatre. When the external transmissions (telemedicine) started, these measures were not sufficient. During the first test transmissions some receivers (at the other side) would come and go according to their local duties. The operation team did not know how many or exactly who the receivers were, or whether they where present or not at any given moment. If there had been quiet time periods without interaction, the operating surgeon might ask explicitly who were watching. As the technology provided a two-way audioand video-connection, an image was actually transmitted also from the receivers site, but this image was (in the beginning of the project) only being displayed in the video room, where the technicians were located. Later this image signal was forwarded also into the operating theatre and displayed on a monitor. Then the whole team could see who the receivers were. It would however, still be possible

13 TELEMEDICINE INFRASTRUCTURE IN SURGERY 13 for a receiver to position oneself outside the camera view and thus be an invisible viewer. If the surgeon wanted to address the audience before the surgical procedure started, he/she needed to know which camera (e.g. the roof- or the wall-mounted room camera) that was currently active in order to simulate eye contact with the audience. The natural impulse is to look at the image of the receivers when talking to them. Also in other situations the surgeon or the team might want to know which camera view was transmitted, how close it zoomed, or which of the other available image sources was transmitted. In other words feedback was required in order to manage and control the self-presentation. Some transmissions were performed to large audiences at medical conferences where not only images from the operating theatre, but also images from other rooms as well as PowerPoint presentations and videos were transmitted. Then the operating theatre was not on air all the time. Even though the images were constantly transmitted to the control room, they were not transmitted any further all the time. At these instances the operating theatre team was provided with the outgoing image (the image sent to the receivers), and they could then relax when they knew that they were not being watched. This image was displayed at the same monitor as mentioned before, and consequently they were not able to see a picture of the receivers. This was not felt to be interesting, as long as one knew that there would be a lot of people watching anyhow. This shows the dynamic character of the work practice s demands to feedback and reciprocity, according to different contexts. The two different genres of communication required different configuration of the network elements (i.e. the monitors and image signals) Less control over the rest of the network External transmissions also implied less control over receivers values and norms, and potential secondary use of information. Discussions arose over protection of the patient s rights and dignity. After some instances of discussions and disagreement, the nurses and the technicians agreed on some rules. There would usually not be any recording or transmission during the early phases of the procedure, where the patient was anaesthetised and naked. After the patient s body was covered, and only the area of interest to the surgeon was exposed, the camera system was turned on. Focusing on the patient s facial features was avoided, and patient information on X-ray images was blanked before they were transmitted. The feeling of less control also led to increased efforts to create a shared awareness in the department through organisational measures. Planned transmissions and off-line video recordings were included as a note on the weekly work schedule, which listed all the patients with their planned procedures, time and location, as well as the participating staff. It was also requested that information about content and purpose of planned transmissions should be given at weekly meetings between all staff.

14 14 MARGUNN AANESTAD 6. Network configurations The issues and problems that were encountered in the operation theatre are well known, as the study of requirements to communication and interaction in virtually collocated teams are central to the field of CSCW (see e.g. Hollan and Stornetta, 1993; Dourish et al., 1996; Mark et al., 1999). The problems of awareness, mutual visibility, reciprocity and feedback, the handling of multiple views, articulation work and coordination (Schmidt and Bannon, 1992; Schmidt and Simone, 1996), and surveillance (Nardi et al., 1996), have all been documented earlier. However, this paper has a somewhat different focus than to provide specific advice and design recommendations. The aim is to more fundamentally investigate the technology s contribution to the continuous design-in-use, tailoring or drifting process. How can the evolving interrelationship between technologies and work be understood? How does the technology acquire its potential for making a change, and how does its capacity for influence play out? Which implications do this have for design? 6.1. INVITATION AND CREATION OF AN ACTOR The reason for the inclusion of the communication technology into the work practice was its designed features or inscriptions; its ability to create image and sound signals that could be transmitted to remotely located receivers. The technology was delegated a role of being the receivers eyes and ears. This role provided the technology with its location in the actor-network and legitimised the costs and disturbances that followed. However, the technology s attributes (e.g. the camera image quality, zoom span, viewing angle, and the possibility for remote control), besides being the reason for inclusion, also introduced unwanted effects like the operation team s perceived loss of control. These effects did not emerge before the technology was included in the actor-network. An uninstalled camera lying in a box wouldn t create concerns of surveillance, but its inscriptions became activated when it was installed in this particular location in the actor-network, between the external room with unseen viewers and the operation room. This illustrates that an artefact s capacity for action or influence is relational and not essential. It occurs or plays out in relations, when the artefact is part of an actornetwork; it is not a feature that is objectively present in an autonomous and isolated entity. We may also see from the empirical material that an artefact s capacity for action may become curtailed or modified by the relations to the network around it. An example of this is the installation of the red light and the control key in the operation theatre, which arose from a wish to adjust this particular configuration of the network with the aim to re-distribute control to the operation team.

15 TELEMEDICINE INFRASTRUCTURE IN SURGERY THE ACTOR IS A NETWORK Inscriptions occur not only within artefacts or network elements, but also from the way these elements are connected. For example were definite inscriptions present in the seemingly neutral and obvious design of the communication facility, in how the equipment was installed with a focus on surgery. These inscriptions reflected more or less explicit organisational values, and they were more or less chosen or intended. We saw, however, that as the network configuration changed, e.g. with different use areas, the effect of these inscriptions varied. In some instances it seemed possible to appropriate and capitalise on the present inscriptions, in other instances the inscriptions seemed to constitute limitations that must be circumvented for example by enrolling other artefacts. An artefact s capacity for influence is thus dynamic, not static and given once for all by the artefact s attributes and its location. What is a powerful location in one configuration of the actor-network may be a disadvantaged position in another. The capacity for action is not only relational and dynamic; it is also a collective feature and cannot be appropriated to specific network elements. Some tasks necessary for successful action were delegated to artefacts (the extra monitor as a representation of the receivers, the red light as a warning of transmissions), and other tasks were delegated to humans (providing input to the technology and producing the context relevant to the receivers). In fact the achievement of alignment was not that easy to pinpoint and allocate to specific actors, it was rather a network effect, where the humans plus the artefacts achieved the result. For example, in order for the monitor in the operating theatre to represent the receivers, several other human and non-human actants needed to act accordingly. A camera had to be present at the receivers site, turned on and connected, and the receivers had to position themselves in front of the camera. The transmitted image had to be transmitted further to the operating theatre, a task that required technicians, additional cables, an unused monitor, somebody to turn it on etc. Correspondingly, in order for the human actants to provide the technology with the required input, they had to enrol microphones and cameras, as well as cables. There was a network of human and non-human actants around or behind each actant; nobody could act in isolation. Successful alignment of the network was a collectively achieved result. Also the effects or results of introducing the new technology may be discovered only at the collective or network level, as displacements DISPLACEMENT OR THE CONSEQUENCES OF INVITING THE ACTOR Displacement can be understood as the direction the collective takes once its shape, extent, and composition have been altered by the enrolment and mobilization of new actants (Latour, 1999, p. 194). Incorporating the technology into the work practice definitely effected change and displacements that may be attributed to the introduction of the technology. However, the changes are not an expression of the technology s pre-given capacity for influence in any deterministic sense, but

16 16 MARGUNN AANESTAD emerge in this particular network configuration, or set of configurations. From this case study we may see at least two features of this displacement Increase in complexity, power, and dependence Through incorporating the new technology a more complex and enmeshed hybrid of human and non-human actants emerged. The work practice expanded and became able to both perform a surgical procedure as before and also to transmit it to external viewers. The actor-network (the new work practice) became more powerful and far-reaching, as it could extend itself and communicate with distant receivers. This was the intended and wanted change. At the same time, control and power was shifted out and became distributed across the network rather than located to specific actants. The dependency of the actants on each other was increased. This expansion thus also reduced the power, in the sense that there was less control, and less predictability than before. We may say that the network technology s capacity of influence has manifested itself in the introduction of network dependencies. The focus of this paper is on the changes in the local work practice, rather than the whole network of different cooperating partners and institutions, but this point has profound implications for the development of telemedicine (or other network technologies). Use of a common network means dependence on other actors and a necessity of relating to their actions and choices. This implies that no actor can act as it wants without constraints, but at the same time no single actor is able to control the others and dictate their actions. Introducing telemedicine or other network technologies, will therefore require open and evolutionary strategies, which are aimed at enrolling allies, rather than control-oriented, specification-driven strategies From procedures to performances A shift or displacement of the work practice towards a performance in front of an audience was evident. The operating theatre literally became a theatre, a stage, as it originally also was. The patient was the one most immediately affected, by having images from the inside of the body transmitted to the outside world. The patients interests were represented by the nurses and the surgeons, and the interests have been translated into new rules for patient information and consent, as well as restricting the camera s view by defining rules for when it was OK to record or transmit. This amounts to defining and managing the patient s role in the performance. The operation team encountered a new work situation, and the discussions around the surveillance issues as well as the wishes for feedback and reciprocity in the communication situation were signs of the shift towards performance aspects of the work. As one of the central persons in the work team the surgeon was affected by being in front of the camera. The camera made the work more visible and thus might increase the level of pressure to produce good results. It also influenced

17 TELEMEDICINE INFRASTRUCTURE IN SURGERY 17 the communication with the receivers (e.g. need for simulation of eye contact when greeting audience, more explicit comments on image use). Other parts of the work practice occurred behind the camera, either in the very literal sense of physical camera control, or more generally the production and support work. In order for the camera to see anything, a whole apparatus of people, cables, connectors etc., had to be present behind it. Its presence demanded new work tasks and new roles previously unknown in the surgical work practice. Other effects of the camera s presence could be seen e.g. in the debates and discussions in the department around routines and the attempts to create general awareness. We thus see that the camera introduced increased visibility of (some) work, and increased work to produce this visibility DESIGN OF CONFIGURATIONS These displacements are expressions of the technology s capacity for influence, but at the same time they are emergent and non-deterministic. The activities described above were situated in nature and attempted to define and delineate the technology s role in particular situations. There was no grand plan, and neither was there only one possible solution to most of the problems encountered. The importance of the configuration of the actor-network should be evident from the above discussion, and the alignment work performed by the technicians and nurses may be characterised as design of configurations. Different connections and relations between the different elements in the actor-network were devised, tested and redesigned. Weak points or missing links were discovered and strengthened, new elements (e.g. technical artefacts) were enrolled into the actor-network, or the capacities of existing elements were worked at (e.g. the competence of users and support technicians). Different configurations of the actor-network had different distribution of burdens and benefits, and some were preferred over others. At some instances, the effects of a given configuration were problematic to live with and required adjustments or re-design. Then there may be several potential ways to solve the problems, as the technicians image mixing illustrates. This iterative and continuous work of creating, analysing, testing, and adjusting network configurations, may be a salient feature of design of such open and generic technologies. When the use is not pre-specified in any detail, the discovery of well-working configurations of users and technology may become the core of the design process. This would include both working with the different network elements, their competencies and capacities, as well as the relations between them, e.g. their location in the network and the links with other elements. Viewing design as design of configurations captures the contingent, continuous and nondeterminate nature of the process, and also the multiplicity and heterogeneity of the participants. In addition it may bring attention to the fact that as the process is iterative and continuous, it will be costly and resource demanding (funds, competence, personnel etc.). To make and keep a network aligned requires care (Ciborra,