University of Maastricht

Geoffrey Bowker

University of Illinois, Champaign/Urbana

Contact address:

Marc Berg

Dept. of Health Ethics and Philosophy

Maastricht University

P.O. Box 616

6200 MD Maastricht

The Netherlands

Phone: +31-43-3881234/3881144

Fax: + 31-43-3670932

E-Mail: Marc.Berg@GW.UniMaas.NL

October 1996

To appear in Sociological Quarterly

THE MULTIPLE BODIES OF THE MEDIC AL RECORD:

Towards a Sociology of an Artifact

This paper argues that the medical record is an important focus for sociological research. In medical work, the modern patient's body Foucault has so aptly described is produced throug h embodied, materially heterogeneous work - and the medical record plays a crucial role in this production. It does not simply represent this body's history and geography: it is a central element in the material re-writing of these. Simultaneously, the re cord fulfills a core role in the production of a body politic. As the record is involved in the performance of the patient's body, it is also involved in the performance of the clinic in which that body comes to life. Finally, we argue that< i>different records, different practices of reading and writing are intertwined with the production of different patient's bodies, bodies politic, and bodies of knowledge. As organizational infrastructure, the medical record affords the interplay and coordination of divergent worlds. Seen in this light, as a site where multiple stories about patients and about organizations are at stake (including the interoperability between these stories), the medical record becomes highly relevant both analytically and politically.

In his Birth of the Clinic, Foucault argues that the classical, pre-modern "medicine of species" required a two-dimensional table as an intermediary between the individual body and medical knowledge (1973/1963). T he table would translate individual symptoms: it would yield the true nature of the disease by showing how they fitted into the eternal scheme of things. Symptoms were not the disease itself: they were pointers to this higher truth, which merely "precipit ated" in individual bodies, and which the table could decode. In contrast to the medicine of species, Foucault argues, the modern clinical gaze requires no such intermediary. Truth is no longer found and organized elsewhere in some grand nosological schem e, but rather in the pathological processes of individual bodies. The gaze deciphers this truth by following the symptoms inwards, eliciting signs, and differentiating the pathological reality that now is the disease.

Yet the development of thi s gaze depended crucially on the development of some new intermediaries. Writing was crucial in this new configuration. "Medicine no longer tried to see the essential truth beneath the sensible individuality; it was faced by the task of perceiving, and to infinity, the events of an open domain" (Foucault 1973, p. 98). In order to develop a body of true knowledge, medicine had to record individual cases: only in the accumulation of such experience, only in the totality of observers/observations coul d true knowledge be generated (Fagot-Largeault, 1989; Dagognet 1970). A cascade of inscriptions, to use Latour's term, typified and produced the possibility of this means of knowing (Latour 1987; 1993, pp. 171-225).

Although Foucault does not discuss this in the Birth of the Clinic, knowing in the practice of medicine is similarly dependent on writing. The power of the gaze, in other words, would not go very far if it stood isolated (cf. Atkinson 1995, pp. 60-65). No longer typifi ed by the metaphorical two-dimensional table Foucault describes, modern medicine could not be imagined without that other object of consultation: the medical record. In this paper, we argue that the medical record is fundamental to the everyday production of that contemporary body whose archeology Foucault describes (a body which hides the essences of the disease in the pathological processes taking place in its tissues; where the symptoms and signs attest to a reality which is never completely accessible in life) and to the everyday production of the organizations which enact and treat it.

How is the patient's body produced in hospital wards? How is its specific geometry and its historicity created? How is the patient's body transformed from the liv ed body of Mr. Thompson into a juxtaposition of organs, parameters, rows of numbers, graphs, and so forth? Is this through a specific way of looking? Or is this a discursive transformation, achieved through talk (as is a frequent focus of so cial constructivist medical sociology)?[1](see e.g. Davis 1986; ten Have 1994) We will maintain that the body is produced through embodied, materially heterogeneous work (Hirschauer 1991; Cussins 1996). Much comes into pla y here: urine containers, infusion pumps, nursing routines, doctor's consultations, and so forth. All these artifacts, individuals and organizational routines are intermediaries which together perform the medical body. Following recent developments in sci ence and technology studies (Bijker and Law 1992; Clarke and Fujimura 1992; Latour 1996; Star 1995), we argue that they constitute the network, or the dispositif within which the body acquires its specific ontology. We focus on the medical record b ecause this artifact occupies a central niche in this network: it is where many of the nurses and physician's tasks begin, end, and are coordinated, where inscriptions accumulate, and where the specific spaces and times we will describe unfold. The record does not merely mirror the bodies it maps, we argue - but neither does it determine them. To emphasize the active role of artifacts without falling into technological determinism, we use the term "mediation" (Latour 1994): the record mediates the relations that it organizes, the bodies that are configured through it.

Following Foucault, we stress that these practices of reading and writing are not only central to the production of a patient's body: they also fulfill a core role in the producti on of a body politic. As the record is involved in the performance of the patient's body Foucault so aptly describes, it is also involved in the performance of the clinic in which that body comes to life. A specific configuration of the body cannot be cut loose from the specific social position different health professionals have within hospitals, and with the type of stories about the work done that can emerge from the records. To strengthen both these points, we make a third argument which departs from the Foucauldian scheme.[2] We argue that different records, different practices of reading and writing are intertwined with the production of different patient's bodies, different bodies politic , and different bodies of knowledge.

In the first section, we look in some detail at a medical record taken from an oncology ward from a Dutch University Hospital. Here we concentrate on the production of patient's bodies. In the second section, we c oncentrate on the way in which the medical record shapes various bodies politic. We explore some more general developments in the area of medical record keeping such as the increasing attention to coding and classification, and the attempts to produce an electronic medical record. In each section we look first at the general process of production of the body/the body politic and then look at the practice of multiplicity (managing the articulation of multiple bodies/bodies politic).

Throughout, we will be using a broad definition of the medical "record" as all written, typed or electronically stored traces of any aspect of patient treatment that has official status within the hospital system and is in principle stored for a period of time (at least equ al to the patient's stay in the hospital; see e.g. Huffman 1990). More often than not, this implies that the "medical record" is not one single object: rather, it is the record the physicians keep in one folder at the outpatient clinic together with the ( also often physically separate) nursing and physician's "in house" record, with the separate forms created and used in the hospital administration offices - and so forth.

This paper is not primarily concerned with practices of reading and writing whic h bring the record to life (see Berg 1996 for this approach); it is concerned, rather, with mapping the configurations it helps bring into being. We will thus not be looking at the real-time articulation work that links the record to the ongoing work, or at the informal organizations that interpenetrate the more formal body politics. These are crucial issues (see e.g. Suchman 1987; Star 1995), but we want to explore here how the structuring of the record speaks to the structuring of the bodies we investig ate. It is a synchronic exercise to explore just how a mundane, boring artifact like a medical record is involved in the politically charged production of human bodies, organizational hierarchies, and selective memories.

The paper is based on two line s of research. Marc Berg has spent several months in different wards in various hospitals in the Netherlands, focusing on the role of artifacts such as records and protocols in the ongoing processes of medical work (see e.g. Berg 1997). Geoffrey Bowker ha s studied the history of medical records and has through a series of interviews and observations traced the development of a nursing classification scheme designed for incorporation within hospital information systems.

PRODUCING BODI ES

Mr. Wood is a patient on the Dutch University Hospital oncology ward. Imagine a newly arrived oncologist, who begins her shift on this ward. She meets Mr. Wood without having his medical record to hand. She sees a middle aged, somber looking man, probably suffering from cancer since this is the oncology ward. Questioning him or the personnel on the ward, she may elicit a story of Hodgkin's disease, which has been treated once before, but which has recurred. Investigating him m ight yield some more clues as to the spread of the disease, the side effects of his treatment, his general condition. Without the record, she might sense the damage the cancer has done, and conclude that the prognosis is poor. Yet she would excuse herself and not take any action before having seen the record. Without the record, she is without memory, without a device to structure her thought; despite all her years of experience, she is barely more able to proceed than the recently graduated resident who stands besides her.[3]

With the record, things are different. This record comprises some 120 pages, producing a fascinating, detailed yet jagged and dispersed memory of the patient. It starts with the temperature list (Figure 1): an unfolding sheet which is structured like a flow-chart, wherein blood pressure, pulse, temperature, medications and so forth are logged. Next is the order form, on which physicians write down any diagnostic steps and changes in treatment for nursing staff to effectuate. Then come the physician's progress notes, starting with a few pre-structured forms which summarize the patient's medical history at the date of admission, and followed by over twenty, unstructured pages of daily notes. These are followed by computer print outs from the laboratory information system (listing rows of numbers indicating outcomes of laboratory tests performed on the patient's blood and urine), results from bacteriological tests and X-rays (altogether over twenty pages), letters written about Mr. Wood, and so forth.

FIGURE 1 HERE

History

The record produces a patient with a medical history: the accumulation of sets of traces configures a medical past for a specific pa tient.[4] The temperature list maps the different parameters against time - the x-axis of this flow chart. Lines divide the sheet into weeks and days, and thinner lines divide the upper, graphical part of the sheet even further into eight-hour periods. Temperature, pulse and respiration rate can be entered along the y-axis. Below this, the y-axis changes its role. It becomes a list of different parameters: tension, weight, specific gravity of the urine , bleeding time (both not filled in on this form), and the medication given. Presenting them in this way clarifies how changes in one are temporally related to changes in another - yet their vertical order is arbitrary. Further below, the y-axis ch anges again: here it incorporates a mathematical operation. The different infusions listed add up to total fluid intake ("totale vochtopname"), followed by the different varieties of fluid loss that result in total fluid loss. These totals are then subtra cted, resulting in the last row on this sheet: the fluid balance.

The forms listing the laboratory results are structured in a similar fashion: columns of figures, each column indicating a time at which the blood was tested. Reading such forms from le ft to right, one enters the past in a most orderly fashion: step by step, day by day, or even hour by hour, the same variable is followed back through time. This produces a linear, stable history; this activity performs the temporality that Foucaul t saw as a crucial innovation of the modern, clinical gaze. It generates a body with a set of variables whose mutual interrelations and deviations can be traced. The body acquires a double continuity: it acquires a past through recurring variables distrib uted throughout an evenly flowing time, and this past itself remains constant over time. Whether a doctor looks at these traces one day or one month later, this past does not change. (As we will see later, this constancy is not characteristic of al l the histories produced in the record, nor does it imply a unified history).

History is also produced by putting new forms (such as X-ray reports) in front of the previous form in that section. This does not produce the graphical oversight that a flo w sheet format provides, but through for example flipping through the bacteriology reports, the doctor again finds herself traveling back and forth through linear time, accelerating and decelerating where necessary. Through compressing several weeks of re ports into a few seconds, the doctor leaves the time zones of everyday medical work and enters the temporal order of pathological processes of Foucault's body: the record produces textbook time. The growth of a tumor as witnessed on successive X-ra ys, the battle between microorganisms and antibiotic regimes, and the rise and fall of blood cells as the chemotherapy does its work - these are not events which take place within the time zones of a working day. Fluctuations in blood cell levels only bec ome meaningful over a several day period, and the "growth of a tumor" unfolds over several weeks. Since the rows of blood cell levels on one page cover some ten days, and the X-ray reports over a two month period usually only cover a few pages, the record elegantly affords this crossing between time zones.[5]

The physician's progress notes and letters, finally, generate a history in a somewhat different way. As the pages of the progress notes fill up, the sequence of e ntries can certainly be read much as the sequence of bacteriology reports or laboratory tests: re-winding time backwards to generate an image of events taking place in a different temporal order. Yet most entries themselves start out with a short "summary " of the case, as follows:

Now 8 days post-reinfusion (of his bone marrow cells)[6]

Last night started amica 2 x 700 (these are antibiotics)

ripera 4 x 4 gr.

Origin? (... of the fever)

(From Mr. Wood's record)

In these summaries, the "current situation" is deduced from other entries in the record, the new laboratory results, information from nurses and the patient, and so forth. The starting point of that d ay's deliberations is formulated in a single phrase uninterpretable out of the context of the other entries. These summaries are read, re-read, re-written in the light of new events, summarized again, and continually constitute a brief patient's history a s relevant at that particular juncture (Berg 1996). Glancing through a few recent pages of progress notes, then, is often the preferred way for a novel physician to "eyeball" a patient's current situation: to build a multi-layered history of pathological processes, diagnostic and therapeutic procedures, intertwined with organizational routines (such as who was the responsible physician when, where was the patient sent to), and so forth.

Geography

The record also produces an anato mical geography. Its different sections detach the blood from its vessels, split its constituents, juxtapose the organs. The case history form separates several "organ systems", as do the nursing history forms; the X-ray reports delineate the "thor ax", the "abdomen", and all the structures that can be seen within these bodily cavities. Orthogonal to the creation of a linear history, here anatomical orders are produced which perform the textbook medical body. The record thus also performs a dislocat ion in space (Latour 1987): it performs an in-vivo dissection, fleshing out a map to the terrain that is hidden under the patient's skin. This map affords a linkage of anatomical realities analogous to the time travel described above. Juxtaposing t he radiographic reports from Mr. Wood with the physical examination and the laboratory results allows the physician to trace the spread of the cancer and to weigh the damage done by the chemotherapy.

But what does it mean, exactly, to say that the rec ord "produces" a patient's history or geography? Of course, Mr. Wood's record is productive of narrative in the banal sense that people tend to read a narrative structure into any set of facts and figures: Boland (1993) has shown how, read by accountants, two different sets of balances sheets can give rise to the most varied conjectures about the personal lives and professional competence of their peers. Yet we want to go further. We are not arguing that the record produces a history or geography b y creating a specific representation of the body, while leaving the "real" patient's body untouched. It is not a matter of merely producing a particular discourse, a specific rhetoric about a body. The medical record is not simply a post hoc depiction of times passed and spaces explored: it feeds into the very constitution of these times and spaces. The medical record is a distributing and collecting device (Berg 1996): work tasks begin and end there. It produces the patient's history by sequential ly demanding that the same measurements be made again and again (according to the density of the x-axis). In order to produce the evenly distributed graphs and tables, these measurements have to be meticulously timed: every day, standard blood test s are performed, bacteriological cultures are taken; and every eight hours, nurses measure the temperature and the pulse. Every day, physicians produce a new summary of the situation which forms the basis for that day's diagnostic and therapeutic interven tions. The end result is not a "medical" history which is clearly distinguishable from the history of Mr. Wood. The production of this representation can only occur with the concurrent transformation of that which is represented. The meticulous organizati onal routines into which the patient is "hooked" totally transform and control the patient's previous time zones - determining when the patient sleeps, structuring the days, and transforming the very experience of the flow of time (Frankenberg 1992; Roth 1979/1963; Star and Bowker 1994). The body is thus re-written in bureaucratic format: the weekly, daily and eight-hour cycles which now structure the patient's time closely match the doctor's and nurses' shifts, the opening and closing hours of the labora tory, and the measurement units of the financial administration (Zerubavel 1979). Again, this is not merely a different, professional "reading" of the body. The body is materially reconfigured - its flesh is part and parcel of the discursive transformatio ns we witness here. In Star and Bowker's terms, we witness a convergence between body and representation: in its production, the representation inscribes itself in the body it represents (1994).

It is this re-written body, subsequently, which i s the site of the diagnostic and therapeutic interventions. At this point, it becomes meaningless to debate whether these interventions address the body "itself" or its representation, since it is in and through this representation that the body "i tself" is known, surveyed and intervened upon. Peaks in the temperature curve will prompt the administration of a certain antibiotic - whose effect subsequently is also monitored in that graph, and in the temporal changes in the laboratory values register ing the organ functions that might be affected by this drug. Not only do the record's pre-formatted time zones inscribe itself into the patient's body: all medical activities are started, followed up and evaluated within the time zones produced by the rec ord. Most of the relevant historical events in the hospitalized patient's life become events triggered by orders written in the record - procedures set in motion by specific forms; therapeutic dosages re-adjusted because of an increase in a certain labora tory variable - as gleaned from the record.

While the patient's time is morphed into the record's time, the patient's geography is rewritten as well. As Hirschauer has so beautifully argued, the surgeon who operates on a body meticulously prepares and carves out the tissues so that they map the anatomical atlas. The atlas does not simply figure as the "ground" of objective knowledge, immediately visible in individual patient's bodies. Rather, this visibility is the outcome of the surgeon's "scu lptured practice": "the proper anatomy of the ideal body is engraved ... into the patient-body" (1991). In an analogous fashion, the record inscribes a geography. The forms physicians and nurses use in their investigations of the patient body invariably d elineate the organ systems. A form used in an intensive care unit, for example, listed the categories under "physical examination" as follows:

general impression

central nervous system

cardial

pulmonal

abdomen

urogenital

extremiti es

Similarly, many laboratory tests are categorized anatomically: "liver function", "renal function", "pulmonary function". In this way, through categorizing observations and substances from the body, the record carves out an anatomy. And it is thi s "geographicalized" body which is subsequently intervened upon: drug dosages are adjusted to correct a decreasing liver function, antibiotics are geared towards a possible focus for infection seen on an X-ray, and tubes and monitor cables are put in plac e to monitor the cardio-vascular circulation. As above, in an ever tightening cycle, the patient's body becomes its representation. Information required by the record or written in the record leads to interventions on the patient's body as it is re presented in the record. Again, the record does not simply describe a patient's body: it structures the way the patient's body is rewritten. The categorization of observations and substances is only a first step in a continuing sequence of events, in whic h the individual anatomical systems become the center for specific interventions by different specialists. Fluids and medication are given to correct urogenital problems, a chest is punctured to restore a deflated lung to its proper form, and postures hav e to be taken to shoot the X-ray right: in mapping the patient, the patient's body is reconfigured so that it matches its map.

The record is of course not the only element active in these rewriting processes. Only as a part of a interlocked se ries of elements does the record come to life; only when linked to nursing and laboratory routines, tubes, infusion bags and cables, the hospital information system and so forth does a network emerge which as a whole performs the transformations de scribed. And the record's structure does not "determine" the nursing routines in any simple way: these routines and the record's current structure have emerged together, mirroring one another and interlocking in historically specific ways (Berg forthcomin g). Yet the record is one turning point in the cycle of inscriptions that circulate through these interlinked entities. It is where the inscriptions end up, are matched and rearranged, and where new inscription-yielding activities begin.

Mul tiplicity

The rewriting processes described above do not produce a single coherent and transparent patient's body, mapped into Euclidean space. Medical practice is not a unitary apparatus "disciplining" the patient's body into some free-floati ng, similarly unitary biomedical entity (Haraway 1991; Hirschauer and Mol 1995). Focusing again on the medical record, we could say that it constitutes multiple histories. It encompasses multiple layers: time flows faster in some sections than in o thers. The temperature curve exists in time measured in hours, while X-ray's exist in time measured in days. Some pages capture a time span of two months, while other pages cannot hold even a single day.

And these histories do not simply fold into one another. The flow chart links several histories by mapping them on a similar x-axis - yet even here the histories remain relatively self-contained. The fluid balance, for example, is produced as a closed system, with a rhythm and an internal logic which is transparent. Daily, a single number results which "represents" the fluid balance of this particular patient. And the temperature curve is similarly clear-cut: it is a graph mapping the evolution of the patient's body temperature over eight-hour periods. Yet the interrelation between these two histories is not self-evident. The only direct linkage (Mol forthcoming) produced by the record is a temporal co-occurence. But how are the peaks related to the numbers? Are they? What is their joint histor y?

Such links are sometimes made in other parts of the record: especially in progress notes. A high temperature reading may be linked to a negative fluid balance record by the story that the former causes dehydration. Yet the progress notes only refer to a fraction of all the data listed in the record. More often than not, the X-ray reports, temperature curve, fluid balance, laboratory values and so forth remain unconnected, in their own time zones. And even if they are connected through the pr ogress notes, another complication of the image of a fully integrated, unitary, Euclidean body arises. Where the flow chart and the series of X-ray reports produce a series of - often separate - linear histories, the progress notes contain non-linear histories. The summaries extract and condense time to match the current situation. Moreover, they continually reconstruct these histories whenever the current situation is changed. The progress notes do not build up linear histories like a flow chart d oes. Rather, a history of the patient evolves in the progress notes, changing whenever the current situation changes. Links are forged to the linear histories evolving elsewhere in the record, but these links are constantly reconstructed, removed a nd retrofitted whenever the current situation alters. What is a crucial temperature peak at time t, may become a singular, non-relevant reading later, and be forgotten in the next summary. And what is described as a sequenced chain of events at som e stage in the progress notes, gets summarized into a singular "event" later on. At the eighth day after the reinfusion of Wood's bone marrow cells, the summary in the progress notes starts out with "now 8 days post-reinfusion", and then states "last nigh t started amica & ripera" (see the fragment of his record above). A whole evening of doctor's deliberations on the right antibiotics to prescribe is summarized in this last sentence. And the eight days between reinfusion and last night flash by uneven tfully, only to slow down "last night", and to stop in the present, where the search for the "origin" of the fever is the order of the day. Time loops and swirls in these reconstructions: post hoc rationalizations after a sequence of dispersed occurrences become prior reasons for a "decision", and episodes get condensed, stretched and rearranged. Although the progress notes do sometimes tie the histories evolving in different sections of the record together, they produce a history which is non-linear, con stantly rewritten, and constantly emergent.

There is, then, no simple mapping between these histories. Mr. Wood's record portrays an array of histories, which do not fold into one another to produce one overall history of a medical body. Rather, the d ifferent histories co-exist, self-contained, sometimes touching and interlinked, but often going their own way. The record performs multiple histories: a series of parallel trajectories, through parallel but also often disconnected time-zones, and a series of statements in which a integrative, historical narrative constantly rewrites itself.

The multiplicity of the patient performed by the medical record is prominent when we look at the patient's geography. Juxtaposing the different parts of th e record does not provide an evenly distributed, three dimensional map of the body. The travel through bodily space is discontinuous, and filled with indeterminacy. The different sections all stand for a series of probes (the blood tests, the radiographic procedures, the physical examination) that are let down into body that, in Foucault's felicitous phrase, hides its inner pathological truth as long as it remains in the "night of life" (1973, p. 165). Different sections perform different geographies: som e organ systems are lumped together in one, and separated in other parts. The blood test forms meticulously separate different cell types, which are lumped together in the fluid loss forms (all under the category "blood"). The bacteriological forms distin guish different sites from which cultures are taken: the urine, the blood, the throat, the prepuce, the feces. The X-ray reports speak about the thorax (including the lungs, the heart) and the abdomen (including liver, spleen, pancreas, and so forth). And sometimes different geographies are performed on one form: the "organ systems" listed on the intensive care's physical examination form (see above), for instance, mapped out regions (the abdomen, the extremities), functional systems (cardial, pulmonal, c entral nervous system) as well as an overall "others" category (general impressions).

Here as well, the different geographies do not unequivocally fold into one biomedical body. There are spaces that remain empty, and there are spaces that are filled with competing tales. A differentiation in "regions" cross-sects a differentiation in "fluids" (blood, urine), which cross-cuts a differentiation in "functional systems"; some "functional systems" might be grouped in one "region", while others (e.g. the c ardio-vascular system, or the nervous system) might not. How are these related? How is what is seen on the X-thorax related to the "cardial system" as felt and heard during the physical examination? Does fluid loss through blood loss result in a loss of e quivalent amounts of the different blood cell types, or does bleeding result in a depletion of some cells more than others? As above, these links are not self-evident. They may be created, in the record, through an explicit juxtaposition and interpretatio n of these observations in the progress notes, or through additional examinations (a more frequently repeated series of blood tests, for example). The point is that these links are often not made; that they are absent, thus leaving the geographies in thei r unconnected realms (Mol forthcoming). And when such attempts are made, the geographies may well contradict each other. The blood might seem infected according to the changing ratio in blood cell types, but not according to the bacterial examination of t he blood; while the physical examination may carve out lungs that are "clean", an X-ray may simultaneously show lungs that are "massively infiltrated". Again, the record does not contain mechanisms for preventing such gaps and conflicts. It performs multi ple geographies, sometimes linked, sometimes oblivious to one another, sometimes conflicting.

MULTIPLE BODIES POLITIC

The record is not only crucial for the planning of interventions and the coordinati on and interaction between care providers. Other parties are also highly interested in the record. Insurance companies want more information about the indications for interventions, researchers want data on the prevalence of combinations of symptoms and t he efficacy of therapies, and governments and hospital boards want to know how many days patients of a specific diagnostic category are hospitalized - and for what reasons. All these parties claim that the patient record can deliver the information they r equire (see e.g. Ball and Collen 1992). In these discussions, the record is often portrayed as a "repository of information" (Dick and Steen 1991), as a vast treasure of facts which only needs to be tapped into. In such a view, the building blocks, the "facts", can only be more or less "adequate" and "complete".

One dream of an electronic medical record from the 1970s was that such a record would provide an unmediated picture of doctors' thought, and produce an untainted, complete, error-free d ocument:

To put a secretary or a computer technician between the physician and the computer is defeating the purpose of a direct natural language physician-computer conversation. If a girl has to interpret what you have said and then put this in s ome form that she thinks the computer will understand, I think we increase the error rate. More important, you prevent the physician from talking to the computer. When the technology becomes sophisticated enough and the expense becomes low enough t hat the physician-computer interface should be direct and in a natural language, there should be no girl, no mark sense card, no keyboard or any other type of artificial instrument between him and his computer. This is why a light pen was devised. It look s like a pencil, it has the physical form of a pencil. The physician has been using the same instrument since he was three years old. (Anderson and Forsythe, 1970, p. 257)

Current, "paper" medical records are harshly criticized precisely for not p roperly reflecting what has happened, and what physicians think:

The medical record is an abomination. ... It is a disgrace to the profession that created it. More often than not, the chart is thick, tattered, disorganized, and illegible; progress notes, consultant's notes, radiology reports, and nurses notes are all co-mingled in accession sequence. The charts confuse rather than enlighten; they provide a forbidding challenge to anyone who tries to understand what is happening to a patient (Bleic h 1993).

The dream of the unmediated record dates from the earliest medical computing applications such as MUMPS (Barnett, 1975) - and has of course never been realized. There are many steps between "what happened" and the final record of t hese events: the summarizing processes described, but also the often invisible but crucial pool of typists who transform the physicians' taped reports into legible, grammatically correct and understandable reports. And most medical records work in hospita ls is actually done by the equally invisible yet crucial medical records professionals - there are some fifty schools in the States training people for bachelor's degrees in Medical Records Administration (Huffman 1990, p. 53). Such professionals signific antly change aspects of the record. For example, Ann Fagot-Largeault (1989, pp. 139-140) traces the mediation process for French death certificates. She cites cases where, like doctors, coders (who take the medical history and apply it to the death certif icate) have their own favorite codes - leading to an "epidemic" of "diffuse intravascular coagulation" in the 1970s (p. 145). These mediations add up: she cites a study from 1970 in Stockholm, where it was shown that the final cause of death was wrong (as demonstrated by autopsy; itself of course a specific mediation) in 45% of cases for "probable" causes of death and in 25% of "certain" causes... (p. 145). It is estimated in France that about one in four certificates filled in by doctors need to be "re-i nterpreted"; with doctors agreeing with the coder's choice only between 80% and 95% of the time (p. 218).7 The image of the record as a more or less adequate representation of facts and events is also problematic since, a s we have seen, the record feeds into the very constitution of these facts and events. We have seen how the record contains many emergent stories about the patient's body: both its temporality and geography. Any one implementation of the record actualizes one type of story; and a set of implementations employing a standardized story format (it is easier to tell some stories than others in the record) inscribes its specific time and space in the patient's body. In this section, we shall see that the same a nalysis can be made of the organizational work done by the medical record.

What do medical records record? The record traces what Strauss and his co-workers have called the "illness trajectory": the course of the illness, the total organization of work done over that period and the impact of that work on those involved (1985, p. 8). In this section, we will take one cut through this totality, arguing that medical records always describe past action in the context of a set of organizational arrangements . It would be meaningless to cite a temperature reading from a record out of the context of what that reading entailed organizationally - a high reading in this kind of chart at this point entrains this kind of response, and so forth. The medical record o f its nature only exists within a certain organizational context: as do all records of past action. But it at the same time feeds into the reproduction of these organizational contexts: the act of recording both enacts the organization that deals with the phenomenon and creates the past within which the phenomenon has occurred. Medical records produce a series of well-defined narrative structures; each of which reflects and produces a particular kind of social organization. We want to look at some of thes e structures to elaborate how the record is involved in the reflection and production of different body politics; how the records are involved in the very production of the descriptions under which they function.8 Here as well as above, the crucial observ ation is that the stories the record produces are not mererepresentations of the work - they are a core, constituent part of it. We will concentrate on three dimensions of the organizational life of medicine in turn: institution al (legal and insurance arrangements), work practice (the organization of the clinic) and processes of professionalization (nursing); at each point we shall be concerned with the deployment of medical records in building a "body politic".[9] Finally, we will - making the same move as in the previous section - discuss the question of multiplicity: how one infrastructural form (the medical record) can produce a range of configurations of these bodies politic.

The body of the law

The medical record is one of the records that produces the legal entity the citizen-in-the-state (Ewald 1986). In order for the record to work, the object of the story has to be a registered entity (the need for registration often being the need for medical attention). Here as elsewhere the primary issue is not whether the records tells a "true" story; though in this case the act of registering in a sense makes it so - it is constitutive of citizenship.H utchins (1995) reminds us of this in the case of navigation: the primary motive for the rich set of records kept by every ship is that the captain needs to be able to show, in case of accident, that everything possible was done.[10] And in a textbook on medical records, the first thing that is said about medical data entry is that these "formulate a legal document that demonstrates the following:

* the patient has a need for service

* the correct service is provided< p> * the service is provided in the proper manner by the proper person". (Waters and Murphy 1979, p. 54).

In current medical work, this legal story is simultaneously the story produced for the insurance companies' interest: it legitimates the actions p erformed as well as their costs. These are particular kind of stories: the legal record should not contain inconsistencies or lacunae (Bowker 1994). It should not refer to things which "just happened". Everything has to be explicable and justifiable. If t he patient died it was despite the full operation of due process; if a test was performed it was because there was a clear-cut medical indication. The outcome of the legal record (the moral of the story) is the meta-affirmation that the medical sys tem is fully penetrated and underwritten by the legal system: that the doctors did everything that they could be expected to, and that all these affirmations were just.

Typically, these "preferred" accounts centralize the physician's agency as core d ecision maker, and reduce the role of the nurse to the provision of primary data and the execution of the doctor's plans. Although nurses spend far more time with the patient, their presence in the medical record is limited to filling in the temperature l ist; whilst the final story emphasizes the logical work (rational reconstruction) of physicians, who are geographically much more removed from the patient. The story that is told over and over again is that patients are cured by doctors' actions; and that doctors' actions reflect a rational thought process. In producing these stories about doctors as heroes and patients as objects, the record configures chains of delegation that render the doctor the representative of the entire heterogeneous assembly com prising the hospital. S/he is performed as the head of the organizational work comprising the crafting of the patient's body, which is done according to the logic of textbook science. This "repository of information", in other words, embodies a differenti al evaluation of whose time counts, whose information matters, and whose work deserves to be mentioned and made visible (Egger and Wagner 1993; Frankenberg 1992; Star 1991).[11]

How do these preferred accounts evolve? It would be wrong to assume that medical personnel "lie", or that there are "cover ups" going on somewhere. Preferred accounts emerge from the record only gradually, through small steps and reconstructions which by themselves are just ordinary moments in the ongoing work. The ordering and polishing of accounts is a natural process - it is a prerequisite for the smooth progression of complex, interactive work processes (Garfinkel 1967).

One of the active elements in the production of such accoun ts is the medical record (Berg 1996). The records lead to a rational reconstruction of the past - just the same kind of rational reconstruction that has been so successfully deconstructed within science studies (see for example Latour and Woolgar 1986). T he record is instrumental in producing such stories that tell "rational", "textbook" narratives about a patient entering a hospital with complaint X, being diagnosed with Y, treated with Z, and who was discharged with or died from Q. Its very forms are st ructured towards such a sequence: the complaint should be filled in first, then the diagnosis, and then the therapy. The forms suggest the reasoning process which should have taken place: the conclusions are logically derived from the data gathered (Barrett 1988).

In addition, the record is the very place where a public account of "what has happened" is created. It is when writing into this potential source for retrospective inspection that physicians and nurses construe narratives that align w hat actually happened with what should have happened - no matter how insignificant these occurrences may seem (Garfinkel 1967, pp. 197-207; Hunter 1991). If a patient has been hospitalized for several days, for example, nurses may omit measuring the blood pressure and just fill in yesterday's measurement in today's column (see Figure 1). Likewise, residents often ask nurses what to prescribe while they complete the order form in the regular fashion: as if it is they who have told the nurses what to do (cf. Hughes 1988). The same phenomenon occurs in and through the summaries that are continually being produced. In this process, details are omitted, and the story is simplified and retold in ways that fit the present situation at hand. This results i n an increasing stylization of past events into a standard cannon: a sign leading to a diagnosis leading to a therapy leading to an outcome. A sentence like "admitted with Hodgkin, now 8 days post-reinfusion" effectively sets the focus of the current atte ntion. Yet in doing so, it also smoothes over any diagnostic uncertainties which might have played a role, deleting the deliberations that went into the selection of this therapy and Mr. Wood's fears and anxieties.

Finally, all this adds to the peculi ar feature of written text that once written, it tends to have a privileged position vis a vis other recollections of these events (see Clanchy, 1993 for the historical genesis of this privilege). Wherever it travels (from the audit committee to the insur ance inspector's desk to the court room), it becomes the trace to the "original event". As Smith aptly sums these issues up, accounts enter "document time" once they are written: "that crucial point at which much if not every trace of what has gone into the making of that account is obliterated and what remains is only the text which aims at being read as `what actually happened'" (1974, p. 260).12 Legal concerns are not the only relevant cross-cutting interferenc es here: the fact that these documents are also (formally and informally) used for quality control purposes and for the evaluation of colleagues and trainees yields similar consequences. And these are not new interferences: the first official, early twent ieth century attempts to standardize and improve medical record keeping in the USA did not primarily derive from the urge to improve the primary care process. Rather, they were driven by the absence of means to decide upon certification and accreditation of hospitals and individual doctors (Stevens 1989; Waters and Murphy 1979, pp. 4-5). From early on, then, the record was a crucial element in the generation and reproduction of this complex body politic.

The body politic of the clinic

Analogous to its involvement in the production of a patient's body, the medical record is involved in the production and reproduction of the body politic of the clinic. First, it plays a core role in the ongoing reproduction of the hospi tal's temporal structure. Triggering sequenced measurements produces both a graph in a record and circumscribed, cyclical work tasks for nurses. The unstructured progress notes forms require physicians to regularly summarize the current state of af fairs, and thus create an organizing structure for their work days. Without the record's core role as a distributing and collecting device, the complex temporal organization of this body politic could not exist. The record lists which actions are undertak en and which actions still to undertake, when which requests or interventions will yield outcomes, and who is responsible for the completion of such a task. The flow sheet's medication and infusion lists, for example, function for both nurses and physicia ns as a list of what has been administered, and as a means to see at a glance how some crucial measurements have been affected by these interventions. In addition, they also organize the work: the ward secretary can see what type of medication might need to be ordered, and nurses can see when to give what, and can subsequently add the infusion dosages for the calculation of the fluid balance. The record, in other words, affords a highly complex, temporally distributed division of labor.

And it thereby also maps a complex geography of the hospital. The record discussed here is a so-called "source-oriented" record: by far the most common means of record keeping. It entails ordering the different sections in the record according to the different institut ional sources the data are derived from: nursing notes, doctor's notes, laboratory results, bacteriology reports, consultations of different specialties, and so forth.[13] Again, this does not yield a map which merely "rep resents": it feeds into the reproduction of these geographies. Having their own forms and sections within the record, the nurses are separated from the physicians - both physically and intellectually. Notwithstanding other alliances or differences, the re cord clearly differentiates the bacteriologists vis-à-vis the radiologists, and sets them both apart, as two isolated groups, from the specialists in charge.

The record, moreover, feeds into the reproduction of the hierarchies between these geo graphical sections. The treating physicians are on top: they write the central pages of the record, defining the outlines of the illness trajectory and incorporating and summarizing information coming from all other sources. They have unlimited access to all the sections, but only write there when the overall course is set. And the other sources are not equal either. Specialists called in for consultation will usually fill in their own, yellow consultation forms, but they are allowed to write their findin gs in the progress notes. No other discipline is allowed to do this. In addition, the hierarchy is evident in the way just how much a form is pre-structured. An unstructured forms leaves it to a named author to create order through the production of a nar rative, and it does not exert demands on the content of these narratives by listing pre-formatted labels - which may remain visibly empty. The physicians' progress notes are totally unstructured, and the consultancy forms have a few, large sections with l abels such as "findings, conclusions and advice" - all designed for free text. The radiographic reports are more brief, but the radiologist is free to describe the image as s/he pleases, as long as it is followed by a conclusion which answers the treating physician's request. The bacteriologist, however, is limited to saying "no growth", "a little growth" or "much growth", and the laboratory workers and nurses can only produce the numbers which end up in the pre-formatted graphs and columns. As Hunter phr ases it, "the hierarchy of disciplines is very alive in the chart" (1991, p. 88).[14]

The record not only plays a generative role in the organization of the clinic: it also legitimates this organization's design. As we have pointed at above, the record tells the story of the work this body politic performs. One way of structuring the record has been to make every record a description of a working organization operating at peak efficiency. Thus record keepers are advised that the purpose of the record is to tell a story that begins with admission, describes an every more precise diagnostic process, goes on to treatment and ends with discharge - or, even more "complete" from the medical record keepers' perspective , death (Waters and Murphy 1979, p. 97). In this case the outcome of the record is again the meta-affirmation that allopathic medicine works - and, ipso facto, that the sites where medicine is performed work. Every record - as individual as may be - is, t hrough its very form, testament to this assertion.

The professional nursing body

The fact that the chart maps - and thus feeds into - the hierarchy of disciplines in medical work has not gone unnoticed by these disciplines themse lves. Turning to our third example, we encounter a professional group striving to enhance its status through attempting to change these record-keeping practices. Here, the constitution of a body politic in and through the medical record has become a site of active contestation. The record-keeping practices at stake concern the nursing classification scheme designed for incorporation within hospital information systems. The scheme, NIC (Nursing Interventions Classification) divides up all of nursing work i nto a series of well-defined actions which are then to be coded (allowing nursing administrators to track the work of their staff). The group at Iowa University School of Nursing published a first edition of their system in 1992, and a revised and expande d version came out in 1996 (Cohen et al. 1991; McCloskey and Bulechek 1996).

NIC itself is a fascinating system. Some categories, like "Bleeding Reduction - Nasal - 4024", are on the surface relatively obvious and codable into discrete units of work practice to be carried out on specific occasions. But what about the equally important categories of hope installation and humor? "Hope Installation - 5310" includes the subcategory of "Avoid masking the truth". This is not so much something that nurses d o on a regular basis, as something that they should not do constantly. It also includes: "Help the patient expand spiritual self". Here the contribution that the nurse is making is to an implicit lifelong program of spiritual development. With respect to" Humor - 5320", the very definition of the category suggests the operation of a paradigm shift: "Facilitating the patient to perceive, appreciate, and express what is funny, amusing, or ludicrous in order to establish relationships"; and it is unclear how this could ever be attached to a time line: it is something the nurse should always do while doing other things. Further, contained within the nursing classification is an anatomy of what it is to be humorous, and a theory of what humor does. The recommen ded procedures break humor down into subelements. One should determine the types of humor appreciated by the patient; determine the patient's typical response to humor (e.g. laughter or smiles); select humorous materials that create moderate arousal for t he individual (for example picture a forbidding authority figure dressed only in underwear); encourage silliness and playfulness and so on to make a total of fifteen sub-activities: any one of which might be scientifically relevant. A feature traditionall y attached to the personality of the nurse (being a cheerful and supportive person) is now attached through the classification to the job description as an intervention which can be accounted for. These process categories fit into the same numbering schem e as more punctual categories such as "Bottle Feeding - 1052" or "Suturing - 3620" - once in the record, there is no distinction between process and single interventions.

The Iowa group (the kernel of whom are teachers of nursing administration) made essentially three arguments for the creation of a nursing classification, initially to nursing informaticians and thence to the hospital information systems community. First, it was argued that without a standard language to describe nursing interventions , there would be no way of producing a scientific body of knowledge about nursing. NIC in theory would be articulated with two other classification systems: NOC (the nursing sensitive patient outcomes classification scheme) and NANDA (the nursing diagnosi s scheme). The three could work together: one could perform studies over a set of hospitals employing the three schemes in order to check if a given category of patient responded well to a given category of nursing intervention. Rather than this comparati ve work being done anecdotally as in the past through the accumulation of experience, it could be done scientifically through the conduct of experiments. The second argument for classifying nursing interventions was that it was a key strategy for defendin g the professional autonomy of nursing. The Iowa nurses are very aware of the literature on professionalization - notably Schön (1983) - and are aware of the force of having an accepted body of scientific knowledge as their domain. The third argument was that nursing, alongside other medical professions, was moving into the new world of computers. As the representational medium changed, it was important to be able to talk about nursing in a language that computers could understand - else nursing work would not be represented at all in the future, and would risk being even further marginalized than it was at present.

The intervention in turn can then be linked by nursing academics to nursing outcomes in a series of clinical studies. The outcome o f the records will be the integration of nursing work into the medical establishment . The expressed fear of professional leaders is that without representation in the information system, the profession of nursing will become ever more marginalized. So th e record is producing a patient - as we stressed in the first part of this work - but it is also producing an organizational result. For the nurses, the record constitutes, then, building blocks for a body of knowledge, which in turn loops back int o the hierarchy. In current storage practices, nurses' notes are frequently separated from doctors' notes, and are often destroyed immediately the patient leaves the hospital (Huffman 1990). A professionalized form of reporting would create a continuing r ecord of their activities - which could feed into the discipline of nursing science and be used to unseat the current hierarchy.15 Nurses who met to develop a "nursing minimum data set" (modeled on the successful medical minimum data set) to be produced by all hospitals in their medical records argued that: "the most frequently focused upon language is that developed by medicine and the numerous natural or physical sciences that contribute to medicine's scientific knowled ge base. This language is focused upon most frequently by nurses, because of medicine's long history of dominance in nursing, its relative specificity about physiological phenomena, its apparent measurability, and its familiarity. It is also valued for it s social status as a scientific language, used by a politically powerful professional group, physicians" (Kritek, 1988, p. 25). It is not enough for nurses to accurately record what they do - they must record it in a language acceptable to physicians and hospital administrators, and productive of a new, higher status. For this reason, "leech treatment" was not initially recognized as a NIC category - even though it was indeed a treatment and was indeed carried out by a number of nurses (Neumann, forthcomi ng). The nursing record had to be seen to be "professional" in order for it to do its work. That work is, in part, the modeling of nursing on the example of medicine; itself modeled (with the help of the medical record, as we saw above) in the image of pu re science.

Multiplicity: Interoperability and Infrastructure

It is turning into a complex thing, our medical record. It produces multiple bodies and underwrites a number of ever-shifting organizational configurations. For the nurses, the medical record is a crucial tool in their organizational struggle for status at the same time as it is a record of what happened to the patient, a legal document, and a tool which organizes their daily work. It must be able to do all these jobs at once - and for all those working with it. In the multiplicity of context the record is drawn into (and helps to constitute), materially different objects are at play: most physicians never get to see the coding forms used by t he coders; most record administration coders know little of the elaborate nursing coding schemes; most nurses primarily draw on the nursing record during their work, and so forth.

As we have seen, medical records are involved in the production of mult iple bodies politic. Consider the legal reading of the record that we have discussed. In this reading, the legal narrative is the chief concern, and some features of the record are activated as relevant for the story to be properly told. It may not matter for the nursing record whether or not a given temperature reading was taken, but this may make a large difference legally (as indicative of contributory negligence). So medical records do not have to do a single job, they have to be configured in such a way that a set of different organizations are serviced by them. To borrow the vocabulary of information science they have to serve as infrastructure (Star and Ruhleder, 1996) to a set of organizations. They have to do this in such a way that these organizations can use the medical records in conjunction with other records that they generate. Thus for example you want the medical record to store information about ethnicity in a way which makes sense to epidemiologists and to public health off icials - so that in turn these groups can deploy the records in conjunction with other records that they produce to track diseases. Again to borrow the vocabulary of information science, we can say that the medical record needs to guarantee interoperab ility. In this section, we shall see that fashioning of medical records into infrastructural tools guaranteeing interoperability involves a series of organizational alignments of great political and ethical importance. Just as for the patients in thep revious section above, there is a real multiplicity of organizations produced by the records. Just as for the patients, this conflicts with a rhetoric of singularity (the convergence of traces on a single ideal patient temporality and geography and on a single ideal organizational context).

A recent history of medical informatics notes that: "... by the 1980s, the provision of health care in the United States had three dimensions of integration for patient care information: local integration of information in hospitals or in physicians' offices; vertical integration of information between affiliated hospitals and medical offices; and horizontal integration among associated hospitals, clinics, and community health and welfar e support groups" (Collen 1995, p. 82). Each of these forms of integration (which cannot be as neatly separated off from each other as Collen implies) entail a different form of interoperability. The record has to be aligned with the needs of the organiza tions associated with them. This involves, for one thing, matching of the different timetables and geographies the records produce. Roth (1979) discusses the difficulty of a tuberculosis patient traveling to a new sanatorium without there being an agreed- upon language for describing the patient's condition; the NIC nurses despair of ever producing nursing knowledge without comparability (Timmermans et al. forthcoming). But this integration also extends to relationships with health care providers, public h ealth officials, the World Health Organization (which maintains the International Classification of Diseases), the legal profession and so forth: as we have pointed out, the record is increasingly seen as a vital source for these "third parties" as well. With the development of computer based records, a further dimension is introduced by the needs of the programmers. Thus at a meeting discussing the development of the NIC classification, for example, nursing informaticians protested the existence of "nest ed classifications" within NIC - that is to say an intervention containing another within it. They argued that this made for difficulties in programming the appropriate screens on the electronic nursing record.

With the extension of the record's reach through different times, spaces and into novel domains, much work is done on the record to ensure that these multiple configurations continue to be possible. Much of this work seems "purely technical"; appears far removed from organizational or po litical issues. Consider the sets of committees that have been involved with standardizing medical informatics:

The American Standard for Testing Materials (ASTM) subcommittee E31.12 subcommittee framed nomenclatures and medical records. In 1988, A STM sub-co E31.12 published standards for patient discharge and transfer data. Health Level Seven (HL7) an organization made up of vendors, hospitals and consultants was organized in 1987 to develop interface standards for transmitting data between applic ations that used different computers within HISs. The message content of HL7 was to conform to the International Standards Organization (ISO) standards for the applications level 7 of the Open Systems Interconnection (OSI) model; the HL7 standard used the same message syntax, the same data types and some of the same segment definitions as ASTM. The Medical Data Interchange (MEDIX) P1157 committee of the IEEE, formed at the SCAMC 1987 was also developing a set of standards, based on the ISO application-lev el standards, for the transferring of clinical data over large networks from mixed sources, such as from a clinical lab and a pharmacy, for both intra- and interhospital communications. (Collen 1995, p. 109)

Each of these interconnections was the product of a continuing effort to maintain and stabilize an infrastructure. Janet Abbate (forthcoming) has demonstrated how the OSI standards are themselves the fragile outcome of a series of negotiations between a computing industry wanting to retain the proprietary edge of a closed system, government regulators, public pressure groups and so forth. As you spiral down into the infrastructure you get beyond a particular database in a hospital to the technical standards (OSI) underwriting that database; an d as you follow OSI in development you witness a titanic struggle between company interest and the public good being written in code. This is typical for infrastructure: the "harder" you go, the closer you get to organizational and political concer ns (see also Hanseth et al. 1996, and see Carlson 1991 who makes this point brilliantly for innovation in the electrical industry).

All these standardization efforts do not do away with the continuing need to balance the needs exerted within the diffe rent networks that interconnect through the infrastructure (Bowker and Star 1994; Timmermans and Berg 1997). The ever unattainable ideal of pure, quantifiable knowledge remains a driving organizing force - even when it is patently absurd. Consider the cas e of fluid balance - this is precisely measured to the milliliter in intensive care units, and then a fudge factor out of all proportion to the data is added in for sweating (which cannot be so simply measured). Why is this ideal so successful in organizi ng infrastructures and guaranteeing interoperability? One answer is that it generates alignments. If we are all striving in the same direction then, whether or not the direction itself is a good one, our sets of fudge factors and uncertainties will tend t o align with each other. An analogy here would be to the story of the emperor's new clothes. If everyone seeks to describe the world in ways acceptable to the others around them, then the emperor will be seen to well dressed - whether or not he is - if on e person is out of alignment, the whole may crumble. This, we would argue, is one reason why formal records entail such deep organizational discipline.

The nurses producing NIC believe deeply that the current system of accounting inscribed in the hosp ital information systems is wrong-headed. Thus in a June 1995 meeting, co-leader of the NIC team JoAnne McCloskey mentioned the rise of activity-based accounting, meaning that individual activities had to be costed. She went on to effectively deconstruct this in much the same vein as Dick Boland (1996) has done from within the field of Management and Information Science. Nurse practitioners, she noted, see patients for longer than doctors - since they are teaching, and preventing re-occurrence etc. Until you pull out productivity against performance one cannot see this. This was also a problem, she noted, for women physicians (who see patients longer, and need to see them less as a result - so on a productivity measure it looks like they are working more slowly). However the productivity measures being used were based on the factory measure of how many widgets an hour you produce. Crucially, she went on to say that: "Nursing needs to speak the language used elsewhere. This is why we need to use a restrict ed definition of productivity". In our terms, she is pointing to the need for interoperability of the nursing records with other medical records. And indeed in the NIC email list she soon floated an attempt to pin times to interventions (difficult in the case of emotional support and active listening; but not obvious even for bowel incontinence care...). And indeed in order to speak the language of accountancy well enough to justify a new nursing slot, the organization had to be recorded as having worked in a particularly well defined set of temporal slots. Here is a sample argument that uses the assignation of time slots to NIC interventions to justify the creation of a new nursing position in a given hospital. This argment was presented to nursing admin istrators and nursing information specialists at the same meeting:

Position justification

Teaching/ discussion/ process time

individ. teaching 8

prescribed medic. 15

prescribed procedure 10

treatment 30

63 mi nutes by 620 new patients per year = 39,060 minutes or 651 hours.

651 hours divided by 2000 = .3 FTE. (FTE being "full time equivalent").

What is happening here is that a key player in the production and maintenance of the medical record - the nurse - does not believe in the validity of the traces that they are producing of their own work in the record; and yet they have to both write records as if such traces were true and write them well enough so that nursing should be seen to be a key playe r in the efficient scientific treatment of patients (rather than a prime candidate for cost cutting). What a strange irony! It is, we would argue, an extreme example of a very common phenomenon. Consider the case discussed at length by Young (1995) and Kirk and Kutchins (1992) of psychoanalysts who in order to receive reimbursement for this procedures need to couch them in a biomedical language that is anathema to them, but is the lingua franca of the medical insurance companies. The nurses have po tential allies - there are radical accountants, radical doctors, critical medical informaticians: but the need to align is immediate; the risks of disappearing forever from the medical record too great to make the risk of challenging the common language w orthwhile. Local use of the record underwrites organizational multiplicity (nurses will continue to be concerned with process; psychoanalysts with the id) at the same time as the constrained form of the record evokes a single large medical organization wi th guaranteed interoperability of its "departments" (nursing, epidemiology, public health, medical insurance and so forth) - just as the patient's body is at once conjured into a single time and space while remaining locally multiple.

Infrastructures underwrite alignment of the organizations that they are infrastructural to; and a key feature of this alignment in the case of the medical record is the evanescent ideal of the scientific operation of medicine. As Latour (1988) has argued in an analysis of Pasteur, much of the power of science comes from such alignment. This alignment is not ever completed or entirely successful: however large techno-economic networks cannot exist without it. The medical record produces organizations which can configure their own records such that a privileged set of stories (due process; medical science; nursing efficiency) can be told again and again.

Conclusion

The medical record, it has become clear, is a complex object, and a fascinating a nd crucial focus for sociological research. It feeds into that what it merely seems to represent; it is a crucial element in the constitution of the patient's body (inscribed in a set of spaces and times) and of the hospital as a formal organization (equa lly inscribed in a set of spaces and times). It is a site where both the constitution of the patient and the constitution of the hospital worker is at work: the bodies and bodies politic they inhabit are reinscribed into novel places, broken apart in new ways, and configured into spaces and times out of their control. Seen in this light, as a site where multiple stories about patients and about organizations are at stake (including the interoperability between these stories), the medical record becomes hi ghly relevant both analytically and politically. There is a strong current "push", for example, towards the implementation of the electronic medical record - and when the record is seen as an innocuous storage device, the appropriate implementation is oft en seen as a "technical problem", or as a matter of finding the "appropriate interface". When it is acknowledged that the medical record is interwoven with the structure of medical work in fundamental ways, that different medical record systems embody dif ferent notions of how work is organized, different modes of configuring patient bodies, and so forth, we are in a position to better understand and intervene upon the issues that are at stake. We are reminded of Latour's (1987) observation that science is bureaucracy or Serres' equation of law and science (1987): the site of the creation of organization and body is the bureaucratic work of creating inscriptions. To end where we began: this site thereby becomes a key site for the arrogation and exercise of power.

Notes

Acknowledgements

Marc Berg would like to thank the Dutch Association for Scientific Research (NWO) for making this research possible. We thank the two anonymous referees, Emilie Gomart, Leigh Star an d the members of the Xerox PARC Work, Practice and Technology group for their valuable comments.

References

Abbate, Janet. forthcoming. "On the history of OSI." The Information Society.

Abbott, Andrew. 1988. The syste m of professions. An essay on the division of expert labor. Chicago: University of Chicago Press.

J. Anderson and J.M. Forsythe, "Information Processing of Medical Records." In: Proceedings of the IFIP-TC4 Working Conference on Informatio n Processing of Medical Records, Lyon, 6-10 April 1970, Amsterdam: North-Holland Publishing Company, 1970

Atkinson, P. 1981. The Clinical Experience: The Construction of Medical Reality. Farnborough: Gower.

Atkinson, P. 1995. Medical T alk and Medical Work. London: Sage.

Ball, M. J., and M. F. Collen, ed. 1992. Aspects of the computer-based patient record. New York: Springer.

Barnett, G. O. 1975. COSTAR Computer-Stored Ambulatory Record. Boston, MA: Massachusetts General Hospital.

Barrett, R. J. 1988. "Clinical writing and the documentary construction of schizophrenia." Culture, Medicine and Psychiatry 12: 265-99.

Berg, M. 1996. "Practices of Reading and Writing. The Constitutive Role of the Patient Record in Medical Work." Sociology of Health and Illness,. 18: 499-524.

Berg, M. 1997. Rationalizing Medical Work. Decision Support Techniques and Medical Practices. Cambridge: MIT Press, in press.

Berg, M. forthcoming. "Urine Containe rs and Infusion Pumps: Some Tools for a Sociology of the Formal." Science, Technology and Human Values.

Berg, Marc, and Annemarie Mol, eds. forthcoming. Differences in Medicine. Unraveling Practices, Techniques and Bodies.< p> Bijker, W. E., and J. Law, ed. 1992. Shaping Technology - Building Society. Studies in Sociotechnical Change. Cambridge: MIT Press.

Bleich, Howard L. 1993. "Lawrence L. Weed and the Problem-Oriented Medical Record." M.D. Computing 10: 69-72.

Boland, R J. 1993. "Accounting and the Interpretive Act." Accounting, Organizations and Society 18: 125-46.

Boland, Richard J., and Ulrike Schultze. 1996. "From work to activity: technology and the narrative of progress." Pp. 308-325 i n Information Technology and Changes in Organizational Work, edited by Wanda J Orlikowski, Geoff Walsham, Matthew R Jones and Janice I DeGross. London: Chapman & Hall.

Bowker, G. 1994. Science on the run. Information management and indust rial geophysics at Schlumberger, 1920-1940. Cambridge, MA: MIT Press.

Bowker, G. forthcoming. "Lest we Remember: organizational forgetting and the production of knowledge." submitted to Accountancy, Management and Information Technology.

< i>Bowker, G, and S.L. Star. 1994. "Knowledge and infrastructure in international information management: problems of classification and coding." Pp. 187-213 in Information acumen: the understanding and use of knowledge in modern business, edite d by L Bud. London: Routledge.

Carlson, W. Bernard. 1991. Innovation as a Social Process: Elihu Thompson and the Rise of Gernal Electric, 1870-1970, Cambridge: Cambridge University Press.

Clanchy, M. T. 1993. From memory to written record, England 1066-1307. 2nd ed. Oxford: Blackwell.

Clarke, A. E., and J. H. Fujimura, ed. 1992. The right tools for the job. At work in twentieth-century life sciences. Princeton, N. J.: Princeton University Press.

Cohen, Marlene Z., Thomas K ruckeberg, Joanne C. McCloskey et al. 1991."Inductive Methodology and a Research Team." Nursing Outlook, 39: 162-165.

Collen, Morris E. 1995. A History of Medical Informatics in the United States, 1950 to 1990: American Medical Informatic s Association.

Cussins, C. forthcoming. "Ontological Choreography. Agency for Women Patients in an Infertility Clinic." In Differences in Medicine. Unraveling Practices, Techniques and Bodies, edited by M. Berg, and A. Mol.

Dagognet, F. Le catalogue de la vie: étude méthodologique sur la taxinomie, Paris: Presses Universitaires de France, 1970.

Davis, F. 1956. "Definitions of time and recovery in paralytic polio convalescense." American Journal of Sociology 61: 582-88.

Davis, K. 1986. "The process of problem (re)formulation in psychotherapy." Sociology of Health and Illness 8: 44-74.

Dick, R. S., and E. B. Steen, ed. 1991. The Computer-Based Patient Record: An Essential Technology for Health Care . Washington, D.C.: National Academy Press.

Egger, E., and I. Wagner. 1993. "Negotiating temporal orders: the case of collaborative time management in a surgery clinic." Computer Supported Cooperative Work 1: 255-75.

Ewald, F. 1986. L' Etat providence. Paris: B. Grasset.

Fagot-Largeault, A. 1989. Les Causes de la Mort: Histoire Naturelle et Facteurs de Risque. Paris: Vrin.

Foucault, Michel. 1973. The Birth of the Clinic: An Archaeology of Perception. New York: Vi ntage Books. (Orig, edition 1963).

Frankenberg, R. 1992. "`Your Time or Mine': Temporal Contradictions of Biomedical Practice." Pp. 1-30 in Time, health and medicine, edited by R Frankenberg. London: Sage.

Garfinkel, H. 1967. Studies in Et hnomethodology. Englewood-Cliffs, N.Y.: Prentice-Hall.

Goody, Jack. 1977. The Domestication of the Savage Mind. New York: Cambridge University Press.

Hanseth, Ole, Eric Monteiro, and Morten Hatling. 1996. "Developing information infrastru cture: the tension between standardization and flexibility." Science, Technology and Human Values 21: 407-426.

Hacking, Ian. 1995. Rewriting the Soul: Multiple Personalities and the Sciences of Memory. Princeton: Princeton University Pres s.

Haraway, D. J. 1991. Simians, Cyborgs, and Women: The Reinvention of Nature. New York: Routledge.

Hirschauer, S. 1991. "The Manufacture of Bodies in Surgery." Social Studies of Science 21: 279-319.

Hirschauer, Stefan, and Annemar ie Mol. 1995. "Shifting Sexes, Moving Stories: Feminist/Constructivist Dialogues." Science, Technology and Human Values 20: 368-85.

Horobin, Gordon, and Jim McIntosh. 1983. "Time, Risk and Routine in Medical Practice." Sociology of Health and Illness 5: 312-331.

Huffman, Edna K. 1990. Medical Record Management. Berwyn: Physicians' Record Company.

Hughes, David. 1988. "When Nurse Knows Best: Some Aspects of Nurse/Doctor Interaction in a Casualty Department." Sociology of He alth and Illness 10: 1-22.

Hunter, K M. 1991. Doctor's stories. The narrative structure of medical knowledge. Princeton: Princeton University Press.

Hutchins, Ed. 1995. Cognition in the wild. Cambridge: MIT Press.

Kavanagh, Donn cha, and Luis Araujo. 1995. "Chronigami: Folding and Unfolding Time." Accounting, Management and Information Technologies 5: 103-21.

Khoshafian, S. Object-Oriented Databases, New York: John Wiley and Sons, 1993

Kritek, Phyllis B. "Conc eptual Considerations, Decision Criteria and Guidelines for the Nursing Minimum Data Set from a Practice Perspective" in Harriet H. Werley and Norma M. Lang (eds), Identification of the Nursing Minimum Data Set, New York: Springer, 1988

Kirk, St uart A. and Herb Kutchins 1992. The selling of DSM : the rhetoric of science in psychiatry New York : A. de Gruyter.

Latour, B. 1987. Science in action. Milton Keynes: Open University Press.

Latour, B. 1993. La Clef de Berlin et aut res leçons d'un amateur de science. Paris: La Découverte.

Latour, B. 1994. "On Technical Mediation: Philosophy, Sociology, Genealogy." Common Knowledge 3: 29-64.

Latour, B. 1996. Aramis - or the Love of Technology. Ca mbridge: Harvard University Press.

McCloskey, Joanne C. and Gloria M. Bulechek, eds. 1996. Iowa Intervention Project: Nursing Interventions Classification (NIC), second edition, St Louis, MO: Mosby.

Mol, A. 1993. "What is new? Doppler and its others. An empirical philosophy of investigations." In Medicine and change: studies of medical innovation, edited by I. Löwy, 107-125. Paris: Les Editions INSERM.

Mol, Annemarie. forthcoming. "Missing links, making links. The performance o f some atheroscleroses." in Differences in Medicine. Unraveling Practices, Techniques and Bodies, edited by Marc Berg and Annemarie Mol.

Orlikowski, W. J. 1992. "The duality of technology: rethinking the concept of technology in organizations."< i> Organization Science 3: 398-427.

Orlikowski, W. J. et al., ed. 1996. Information Technology and Changes in Organizational Work. London: Chapman & Hall.

Richman, J., and W. O. Goldthorp. 1977. "When was your last period? Temporal as pects of gynaecological diagnosis." In Health Care and Health Knowledge, edited by R. Dingwall et al., 160-84.

Roth, Julius A. 1963. Timetables: Structuring the Passage of Time in Hospital Treatment and Other Careers. Indianapolis: Bobbs -Merrill.

Schön, Donald. 1983. The reflective practitioner : how professionals think in action. New York: Basic Books.

Serres, Michel. 1987. Statues: le second livre des fondations. Paris : Editions F. Bourin.

Shortliffe, E H , and L E Perreault, eds. 1990. Medical informatics. Reading, MA: Addison-Wesley.

Smith, Dorothy. 1974. "The social construction of documentary reality." Sociological Inquiry 44: 257-68.

Smithson, S., R. Baskerville, and O. Ngwenyama. 1994. "Perspectives on Information Technology and New Emergent Forms of Organizations." In Transforming Organizations with Information Technology, edited by R. Baskerville et al., 3-15. Amsterdam: North-Holland.

Star, Susan Leigh. 1991. "The So ciology of the Invisible: The Primacy of Work in the Writings of Anselm Strauss." Pp. 265-283 in Social Organization and Social Process: Essays in Honor of Anselm Strauss, edited by David R. Maines. Hawthorne: Aldine de Gruyter.

Star, Susan Leig h, ed. 1995. Ecologies of Knowledge: Work and Politics in Science and Technology. New York: State University of New York Press.

Star, Susan Leigh, and Geoffrey C Bowker. 1994. "Of lungs and lungers: the classified story of tuberculosis." in A nnual Meeting of the Society for the Social Studies of Science. New Orleans, USA, October.

Star, S L, and J R Griesemer. 1989. "Institutional Ecology, `translations', and boundary objects: amateurs and professionals in Berkely's Museum of Vertebrat e Zoology, 1907-39." Social Studies of Science 19: 387-420.

Star, S. L., and K Ruhleder. 1986. "The Ecology of Infrastructures: Problems in the Implementation of Large-Scale Information Systems." Information Systems Research 7(1):111-134.

Stead, W. W. et al. 1992. "Database systems for computer-based patient records. Aspects of the computer-based patient record." In Aspects of the computer-based patient record, edited by M. J. Ball, and M. F. Collen, 83-98. New York: Springer.St evens, Rosemary. 1989. In sickness and in wealth: American hospitals in the twentieth century. New York: Basic Books.

Stevens, Rosemary. 1989. In sickness and in wealth: American hospitals in the twentieth century. New York: Basic Books.< p> Strauss, Anselm, Shizuko Fagerhaugh, Barbara Suczek, and Carolyn Wieder. 1985. Social Organization of Medical Work. Chicago: University of Chicago Press.

Suchman, Lucy. 1987. Plans and situated actions. The problem of human-machine communi cation. Cambridge: Cambridge University Press.

Ten Have, Paul. 1994. "Disposal negotiations in general practice consultations." in Negotiations in workplaces: discourses and interactional perspectives, edited by A. Firth. New York: Simon &am p; Schuster/Harvester Press.

Timmermans, Stefan, and Marc Berg. 1997. "Standardization in action: achieving universalism and localization in medical protocols." Social Studies of Science, in press.

Timmermans, Stefan, G C Bowker, and S L Star . forthcoming. "The architecture of difference: visibility, controllability, and comparability in building a Nursing Intervention Classification." in Differences in Medicine. Unraveling Practices, Techniques and Bodies, edited by Marc Berg and Anne marie Mol.

Waters, Kathleen A, and Gretchen Frederick Murphy. 1979. Medical Records in Health Information. Germantown, Maryland: Aspen Systems Corporation.

Weed, L L. 1971. Medical Records, Medical Education, and Patient Care. Chicago: Year Book Medical Publishers.

Zerubavel, E. 1979. Patterns of Time in Hospital Life. Chicago: University of Chicago Press.

Zuboff, S. 1988. In the age of the smart machine. The future of work and power. New York: Basic Books.

< /body>