reflexive standardization: side effects and complexity in standard making
TRANSCRIPT
1
1
REFLEXIVE STANDARDIZATION2
SIDE-EFFECTS AND COMPLEXITY IN STANDARD-MAKING13
4
Ole Hanseth5Department of Informatics,6University of Oslo, Norway7
(e-mail) [email protected]
Edoardo Jacucci10Department of Informatics,11University of Oslo, Norway12(e-mail) [email protected]
14Miria Grisot15
Department of Informatics,16University of Oslo, Norway17(e-mail) [email protected]
19Margunn Aanestad20
Department of Informatics,21University of Oslo, Norway22(e-mail) [email protected]
242526
Abstract27
In this paper, we address the general question proposed by the call for papers of this special 28
issue: “What historical or contingent events and factors influence the creation of ICT 29
standards, and in particular, their success or failure?” Based on a case study conducted over 30
1A previous version of this paper was presented at the MISQ Workshop on "Standard-Making: A Critical Research Frontier for Information Systems," Seattle 2003.
2
a period of three years in a Norwegian hospital on the standardization process of an 1
electronic patient record (EPR), the paper contributes to the current discussion on the 2
conceptualization of standard-making in the field of Information Systems. By drawing upon 3
the concepts of logics of ordering adopted from Actor-Network Theory and upon reflexivity 4
and unexpected side-effects adopted from reflexive modernization the paper makes three5
key contributions: 1) it demonstrates the socio-technical complexity of IS standards and 6
standardization efforts; 2) it reports of an empirical case study that shows how the7
complexity can generate reflexive processes that undermine standardization aims; and 3), it8
suggests a theoretical interpretation of standardization complexity by using of ideas from9
complexity theory and the theory of reflexive modernization. These research questions are10
addressed by offering an historical and contingent analysis of the complexity dynamics 11
emerging from the case.12
13
14
Keywords: standards, reflexive modernization, side-effects, socio-technical theory, 15
electronic patient records16
17
18
19
INTRODUCTION20
The research presented in this paper explores the borders and limitations of modern 21
standardization in the context of developing a pan Norwegian standard for electronic patient 22
3
record systems (EPR). An EPR is a computer-based information system for storing and 1
presenting patient clinical data in hospitals.2 We interpret the failed effort to replace the 2
fragmented (mostly paper-based) patient record and information system collage with one 3
integrated electronic record system as an inherent element and outcome of the complexity of 4
the standard setting. Moreover, we show that standardization processes can be reflexive-5
they result in outcomes which are antagonistic to the original aims. For example the 6
standard setting in EPR ended up producing a more fragmented record and IS portfolio.7
8
We will interpret this case and narrative contained therein by highlighting the associated9
complexity of the standardization effort. Our theoretical tool in this interpretive act is Actor-10
Network Theory (ANT) combined with the concept of reflexivity adopted from the theories of 11
risk and reflexive modernization (Giddens 1991; Beck et al. 1994; Beck 1986; Beck 1994;12
Beck, 1999; Beck et al. 2003). The mobilized concept of reflexivity reveals unexpected side-13
effects and how such side-effects can trigger new actions which will have their own side-14
effects, and so on. Initial actions with good intents may thus lead to self-destructive 15
processes whereby side-effects propagate ways that are “reflected” back on their origin and 16
cause the end result to be the opposite of what was initially intended.17
18
The paper makes three key contributions. Firstly, it demonstrates the socio-technical 19
complexity of IS standards and standardization efforts. Secondly, it provides an empirical 20
2 We have chosen to use the term “patient health record” or “medical record” used by core standardization organizations such as CEN TC251, HL7, or ASTM indicates a wider scope of the system that may go beyond single organizations.
4
case showing how this complexity may generate reflexive processes that undermine the 1
initial aims of standardization. Finally, the paper suggests a theoretical interpretation of this 2
phenomenon by means of complexity theory and the theory of reflexive modernization.3
4
The paper is structured as follows. In the first section we show how the development and 5
implementation of an EPR can be seen as a standardization process and how such a6
standard can be considered a complex system. In the second section we present our7
theoretical framework, conceptualizing standards as complex socio-technical systems. In the 8
third section our research design and methodology are outlined. In the fourth section our 9
case is presented, followed by analysis and discussion in the fifth section, including 10
implications and conclusions.11
12
EPR STANDARDIZATION AND ITS COMPLEXITIES13
EPR systems can be used by individual doctors, as a common system in a clinical 14
department, or as a shared and common system in an entire hospital or even among a set of15
interconnected hospitals. EPRs can be an off-the-shelf product, a proprietary system, or (as 16
in our case) a system co-developed between a group of hospitals and a vendor. An EPR 17
system is used to specify, routinize, and uniformize the type and format of clinical 18
information to be collected. Moreover, it is meant to support coordination and cooperation 19
between departments, professions and medical specialities, and hospitals. A hospital-wide 20
EPR could reduce redundancy and inconsistency of patient information, as the information 21
would be stored in one single location accessible from any place at any time. 22
Standardization activities have aimed at defining the appropriate design of the EPR as an 23
information system, e.g., with respect to fundamental architecture, access control and data 24
storage. In order to allow new users to get quickly into work or to avoid fumbling in 25
5
emergency situations, standardization of user interface and data presentation is also 1
advocated. However, an EPR can be also be conceptualized as a “package” of standards. It 2
builds on existing technical standards, e.g., with respect to operating systems, databases3
and network standards. It embeds clinical procedural and performance standards as well as 4
numerous classification schemes and terminologies (Timmermans and Berg 2003). These 5
standards go beyond the EPR system as such.6
7
We conceptualize the EPR standardization attempt as a process of alignment. Successful 8
standardization entails the achievement of stabilization and closure in the definition and 9
boundaries of the standard (Law & Bijker 1992; Bijker 1993; David and Greenstein1990). 10
This is not easily achieved, because of the socio-technical nature of standards, as well as 11
the number and variety of standards and their interrelations (Fomin et al. 2003; Bowker and 12
Star 1999; Brunsson and Jakobsson 2000). In our case a number of different actors were 13
involved, both within the individual hospitals and inside and outside the consortium or 14
national project.15
16
Delineating the intricacies involved in standard development processes, de Vries (2003) 17
defines standardization as:18
the activity of establishing and recording a limited set of solutions to actual or 19
potential matching problems directed at benefits for the party or parties involved 20
balancing their needs and intending and expecting that these solutions will be 21
repeatedly or continuously used during a certain period by a substantial number of 22
the parties for whom they are meant (p.155)23
24
6
Taking this definition as a reference, we submit that the observed phenomenon reported in 1
the case study can be seen as a process of standardization (ibid. p. 156) for the following 2
reasons: (1) the activity was aimed at finding a common set of information needs to be 3
addressed by balancing possible differences and views; (2) the clear aim was to create a 4
single solution (with the possibility of some degree of adaptation) to be shared by hospitals, 5
and, internally, by clinical departments; (3) the intended solution was aimed at addressing a 6
set of matching problems, that is, at harmonizing the problem of collecting, presenting, and 7
sharing clinical information between departments and possibly hospitals; (4) the 8
development was a clear long-term investment aiming at achieving a shared and long-lasting 9
electronic solution to recording, storeing, and sharing clinical information.10
11
We also propose that it is appropriate to conceptualize the EPR standard as a complex 12
system. We do so by means of Schneberger and McLean’s (2003) definition of complexity 13
as dependent on a system’s number of different types of components, number of types of 14
links and its speed of change. As noted above, a standardized EPR system includes or 15
builds on a large number of different standards which define very different types of objects. 16
These individual standards embed a large range of different work practices, and they are 17
developed by many different standardization “bodies” involving a large range of different 18
user groups and technical expertise. Further, the standards contain a huge number of 19
different types of links and relations between these elements, and, as will be demonstrated 20
later in this paper, many of the individual components and their relations are rapidly 21
changing. The EPR system standard is also in line with Paul Cillier’s (1998) more detailed 22
definition of a complex system as one made up of a large number of elements interacting in 23
a dynamic and non-linear fashion, forming loops and recurrent patterns which involve both 24
positive and negative feedback; it is open in the sense that it is difficult to define the borders 25
7
between it and other systems; it has “history”: its past is co-responsible for its present as 1
well as its future; and each element is ignorant of the system as a whole, responding only to 2
information available locally. This broad definition will underlie our conceptualization of the 3
systemic nature of the EPR throughout the paper.4
5
STANDARDS, SOCIO-TECHNICAL COMPLEXITY AND REFLEXIVITY6
The socio-technical complexity of IS standards has been studied from a variety of 7
perspectives. One strand has primarily addressed the role of network externalities – in 8
particular focusing on how these make standards increasingly difficult to change as their 9
installed base grows (Shapiro and Varian 1999). Another strand has focused on the 10
complexity - including local specificity and variety - of work practices and organizational 11
structures, and their embedding into standards (Bowker and Star 1999; Forster and King 12
1995; Hanseth and Monteiro 1997). A third strand has addressed the increased 13
heterogeneity of the actors involved in standardization on the one hand, and increased 14
speed of technical change on the other, and the challenges this leads to regarding the 15
organization of standards setting. These issues have caused a change from traditional 16
formal standardization bodies towards industry consortia as a preferred institutional 17
framework for standards setting (Vercoulen and Weberg 1998; Hawkins 1999; Shapiro et al. 18
2001).19
20
Overall, these studies have pointed at different aspects of complexity in relation to IS 21
standards and standardization dynamics. We suggest now to move one step further by 22
looking at the interdependencies and interactions between these forms of complexities 23
which, we argue, (sometimes) lead to reflexive, i.e. the combination of self-reinforcing and 24
self-destructive, processes. We will now present the theoretical framework we will draw 25
8
upon. The framework is primarily based on Actor-Network Theory (ANT). ANT has by and 1
large been developed and used to analyze the alignment of social networks or what we may 2
call the making of order in a complex world. This world has been seen as including humans 3
and non-humans, or technological and non-technological, elements. The kind of order-4
making that has been studied includes the development and acceptance of scientific 5
theories (Latour and Woolgar 1986), working technological solutions (Law 1987) and 6
organizational structures and strategies (Law 1994). Standardization is order-making par 7
excellence.8
9
Central concepts in early ANT research are closure (Law and Bijker 1992), stabilization (e.g.,10
Bijker 1993) and enrollment and alignment (e.g. Callon 1991). Specifically, closure indicates 11
a state where consensus emerges around a particular technology. Closure stabilizes the 12
technology by accumulating resistance against change. It is achieved through a negotiation 13
process and by enrolling actors/elements of various kinds into a network and translating (re-14
interpreting or changing) them so that the elements are aligned in a way that supports the 15
designers’ intentions. 16
17
The early ANT studies can be said to have focused on complexity in the sense that they 18
spelled out the rich and complex relations between the scientific and the technological on the 19
one hand, and the social on the other, related to the making of scientific theories and 20
technological solutions. ANT has been used in research on the negotiation of IS standards 21
and the embedding of their local context of development and use (Bowker and Star 1994, 22
1999; Timmermans and Berg 1997; Star and Ruhleder 1996; Hanseth and Monteiro 1997; 23
Fomin et al 2003).24
25
9
Since their emergence in the early 1980s, ANT and ANT research have evolved beyond1
their (so-called) “managerial” approach which focuses on how a single actor-network is 2
aligned by a dominating central actor (Law 2003b). Complexity has been addressed more 3
explicitly as the focus has turned to the dynamics unfolding when independent actors try to 4
align different but intersected actor-networks (Latour 1988; Star and Griesemer 1989; Law 5
and Mol 2002; Law 2003a; Law and Urry 2003). This has happened as has attention moved 6
towards more complex cases where order and universality can not be achieved in the 7
classical way.3 These cases are described as “worlds” which are too complex to be closed 8
and ordered according to one single mode or logic. There will only be partial orders which 9
are interacting in different ways, or interconnected and overlapping sub-worlds which are 10
ordered according to different logics.4 The interconnectedness of the sub-worlds means that 11
when one is trying to make order in one subworld by imposing a specific logic, the same 12
logic is making dis-order in another - an order also has its dis-order (Law 2003b; Berg and13
Timmermans 2000). Rather than alignment, stabilization and closure, the keywords are now 14
multiplicities, inconsistencies, ambivalence, ambiguities (Law 2003a, Law and Mol 2002). 15
Mastering this new world is not about achieving stabilization and closure, but rather about 16
more ad hoc practices - “ontological choreography” of an ontological patchwork (Cussins 17
1998). This approach has been applied to studies of cases such as train accidents (Law18
3 John Law and Annamarie Mol (2002, p. 1) define complexity as follows: “There is complexity if things relate but don’t add up, if events occur but not within the process of linear time, and if phenomena share a space but cannot be mapped in terms of a single set of three-dimensional coordinates.” This definition is very brief and rather abstract, but is in perfect harmony with Cillier’s definition presented above.
4 For more extensive demonstration of the logics of EPRs, see also (Gregory 2000, 2004).
10
2003a), a broad range of high-tech medical practices (Mol and Berg 1998), interdisciplinary 1
research [Star and Griesemer 1989]). This approach to complexity has also been applied to 2
analyzing the challenges, not to say impossibility, of achieving closure and stabilization in 3
relation to complex IS, and IS standards (Aanestad and Hanseth 2000).4
5
The evolution of ANT has brought it to closer affinity with the theory of reflexive 6
modernization (Beck et al 2003). The similarities between ANT and reflexive modernization 7
are observed by Bruno Latour when he states that a “perfect translation of ‘risk’ is the word 8
‘network’ in the ANT sense, referring to whatever deviates from the straight path of reason 9
and of control to trace a labyrinth, a maze of unexpected associations between 10
heterogeneous elements, each of which acts as a mediator and no longer as a mere 11
compliant intermediary” (Latour 2003, p. 36).5 From Beck’s perspective, the recent change of 12
focus within ANT signals a move from the first modernity to a second reflexive modernity,613
which is reflexive in the way that modern society itself is now modernized: the change is 14
happening not within social structures but to them. This leads to a “pluralization of 15
modernities” (Beck et al. 2003, p. 2): “ ‘meta-change’ of modern society results from a 16
critical mass of unintended side-effects … resulting from … market expansion, legal 17
5 The most substantial difference between the two is maybe the status they attribute to theories. Reflexive Modernization is presented as a theory in the classical sense which describes the world “as it is”, while ANT has adopted the ethnomethodological position and sees itself just as one “ethnotheory” having the same status as other such theories (Latour 2003).6 According to Beck two processes of modernization of our society can be distinguished: a first one called “first” or “simple” modernization, characterized by a stable system of coordinates as, for instance, the nation state, the gainful employment society - a concept of rationality that emphasizes instrumental control; a second one called “reflexive modernization,” characterized by a fundamental societal transformation within modernity which revolutionizes its very coordinates and calls into question its own basic premises (Beck 1999).
11
universalism and technical revolution“ (ibid. p. 2, emphasis in original)–what we normally 1
refer to as globalization. Beck defines “side-effect” more precisely as “effects that were 2
originally intended to be more narrow in their scope than they turned out to be” (ibid. p.2).3
4
The term “reflexive” connotes, in Bruno Latour’s interpretation, that “the unintended 5
consequences of actions reverberate throughout the whole of society in such a way that they 6
have become intractable” (Latour 2003, p. 36). Side-effects can thus be reflexive to the 7
extent they propagate through multiple, separate networks and finally become reflected -8
hence the term reflexive- back onto what initially triggered them. The end result can be one 9
that is the opposite of what was originally intended.7 In ANT terms, the propagation of side-10
effects results in the dis-ordering of networks created by an initial ordering action. 11
12
Standardization forms a key feature of modernization. Consequently, if the theories of Beck 13
and Giddens are valid, we should find reflexive processes unfolding in standardization. And 14
indeed, such examples are not hard to find – standardization is prone to escalated 15
processes of dis-ordering Hanseth et al. (2001), Ciborra et al. (2000), Ciborra and Osei-16
Joehene (2003), Rolland (2003). Hanseth and Braa (2001) in the context of corporate IT 17
standards, denote standardization vividly as “chasing the rainbow.”18
19
7 Reflexivity may be seen as a form of path-dependence. This concept has more recently emerged as influential within broader discussions of complexity theory. It has diffused from economics into other scientific fields – first historical sociology (Mahoney 2000), then sociology and social sciences more broadly (Urry 2003). Path-dependency in terms of self-reinforcing processes leading to lock-ins has been widely studied in relation to standards.
12
By the following case, we will narrate efforts to standardize an EPR system. This process1
qua standardization unfolded as a prototypical narrative of modernity, It can be interpreted2
as a control seeking process where the actors attempt to create a universal order by 3
enrolling heterogeneous elements, and thereafter translating and aligning them into one 4
closed and stabilized network. We will show how the orders that actors sought to create 5
were linked to multiple “worlds.” The ordering effects originating from each of these worlds 6
created disorders in others. The end effect was the undermining of the order-making and 7
increased disorder. This is what we will call “reflexive standardization”-that efforts and 8
actions taken toward standardization and stability lead to an opposite result. We standardize 9
in order to integrate, order and control a fragmented world, and to reduce its complexity – to 10
forge order out of chaos. Reflexive standardization, then, shows that when we try to achieve 11
order, and closeness, we get chaos, openness and instability.12
13
RESEARCH APPROACH AND SETTING14
The paper addresses the following research question suggested by the call for papers of the 15
special issue: what historical or contingent events and factors influence the creation of ICT 16
standards and in particular, their success and failure? By addressing this question, the 17
research focuses, in particular on three aspects, of standardization: 1) socio-technical 18
perspective on IS standards and standardization as outlined above, 2) identification and 19
analysis reflexive standardization processes, and, 3) interpretation of these processes with 20
the theory of reflexive modernization as discussed above.21
22
The socio-technical perspective provided us with conceptual lens that was valuable in 23
understanding and interpreting complexity associated with standardization. In particular, the 24
concepts of logics of ordering, orders, and dis-orders, provided a helpful sensitizing device to 25
13
interpret ordering processes of reflexive standardization. Theory of reflexive modernization, 1
in contrast has guided us discern of mechanisms of the standardization that create disorder: 2
how the uncertainties and difficulties with the standard implementation were not due to 3
external factors, but, in contrast, were internally and reflexively produced. By directing our 4
attention to the critical role of side-effects, and their production mechanisms we could thus 5
analyze how standardization situations were created that lapsed out of control. 6
7
We will examine these questions in the context of a case study that focused on the 8
development of a national standard EPR system and its initial implementation in the 9
Rikshospitalet in Oslo, Norway. This is a specialized University research hospital with 10
approximately 600 beds, 17 clinical departments, and approximately 3500 potential users of 11
clinical information systems. We studied both the intended and unintended consequences of 12
this standard implementation with a particular emphasis on the possible side-effects of on-13
going standardization actions.14
15
Research Design and Context16
The data for this paper were collected in the period 2001-2004. Other information is derived 17
from our previous collaborations between 1996-2001, where the EPR implementation was18
one topic for Master student projects and three master theses. Since 2001 the cooperation 19
has evolved into a structured research program with the purpose of studying the 20
implementation and use of clinical information systems in particular the EPR system. From 21
2001 up to the time of writing, the project has involved two professors, three post-doctoral 22
and doctoral researchers, and three Master students.23
24
14
The fieldwork has been structured as a longitudinal case study in order to follow the 1
implementation process in its various stages. To gain the valid knowledge for the case 2
analysis we grounded data analysis and collection in interpretive principles of case study 3
method (Klein and Myers, 1999; Walsham, 1993, 1995). In line with these principles the 4
focus of the research has evolved over time while the authors gathered and analyzed more 5
data which has influenced next round of data collections, and while the case progressed into 6
new stages. 7
8
As noted the case deals with developing an electronic patient ecord (EPR) that was 9
accompanied with the design and implementation of a specific product (called first DocuLive, 10
then IntEPR, then GlobEPR8) by Alpha Medical Solutions (the medical division of Alpha9), 11
as well as the adoption of this standard via the implementation at our study site. The EPR 12
was intended to be the electronic equivalent of the paper-based patient record. For decades, 13
EPR systems have been a major topic in the field of Medical Informatics internationally. 14
Their design, development, and implementation entail considerable complexity and 15
challenge (see e.g. Berg and Bowker 1997; Ellingsen 2002). In our study context, EPR 16
systems had for some years been widely used in general practitioners’ offices and in smaller 17
hospitals. Specialized and limited systems have also existed within single clinical 18
departments in larger hospitals. However, developing a hospital-wide centralized EPR 19
system, which involves standardizing the local systems and practices across clinical 20
8 The names of the product and the company have been disguised.
9 Alpha is a global multinational employing about 430,000 people generating revenue of over 75 billions Euros. Alpha is engaged in diverse industries such as healthcare, manufacturing, services, transportation and telecommunications.
15
departments, had proven to be a different task, not to mention efforts to develop generic 1
systems used in hospitals across multiple countries. 2
3
Research Methods and Data Analysis4
We gathered data from seven clinical departments (out of 17), and from the archive 5
department and the IT department by following theoretical replication. These clinical 6
departments were chosen using the implementation stage and department size as a 7
sampling criterion. Specifically, we sought to increase variance in these conditions and8
therefore gathered data from departments where the EPR system had been in use for a long 9
time, a short time, or where it was still under implementation. 10
11
We collected data with more than 35 formal interviews with 23 different employees of the 12
hospital including medical doctors, nurses, and secretaries in clinical departments, and 13
project leaders, heads of hospital units and senior managers in the IT department, including 14
the former CIO of the hospital. The interviews were semi-structured and lasted from one to 15
two hours each. Most interviews were audio recorded, while notes were taken as well. All 16
interviews were summarized and circulated within the research group, and key interviews 17
were fully transcribed. 18
19
Data were also collected in 18 in direct observations by participating in various discussions 20
and meetings, as well as from document analyses. The length of these observations varied 21
between two to eight hours. These observations covered, for example, tracing a patient 22
trajectory in or between clinical departments (and the production and use of information); 23
use of the paper-based and electronic patient record; observation of individual- and team-24
work in relation to information artifacts and IT support; observation of nursing activities and 25
16
use of information before and after implementation of the EPR (in the form of “shadowing”); 1
attendance at project meetings; attendance at EPR courses for user groups; and attendance 2
at preliminary meetings by the IT department and clinical departments before the actual 3
implementation.4
5
Regarding document analysis, we made use of primary and secondary sources. Primary 6
sources of documents included EPR project documents and other material produced mainly 7
by the managers of the IT department. We also analyzed policy documents and contracts 8
from the Norwegian consortium project. We had full access to the intranet of the IT 9
department which contained relevant documentation on the department’s budget and 10
strategic plans. The main secondary source of documental information comprised the 11
fieldwork reports and theses written by Master students during the period 1996-2001, and 12
reports, articles and theses written by members of the EPR research group since 2001. 13
Finally, relevant data were gathered during numerous meetings between members of the IT 14
department of Rikshospitalet and the research team. A weekly meeting of the EPR research 15
group was organized to discuss fieldwork, preliminary findings and further research 16
activities. The head of research of the hospital IT department joined these meetings at least 17
once a month, providing continuous updates on the ongoing activities in the project and 18
proposing new themes of research.19
20
Finally, we note limitations of our fieldwork. We recognize that the fieldwork could have been 21
extended to both the software company and to other hospitals implementing the same 22
system. We approached the software company but could not reach an agreement regarding 23
participation in the research. We intentionally decided to focus the fieldwork on the one 24
17
hospital discussed herein. Through our contacts to other researchers throughout Norway, we 1
were updated on the progress of concurrent implementations in the other hospitals involved. 2
3
DETAILED CASE DESCRIPTION 4
In Norway, work on the definition of a national standard for EPR’s started in the late 1980s. 5
A new organization called the Competence Center for IT in Health Care (with the Norwegian 6
acronym KITH) was established with standardization (definition as well as adoption) as its 7
main responsibility. It aimed at defining EDI (Electronic Data Interchange)-based8
communication standards for routine message exchange (e.g., lab orders and results, 9
prescriptions, admission, and discharge letters), closely linked with the work of CEN TC2511010
to which the EU commission had delegated the responsibility for development of European 11
IT standards for health care.12
13
In the early 1990s, two of the five Norwegian regional university hospitals and a small 14
Norwegian software company initiated a project aimed at developing an EPR system called 15
MEDINA. A project manager from KITH was hired. At this time KITH also started work on the 16
specifications of the Norwegian standard for EPR systems (KITH 2004), conforming to the 17
Norwegian standard for the paper record (Statens Helsetilsyn 1993) and the CEN EPR 18
standards as closely as possible. In 1996, the project enrolled the Rikshospitalet and the 19
other two regional university hospitals not already involved, whihc produced a consortium of 20
the five largest hospitals in Norway (see figure 1 to follow the timeline). This led KITH to see 21
10 CEN is the European branch of the International Standardization Organization (ISO).
18
this project as an important opportunity to develop a standardized Norwegian EPR system–1
not only a specification of some of its elements. To do so, they wanted to merge MEDINA 2
with another system DocuLive, which had been developed for about a decade, and had 3
been hosted by several software companies. Recently it had been acquired by Alpha-4
Norway. After the project organizations were merged, Alpha, as the largest and financially 5
strongest company, eventually bought MEDINA from the other vendors and took over the 6
entire product development project. 7
8
92 93 94 95 96 97 98 99 00 01 02 03 04
Alp
ha
Pro
du
cts
Pro
ject
sE
ven
ts
92 93 94 95 96 97 98 99 00 01 02 03 04
Pap
erR
eco
rd
Medina
Norwegian ConsortiumProject
Distributed
Centralized
CSAM (portal)
Scanning
Rik
sho
spit
alet
DocuLiveIntEPR
GlobEPR
Globalization process
No
rwe
gia
nG
uid
elin
es
fo
r ce
ntra
lized
pa
pe
rrec
ord
pu
blis
he
d
Alp
ha
ac
qu
ires
US
ba
se
dA
MS
No
rwe
gia
nH
ea
lthR
efo
rm
Cris
isa
t arc
hive
de
pa
rtme
nt
at R
H
Pla
nn
ed
de
livery
of
Do
cu
Live
5.0
(an
d e
nd
of
prjc
t)
Timeline
Scand. EU Global
No
r. Co
ns
. Prjc
ts
tarte
dw
ithA
lph
a a
nd
D
oc
uL
ive
910
Figure 1 The Timeline of products, projects, and events for our case11
12
19
In the new project, Alpha keept the DocuLive name and the deadline for the delivery of the 1
finalized system was set for the end of 1999.11 The DocuLive project started with the best 2
intentions of involving users, acknowledging current work practices, and favoring a bottom-3
up development strategy. Yet, as the number of involved users grew, large-scale 4
participatory development became unmanageable. After a few years, only a small number of 5
user-representatives from each hospital continued to actively participate in the development. 6
Moreover, the need to continuously find common agreements between the hospitals turned 7
the intended bottom-up approach into a top-down one.8
9
Overall, the strategy of the DocuLive project can be summarized as follows:10
The EPR should be developed to satisfy the needs of the five regional university 11
hospitals (with the implicit plan that with the successful completion of the project, the 12
EPR would also satisfy the needs of all other hospitals in Norway and accordingly 13
would be adopted by them). 14
The EPR should be “complete,” i.e., include all information about a patient.15
The EPR should be realized as one shared, integrated IS for all departments.16
17
This joint project between the five hospitals and Alpha was terminated early in 2004 without 18
the realization of the initial goal: the implementation of a complete EPR system. The version 19
of DocuLive currently in use has limited functionality in comparison to the project’s aims. 20
11 Ellingsen and Monteiro (2003a) have outlined the overall history of DocuLive from the early 1980s. They analyze other aspects of complexity in the Norwegian EPR project than those on which we focus in this article (Ellingsen and Monteiro 2003b).
20
Eventual further development of the system, at the time of writing, is regulated by separate1
contracts between the vendor and the individual hospitals. At the regional level, four of the 2
five regions in Norway (including the one which contains Rikshospitalet) have decided to 3
standardize on other EPR systems than DocuLive.4
5
The focus of our study is to analyze the role of one important factor behind the failure: 6
complexity. We have organized the empirical material into four stories. Each of them will be 7
on one or more of the “modes of ordering” exemplified by the strategy elements mentioned 8
above. The purpose of each is to illustrate how efforts aimed at making order interfered with 9
conflicting orders or order-making, ultimately producing more dis-order.1210
11
Alpha and the Stabilizing of the Scope of the Standard12
The first story focuses on the role of Alpha in relation to the shaping of the project trajectory. 13
Alpha is a large company and its international orientation challenged the stabilization of the 14
Norwegian standard and the creation of the EPR system. 15
16
When the project started in 1996, it was strongly expected to have the economic, political, 17
technical and medical capacity, both in terms of support and competence, to establish a 18
national standard EPR system in Norway. In the end, this did not happen because the 19
project evolved in unexpected directions. Shortly after the DocuLive project began, the IT 20
managers of Rikshospitalet became aware that Alpha UK was also engaged in EPR 21
12 The reader should refer to the timeline figure above to better follow the unfolding of the four stories.
21
development. Asking Alpha Norway for more clarification, they found out that within Alpha, 1
several EPR development projects co-existed: at least five EPR projects were underway in 2
Sweden, UK, Germany, India, and Norway respectively. The IT department at Rikshospitalet 3
realized that the Norwegian project was not at the top of Alpha’s priorities since Norway 4
represented the smallest market. Within Alpha, the DocuLive project ran the risk of being 5
overrun by other internal projects for more profitable markets. As a consequence, the project 6
consortium, together with Alpha Norway, decided to make the first move to internationalize 7
the project, first to a Scandinavian level, and later to a European one. 8
9
A senior IT manager commented:10
Alpha decided and the hospital agreed to internationalize the product. At that time 11
there were different competing systems within the Alpha company. We saw potential 12
danger to our system and our development and requirements. We supported Alpha13
in bringing this up on the corporate level and getting DocuLive and the Norwegian 14
product to become a main product for Alpha internationally. Because that would 15
secure further development on our system. It was a strategic decision.16
17
The strategy of the consortium was to push the project to a larger dimension in order to 18
secure its continuity. On the other hand, this decision weakened the hospital consortium’s19
position with respect to Alpha, since now it was not the only client with system requirements. 20
Requirements from all other EPR projects in Alpha had to be merged and a new architecture 21
had to be designed. Furthermore, since the original deadline for the final delivery (1999) was 22
approaching, the project consortium agreed with Alpha to extend the time frame to include 23
the development of the new internationalized EPR solution (called IntEPR).24
25
22
At the time the IntEPR project started in 1999 (see timeline in Figure 1), Alpha decided to 1
acquire AMS (American Medical Systems13), a large US software development company. As 2
a consequence, the scope, resources, and balance of the Alpha medical division changed: 3
the division’s headquarters was moved from Europe to the US, and the project’s scope 4
became global. In this scenario, the project consortium supported Alpha to internationalize 5
IntEPR. However, as the project became global, the IntEPR architecture was dropped in 6
favor of a new system called GlobEPR. The basic requirements previously defined for the 7
Norwegian customers of DocuLive, were partly supported by the new architecture. 8
9
From this story we can see the meeting of two different “worlds,” each with their own mode 10
of ordering: the one of the Norwegian project for the Norwegian hospitals, and the one of 11
Alpha and its international scope. To Alpha, achieving economies of scale by targeting12
international markets is a key concern. In addition, the medical division within Alpha is large, 13
with a traditional base within medical imaging technologies. As the imaging instruments have 14
become digital, supplementary software systems have been built. As the EPR development 15
activities were increasing within Alpha, it became more and more important to align and 16
integrate the EPR strategy and product(s) with other Alpha products and strategies.17
18
Within this world, Norway becomes marginal, as the appetite for larger markets escalates in 19
a self-feeding process. From the Norwegian point of view, the original interest in creating a 20
Norwegian standard had to be reinterpreted in a Scandinavian, then European, and finally 21
13 The name of the company has been disguised.
23
global context. A side-effect of the expansion of ambitions and scope was increased 1
complexity: the larger the market Alpha was aiming at, the more diverse the user 2
requirements, and accordingly, the more complex the system had to be in order to satisfy 3
them. This implied that the development costs were growing, which again implied that a 4
larger market was required to make the whole project profitable.5
6
The Complete EPR7
In our second and third stories we look more closely into the implementation process inside 8
the hospital. The efforts aimed at replacing the fragmented paper-based record with a 9
complete and smoothly integrated electronic one turned out to be more challenging than 10
foreseen. In the end, the volume of paper records increased (second story) and the patient 11
record became more fragmented (third story). This in turn increased the overall complexity 12
and consequently slowed down the standardization and implementation processes.13
14
Before 1995, the main problem at Rikshospitalet (as well as most other hospitals) was the 15
fragmentation of the medical record system: each department had its own record archive. If 16
a patient was admitted to several departments, several records would be created, each one 17
containing its specific information. In addition, various smaller local information systems, 18
partially overlapping with the paper records, were in use. In this picture, a long time might be 19
needed to retrieve critical information on patients. This could lead to situations where critical 20
decisions were made without possessing vital information contained in all the relevant 21
medical records.22
23
In 1996, the same year as the DocuLive project started, Rikshospitalet standardized and 24
centralized its paper-based patient records according to the principle: one patient, one 25
24
record. The new centralized paper standard followed the recently published Norwegian 1
guidelines for paper-based patient records. This centralization process was not without 2
problems. In particular, a major complaint was familiar: the long time needed to retrieve a 3
patient record from the central archive. Doctors also complained that, due to the 4
centralization, the merged patient records had become less easy to browse quickly, as a lot 5
of the information was not relevant to their specific interests. In this situation it was widely 6
assumed among doctors, as well as IT people at the hospital, that a standardized, complete7
EPR system would make information instantly available anywhere at anytime, as well as 8
avoid duplication of information and inconsistency of data. 9
Basically, the aim of the DocuLive project was to replicate and replace the recently 10
standardized and centralized paper-based patient record. 11
12
However, at the time of writing (autumn 2004), a full transition from the paper to the 13
electronic record has not yet been accomplished. The DocuLive system mainly contains 14
textual information, while much other information is still on paper forms (lab results, radiology 15
reports, images and other printouts from various equipment). 16
17
A manager from Rikshospitalet’s IT department helped to quantify the situation: “Currently 18
DocuLive covers about 30-40% of information contained in an average paper-based record. 19
Basically most of the information is text.”20
21
From an October 2002 internal report, we found an even more pessimistic estimate of 22
DocuLive’s coverage:23
[It] covers 18 of a total of 66 document groups defined in the Norwegian standard for 24
the paper patient record. In terms of volume – with a high degree of uncertainty – that 25
25
accounts on average for about 10% of the total volume of a paper-based record.1
(Translated from Norwegian by the authors)2
3
Although the implementation of the EPR aimed at reducing paper and eventually replacing 4
the paper system, the paper-based record still remained an important tool. Paradoxically, the 5
production of paper documents increased markedly after the implementation of DocuLive. 6
First, new laws on medical documentation required detailed records from professional 7
groups not previously obliged to maintain a record, such as nurses, physiotherapists and 8
social workers. Second, for legal reasons the hospital kept the paper version of the record 9
updated. Thus, each time a clinical note was written in the EPR, a paper copy was also 10
printed and added to the paper record. Printout efficiency was not a design principle for the 11
current EPR, causing non-adjustable print layouts that could result in two printed pages for 12
one electronic page form. Third, multiple printouts of preliminary documents (e.g. lab test 13
results) were often stored in addition to final versions. The result was that the volume of 14
paper documents increased. This growth created a crisis at the paper record archive 15
department. The hospital had moved into new facilities designed with a reduced space for 16
the archive as it was supposed to handle electronic records only. In 2003 the archive was full 17
and more than 300 shelf meters of records were lying on the floors. This situation also 18
affected the time needed to find records, and often requests failed to be satisfied.19
20
From an internal report:21
In .. 2002 a daily average of 790 requests for paper records were received. … About 22
half of the requests did not turn out as an actual delivery. There are several reasons 23
for this. The most common are that the record has already been delivered in another 24
department or has already been collected; that it is not possible to locate the record25
26
(due to wrong archiving); or that the archive never had the record for that patient 1
(usually because it is a new patient). (Translated from Norwegian by the authors)2
3
To alleviate this situation, a scanning project was started in 2003, with the aim of reducing 4
the amount of paper documents sent to the archive. However, even after the documents 5
were scanned they had to be kept. One reason was that DocuLive’s storage solution was not 6
yet accredited; for that one had to wait for release and implementation of a new version of 7
the software. Another reason was that the existing communication and coordination of work 8
practices were based on a flow of paper documents, and DocuLive did not (yet) contain 9
functionality that would allow the paper to be removed from the daily work practices. The 10
result is that the benefits from the scanning activities were slow to be realized.11
12
This story may be seen as a confrontation between what we might call “the order of 13
computers” and “the order of paper.” Computers, we can argue, are best exploited if they are 14
allowed to work the way that fits them best: where all information is stored in a shared, 15
consistent and non-redundant database. However, the paper record is ordered according to 16
different principles in order to also be an efficient tool for local work practices, and the 17
assumption that all patient related information could be ordered according to the “computer 18
order” has not yet been proven. At best, the transition period from paper-based to digital 19
information will be long. During this period the electronic and the paper-based record have to 20
coexist and “cooperate.”21
22
One Record (per Patient) - One (Integrated Information) System23
The third story focuses on the relation between the new EPR system and the other clinical24
information systems in the hospital. When the implementation of DocuLive started, a few 25
27
local systems containing clinical patient information already existed. Those systems were 1
often overlapping with DocuLive’s (planned) functionality. The original plan as revealed by 2
project documentation was to replace these with DocuLive so as to have the EPR as one 3
integrated information system:4
5
[DocuLive should] create a common platform for a multitude of customized EPR 6
modules,7
[be] powerful enough to support all health-related information and legal aspects8
[be] general enough to serve as a basis for a wide variety of hospital information 9
systems …10
(Technical Overview Document, 1998, translated by the authors)11
12
In this story, again the project’s ambitious plan ended up producing the opposite outcome, 13
which contributed to intensifying the degree of fragmentation of the medical record. The 14
main ordering principles for achieving the integrated record were that there should be one 15
record for each patient, containing all patient related information, that this record should be 16
shared among all units within the hospital, and that all patient records should be maintained 17
by and stored in one single integrated information system. In order to achieve this, DocuLive 18
was planned to be integrated with a few other systems: the central Patient Administrative 19
System (PAS), and certain “information supply” systems, notably laboratory systems that 20
store and deliver laboratory test results, and image archives for radiological images. The 21
idea of entirely substituting the other local EPR-like systems was slowly abandoned. Firstly, 22
to include the functions of all these systems into the EPR would have made its development 23
unmanageable. Secondly, users generally perceived local systems to better support their 24
28
work routines and refused to give them up. For instance a doctor in pediatric cardiology 1
stated referring to a local system:2
3
If you have congenital heart defects it is very likely that you have also other 4
congenital defects. So it is a very complex logistics of patients. These two reasons, 5
the very detailed diagnostics, and the need of keeping track of the patient are the 6
bases for the design of our system.7
8
Rather than replacement, various solutions for technical integration of DocuLive with some 9
of the local systems were considered. These intentions were realized only for a few systems, 10
leading to a situation where users had to either perform double entries or cut and paste 11
information between the systems. Simultaneously the number of specialized information 12
systems were growing, based on well justified needs of the different medical specialties and 13
departments. For example, the in-vitro fertilization clinic needed a system that allowed them14
to consider a couple as a unit, as well as allow tracking of information from both semen and 15
egg quality tests through all procedures involved, up to the birth of the child. The intensive 16
care unit acquired a system that allowed them to harvest digital data from a vast array of 17
medical equipment and thus eliminate the specialized paper forms previously used to 18
document events and actions. Moreover, new digital instruments in use in many different 19
departments include software components with medical record functionality.20
21
Hence, the envisioned role of DocuLive has changed from being the only system to being 22
one among a large (and increasing) number of systems (see figure 2: from “Original Vision” 23
to “Later Vision”). As the problems and the challenges with the original integration strategy 24
emerged, the popularity of the Internet and its technology triggered the IT department at the 25
29
hospital to start thinking about other potential strategies. It started tinkering with portal 1
technology, and this led to the idea of an integrated EPR system achieved by means of a 2
more loosely coupled infrastructure where the many clinical and laboratory systems (and 3
DocuLive itself) were brought together under the common umbrella of a portal (see figure 2: 4
“Current Vision”). The portal was part of a larger change in strategy which went under the 5
acronym CSAM: Clinical Systems All Merged. Thus, while visualization and access to the 6
systems were integrated, the systems themselves did not need to be integrated with each 7
other.8
9
Original vision Later vision Current vision
PAS
LabSystem
LabSystem
...
LocalEPR
LocalSystem
LocalEPR
LocalSystem
DocuLive ”Umbrella” PAS
LabSystem
LabSystem
...
LocalEPR
LocalEPR
LocalSystem
LocalSystem
DocuLive ”Umbrella”
All systems integrated within DocuLive Some Systems integrated(loosely or tightly)
PAS
LabSystem
LabSystem
...
LocalEPR
LocalSystem
LocalEPR
New Portal ”Umbrella”
DocuLive
Variable levels of integrationunder the New Portal
10
Figure 2 Three stages of the envisioned role of DocuLive11
This story shows, again, how the world of DocuLive and its order was confronted with other 12
worlds with different orders and ongoing ordering processes. Different medical specialties 13
30
focus on different types of information (this being even truer in the case of specialized or 1
tertiary hospitals like the one in this case). The ordering principle of “one patient – one 2
record” is non-problematic for many, but may interfere with and create dis-orders for others. 3
As a result, the attempt to achieve a tightly coupled integration of systems (in the view of the 4
logic of ordering of DocuLive) clashes against different logics which reflect the actual 5
complexity and diversity of the work practices to be standardized. From this clash–this 6
interference of orders–comes the generation of a new logic, implicit in the portal strategy 7
CSAM.8
9
To a certain degree, the novel portal strategy appears promising. It is a less strict and 10
accordingly a more flexible way of standardizing and ordering. It seems more likely that this 11
IS strategy can deliver a “complete” system for accessing information. To implement such 12
strategy is far from trivial or without risk. It entails further development work, as adapters13
need to be developed between the portal software and the different applications. Moreover, 14
the laws and regulations concerning documentation of patient information are clearly based 15
on the envisioned all-encompassing EPR. One complete patient record is recognized to be 16
the legal document, while the idea of keeping information in different sources as the CSAM17
strategy proposes is legally problematic: not all of the underlying systems are designed with 18
adequate security of patient data in mind; therefore they do not conform to the standards of 19
the privacy laws, neither when it comes to access control solutions nor long-term storage of 20
confidential data.21
22
The Role of the Regional University Hospitals Revisited23
The fourth story describes how a health sector reform in Norway interfered with the ongoing 24
EPR standardization process. When the DocuLive project started, new procedures and 25
31
technologies were usually developed or first adopted by the five university hospitals, and 1
subsequently by the other hospitals. The standardization of the EPR was expected to follow 2
this pattern, but a major reform in the health sector, initiated in 2001, affected the 3
standardization process significantly.4
5
Before the reform, the hospitals in Norway were owned by the country’s 19 counties. There 6
was a widely held view that the health system was too fragmented, did not encourage 7
smooth collaboration and did not maintain a rational division of labor. The reform implied that 8
the government was taking over the ownership of all hospitals by means of five regional 9
health enterprises, which again owned the individual hospitals. The reform significantly 10
altered the dynamics in the sector. As a health enterprise, each region had to define cost-11
efficient and effective strategies to realize benefits of scale. A key concern was then to 12
standardize on IT, and specifically to select one standard EPR system among the three 13
systems in use. Inevitably, this created competition among the hospitals for each to have its 14
own EPR system prevail over the others.15
16
At this point the DocuLive system progressed slowly due to continuously emerging new17
elements and demands originating from Alpha’s activities and shifting strategies. In this 18
situation, the IT department at Rikshospitalet attempted to market DocuLive as the standard 19
system to be adopted by other hospitals, even though its development was far from 20
complete. Moreover, Rikshospitalet sought to become the reference center for delineating 21
and implementing regional IT strategies. However, this strategy of promoting DocuLive was 22
soon changed, as the IT department at Rikshospitalet acknowledged its rather weak position 23
in the regional “battle of systems” (Hughes 1983). Accordingly, they made a strategic move, 24
promoting the portal concept rather than DocuLive. This turned out to be a more flexible and 25
32
robust strategy in order to enroll the other hospitals in the region into a collaborative rather 1
than competitive standardization effort. The strategic move to promote the portal strategy 2
also implied that DocuLive was “buried” as “the standard” EPR (although not as product).3
4
ANALYSIS AND DISCUSSION5
In this section we discuss the complexities that this case exhibits, and identify three research 6
contributions. First, we show the socio-technical complexity of IS standardization by 7
highlighting how the standardization was shaped by the different orders, multiple actors at 8
play and the interferences between orders. which created dis-orders. Second, by analyzing9
how dis-orders may undermine the creation of a possible stable ordering logic or solution 10
and of a closure of the standard, we show how this complexity generates reflexive 11
mechanisms. We call this phenomenon reflexive standardization. Third, we discuss how the 12
reflexive mechanism is an instance of the complexity inherent in IS standardization, and we 13
contrast our findings with traditional approaches to IS development and standardization.14
15
Interference and Propagation of Side-effects16
The four stories presented above provide an account of the multiplicity of perspectives, 17
intentions, constraints, challenges, and agendas at work in the socio-technical network of the 18
standardization process. As Law points out, orders are not “simply told, performed and 19
embodied in agents, but rather they speak through, act and recursively organize the full 20
range of social materials” (Law 1994, p. 109). This perspective helps us to go beyond the 21
intentionality of single actions and to see how ordering logics are embedded in 22
heterogeneous social networks. Thus, their character is contingent and, in part, a matter to 23
be determined empirically (Law 1994). 24
25
33
Different orders interact with and re-organize one another: they may create dis-orders, or 1
reinforce existing orders. While it is true that standardization processes may eventually 2
stabilize- we have a large number of standards embedded in our practices- this case 3
suggests that, under certain circumstances, interferences between orders will reflexively 4
produce additional interferences with greater complexity that will ultimately destabilize the 5
initial order. 6
7
In the first story we see how the two main actors, Alpha and the Norwegian EPR project, 8
mutually and iteratively redefined their aims, strategy, and design of the standard as the 9
project gradually escalated to a global level. The overall result of this dance of orders and 10
redefinition of interests was that the Norwegian standardization project, as initially 11
conceived, did not succeed. The second story highlights conflicts between (the order of) the 12
electronic patient record and (the order of) the paper-based record. As a result, the strategy 13
for creating an integrated record created, as an unintended consequence, a more 14
fragmented one. The third story illustrates how the ordering principle of “one record (for each 15
patient) – one integrated Information System” created dis-orders in terms of making it more 16
difficult for workers to have easy access to the specific information they needed. The final 17
story illustrates how a new order enforced by the government interfered with the ordering 18
principles of the project. 19
20
The failure of DocuLive, at least as a standardization story, can be seen as a failure in 21
attempting to control complexity. Arguably, the main mistake was to follow a “traditional” 22
standardization approach–typical for (first) modernity that is, overemphasizing criteria of 23
universality, uniformity, and centralization of control to achieve alignment, stabilization and 24
closure. In line with our theoretical framework, our case data suggest that the complexity 25
34
defines standardization as the emergence of multiplicities, inconsistencies, ambivalence, 1
and ambiguities (Law 2003b, Law and Mol 2002). Ironically, what happened became the 2
opposite of the initial aims. When actors tried to stabilize the standard by enrolling more 3
actors, this made it less stable. Attempts to improve fragmented records by means of one4
integrated EPR made the records more fragmented. The complexity of DocuLive turned out 5
to be one where the ordering efforts created dis-orders. The side-effects triggered new ones, 6
which again were reflected back on the origin–the standardization process turned out to be 7
reflexive and self-destructive. The dynamics of reflexive processes at work are summarized 8
in figure 3.9
More varied and complex work practices to be supported
More complex and fragmented IS portfolio constituting the EPR
More complex EPR product
More fragmented medical record
Larger market to get economies of scale
10
Figure 3 The reflexive standardization process11
The concept of reflexivity offers thus an interpretation of the dynamics of the case. The 12
theory of “high” modernity helps to observe how the logics of the “first” industrial modernity 13
find their limits (Beck et al. 1994). The intensified interconnectedness of social practices with 14
technical artifacts on the one hand, and the need to align geographically dispersed actors on 15
the other hand, effectively undermines the reductionist approach to control complexity. The 16
35
weakness of such approach become visible when the control itself reflexively reproduced the1
complexity- thus creating the immanent paradox of modernity Reflexive standardization 2
seeks to highlight the need to develop alternative standardization approaches that better 3
overcome the paradoxes to deal with complexity as pointed out below.4
5
On the Validity of the Case Analysis6
We have so far attributed the failure of the standardization effort to the inherent complexity of7
the standardization process and the ways in which how such complexity was addressed. But 8
this may not be the only possible explanation. It may be argued that the effort failed because 9
of poor project management, insufficient user participation, incomplete requirements 10
specifications, bad decisions, historical circumstances like stupid people, following wrong 11
standardization approach, and so on. In our view, a standardization effort like the one 12
described here almost never fails for one reason only. We do believe that the EPR project 13
management followed strategies and made decisions that were well in line with “best 14
practices” in software engineering, based on traditional standardization models. Without 15
doubt, questionable decisions were made. Yet, we think that improved user participation or 16
requirements specifications would not have saved the effort. In the similar vein sophisticated 17
risk management methods would not have been of much help, rather they, most likely, would18
have added to the overall complexity and triggered new reflexive dynamics. In fact, when 19
dealing with complex phenomena we will always be confronted with unpredictable events 20
and problems (Perrow 1999). We think that a different approach to complexity could have 21
improved the chances of a successful standardization effort. Still, the people involved are not 22
to blame. Rather, we submit, the problem is the poor understanding of complexity and 23
standardization within the software engineering, information systems, and standardization24
fields.25
36
1
With regard to the external validity of our case, we can ask: is our case representative of a 2
new class of standardization problems? We believe so, especially in the health care domain 3
where plans for developing electronic health records grow continuously bigger and more 4
ambitious. For example, in his 2004 State of the Union address, George W. Bush Jr. 5
envisioned “an EHR for all Americans within the next decade” (The White House, 2004). In 6
addition, to building a national health information infrastructure, establishing data 7
interoperability and comparability for patient safety data is seen as crucial. This is expected8
to be facilitated through adopting standards that allow medical information to be stored and 9
shared electronically while assuring privacy and security. Similarly, the British “Connecting 10
for Health” initiative proposes to establish the NHS (National Health Service) Care Record 11
Service. For each individual patient the Patient Clinical Record will be used to deliver direct 12
patient care, and in addition a centralized database (“The Spine”) will contain a National 13
Summary Record in order to support urgent and emergency care (NHS, 2005). For both 14
initiatives, huge challenges can be recognized and critical voices emanating from our 15
analysis can predict significant obstacles. However, from official documents and 16
presentations of these projects, the general perception of standards is the value of increased 17
control: developing and adopting standards is definitely seen as part of the solution, rather 18
than part of the problem.19
20
We do not claim that standardization of medical information systems is impossible or 21
undesirable. Because our case study site is a specialized hospital, it represents a 22
paradigmatic example of the socio-technical complexity arising from the close intertwining of 23
technical standards with local and highly professional work practices (in terms of 24
professional disciplines and geography). To some extent our case represents a general25
37
class of problems associated with the interactions between the complexity of information 1
infrastructures, information processing, and local work practices. Thus, the common shared2
complexity in these classes is the immense heterogeneity and multiplicity of actors involved3
and the need to coordinate and standardize their behaviors. Not surprisingly, challenges 4
associated with complexity similar to those described here, have been recently identified in 5
multiple areas including Enterprise Resource Planning (ERP) systems, corporate IT 6
infrastructures in the oil and chemical industry (Hanseth et al. 2001; Ciborra et al. 2000), the 7
financial sector (Ciborra and Osei-Joehene 2003), ship classification (Rolland 2003), and e-8
government (Ciborra 2003). All these are responses to a “quest for integration” (Dechow 9
and Jan Mouritsen, in press) and will unavoidably lead to increases in IS complexity. 10
Accordingly, we believe that the reflexive processes as illustrated here will grow in number 11
and importance.12
13
Implications for Research and Practice14
What are the implications of our findings for practice? Significant research efforts into the 15
complexity of IS standards and their dynamics are called for before detailed concrete advice 16
can be given. We agree with the authors of the British Computer Society 2004 (BCS 2004) 17
report that complexity is the single most important issue for software engineering and 18
information systems design. Research on the “duality of standards” regarding complexity, 19
i.e., how standards may reduce as well as increase complexity, forms an important part of 20
this. We need also further research on specific reflexive effects of complexity in different 21
situations. Here, we will offer only tentative answers to this question. In doing so, after22
Perrow (1999) we assume that identified problems are inherent to complex systems and 23
better structured methodologies or management tools for control will not solve them. To 24
address such problems, we have to seek to avoid creating complex systems. 25
38
1
Avoiding the kind of socio-technical complexity we have pinpointed may be possible by, first 2
resisting the temptation to gain perfect order. The case study of the development of an EPR 3
standard illustrates that traditional engineering approaches are risk prone in complex areas 4
like designing technologies that span across multiple work practices. These approaches tend 5
to overestimate the universality of work practices, thus seeking order by simplification and 6
abstraction and thus laying strong emphasis on design criteria such as consistency, 7
completeness, and non-redundancy. These are all sound engineering principles and central 8
to modernity. These criteria have been explicitly emphasized within the development of IT 9
standards for health care (De Moor et al. 1993, McDonalds 1993),. They all work well if we 10
can start out by delimiting a part of the world that can be treated as closed and in isolation.11
But they can become a risk, if and when such assumptions do not hold as identified in12
failures of IS-related standards in various areas (Graham et al.1995; Graham et al. 1996; 13
Hanseth et al. 1996; Hanseth and Monteiro 1997; Hanseth and Braa 2001). In the case of IS 14
standards-making, the more the object of standardization is close to local work practices, 15
and the more knowledge-intensive the work practice is (e.g. a specialized hospital), the less16
likely the traditional approach is to succeed, possibly generating a reflexive self-destructive 17
process. AS Law (2003a, p. 14) notes: “[T]he search for system perfection is not only 18
impossible but, more strongly, it may be self defeating” (Law 2003a, p. 14). We need to 19
accept in our standard making that our complex worlds are populated with a multiplicity of 20
orders that are inconsistent. We need to be able to live with such multiplicities and 21
inconsistencies–we need to master the trade of what Charis Cussins (1998) calls 22
“ontological choreography.”23
24
39
Another key element of a strategy accepting a multiplicity of orders is to identify sub-worlds 1
which can be properly ordered and which are interfering with each other as little as possible, 2
i.e., making sub-worlds that are loosely coupled (Perrow 1999). Maintaining loose coupling 3
between the social and the technical is perhaps the most important strategic element. And 4
what should be avoided is the embedding of specific working practices into the standards. 5
To do so, one needs to be well aware of the local specificity of work practices, and the fact 6
that the practices are embedded into the technology. 7
8
Another way to reduce socio-technical complexity is to reduce the organizational complexity9
that results from the large number and heterogeneity of actors and their interdependencies. 10
This is caused by the reach and range (Keen 1991) of the standard, which we see as a key 11
source of its technical complexity. One strategy to reduce the scope is to split the standard12
into separate and independent packages, each of which has a more restricted reach and 13
range. Rather than one universal standard in which the elements are tightly coupled, we 14
should aim at a multiplicity of simpler standards that are loosely coupled. Such loose 15
coupling between standards and infrastructures based on them can be achieved by means 16
of gateways (Hanseth 2001). In the domain of EPR systems this means that one should 17
develop separate systems for different countries, and that these systems could be further 18
split up into individual systems for specific medical specialties or hospital departments or 19
functions. Different medical record systems in a hospital can then be integrated either by 20
sorts of gateways enabling the exchange of shared data between them, or by means of a 21
portal which provides a shared interface towards all of them.22
23
Complexity is a vague term. It is hard to measure. We have learned in both the natural and 24
social sciences that everything is indefinitely complex when we look at it carefully – if we 25
40
“open the black box” (Latour 1988). Black-boxing is a strategy for reducing complexity which 1
is closely related to modularization. It is indeed a potentially powerful strategy, but it works 2
well only under conditions of stability. A simple description of a complex system (in terms of 3
an interface or an abstract specification) may be sufficient for a specific purpose. However, if 4
the complex system is changing or the needs change, the simple description may not be 5
appropriate any more. From this we can derive another strategy to reduce complexity: look 6
carefully for elements in the world -medical practices, instruments, ICT solutions- that will not 7
change. These are the elements that may be black-boxed, ordered and turned into 8
standards.9
10
The points made above can be illustrated by contrasting DocuLive on the one hand and the 11
portal on the other. The portal strategy is certainly very promising in terms of its robustness 12
and flexibility. Its guiding principle is loose coupling between the different parts of the EPR 13
system in contrast to the tight coupling strategy of DocuLive where all information was stored 14
in one shared data base and all functions integrated into one single software system. The 15
portal strategy potentially makes it easy to include any kind of information and illustrations 16
produced by new IS into the electronic record. For the time being, this strategy also 17
embodies its risks. We do not know much about the long-term requirements of such portal 18
and how to structure and design it. Such a portal can also be complex as more applications 19
are integrated and its sophistication grows. It may become so complex that reflexive 20
processes are triggered.21
22
CONCLUSION23
In this article, we have argued that (1) socio-technical complexity is a major issue in 24
information system standardization. (2) Through a case study of EPR standardization critical25
41
dynamics related to this complexity relate to reflexivity. (3) Reflexivity (Beck 1994) explains 1
how under certain circumstances, efforts aiming at reducing complexity through 2
standardization may generate the opposite outcome. The circumstances described by the 3
case represent a paradigmatic example of increased intertwining of technical standards with 4
local, heterogeneous and dispersed work practices. Through our analysis we argue that 5
traditional standardization approaches can not deal with such complexity appropriately. Not 6
only will such approaches fail to not deliver the intended outcome- order-; they can also lead 7
to the opposite effects of greater dis-order, and instability. The need of future research on IS 8
standardization, is therefore critical in approaches that help mitigate the increasing 9
complexity IS standardization.10
11
Acknowledgements12
We would like to thank the IT department of Rikshospitalet, Oslo, Norway. We owe special 13thanks particularly to Ivar Berge and Arve Kaaresen for setting up the research project and 14for the long-lasting research relationship. We thank Eric Monteiro, Marc Berg, Geoff 15Walsham, Judith Gregory, Sundeep Sahay, Ola Henfridsson and the participants in the 16Workshop on Standard-Making organized by MISQ in Seattle for providing useful comments 17on earlier versions. Special thanks go to two anonymous reviewers, the editorial board, and 18the senior guest editors of the special issue for their insightful comments and 19encouragement.20
21
.22
23
REFERENCES24
Aanestad, M., and Hanseth, O. “Implementing Open Network Technologies in Complex 25
Work Practices: A Case from Telemedicine,” in Organizational and Social 26
Perspectives on Information Technology, Kluwer Academic Publishers, Boston, MA, 27
2000, pp. 355-369.28
BCS. The Challenges of Complex IT Projects. British Computer Society and Royal Academy 29
42
on Engineering. Report. http://www.bcs.org/NR/rdonlyres/3B36137E-C5FE-487B-1
A18B-4D7281D88EF7/0/complexity.pdf. Accessed 31st of August 2005.2
Beck, U. Risikogesellschaft: Auf dem Weg in eine andere Moderne, Suhrkamp Verlag, 3
Frankfurt am Main, 1986.4
Beck, U. “The Reinvention Of Politics: Towards a Theory of Reflexive Modernization,” in 5
Beck, U., Giddens, A., Lash, S., Reflexive Modernization: Politics, Tradition and 6
Aesthetics in the Modern Social Order, Polity Press, Cambridge, UK, 1994.7
Beck, U. World Risk Society, Polity Press, Cambridge, UK, 1999.8
Beck, U., Bonss, W., and Lau, C. “The Theory of Reflexive Modernization: Problematic, 9
Hypotheses and Research Programme,” Theory, Culture & Society, (20:2), 2003, pp. 10
1-33.11
Beck, U., Giddens, A., Lash, S. Reflexive Modernization: Politics, Tradition and Aesthetics in 12
the Modern Social Order, Polity Press, Cambridge, UK, 1994.13
Berg, M. and Bowker, G. “The Multiple Bodies of the Medical Record: Toward a Sociology of 14
an Artifact,” The Sociological Quarterly (38:3), 1997, pp. 513-537.15
Berg, M., and Timmermans, S. “Orders and Their Others: On the Construction of 16
Universalities in Medical Work,” Configurations (8), 2000, pp. 31-61.17
Bijker, W.E. “Do Not Despair: There is Life After Aonstructivism,” Science, Technology & 18
Human Values (18), 1993, pp.113-38. 19
Bowker, G., and Star, S.L. “Knowledge and Infrastructure in International Information 20
Management: Problems of Classification and Coding” in Information Acumen: The 21
Understanding and Use of Knowledge in Modern Business, L.Bud-Frierman (ed.), 22
Routledge, London, 1994.23
Bowker, G., and Star, S.L. Sorting Things Out, MIT Press, Cambridge, MA, 1999.24
Brunsson, N. and Jacobsson, B. A World of Standards, Oxford University Press, 2000.25
43
Callon, M. “Techno-economic networks and irreversibility,” in A Sociology of Monsters: 1
Essays on Power, Technology and Domination, J. Law (ed.) Routledge, London, 2
1991, pp. 132-61.3
Ciborra, C. “E-government: Between Development and War,” In People and Computers, T. 4
Järvi and P. Reijonen (eds.), TUCS Publications, Turku, Finland, 2003.5
Ciborra, C. and Osei-Joehene, D. “Corporate ICT Infrastructures and Risk,” Proceedings of 6
the European Conference of Information Systems, Naples, June 2003. 7
Ciborra, C., Braa, K., Cordella, A., Dahlbom, B., Failla, A., Hanseth, O., Hepsø, V., 8
Ljungberg, J., Monteiro, E., Simon, K. From Control to Drift. The Dynamics of 9
Corporate Information Infrastructures, Oxford University Press, Oxford, 2000. 10
Cilliers, P. Complexity and Postmodernism: Understanding Complex Systems, Routledge, 11
London, 1998.12
Cussins, C. “Ontological Choreography: Agency for Women Patients in an Infertility Clinic,” 13
in Differences in Medicine: Unravelling Practices, Techniques and Bodies, M. Berg 14
and A. Mol (eds.), Duke University Press, Durham, NC, and London, 1998, pp. 166-15
201.16
David, P. and Greenstein, S. “The Economic of Compatibility Standards: An Introduction to 17
Recent Research,” Economics of Innovation and New Technology (1), 1990, :pp. 3-18
41.19
De Moor, G.J.E., McDonald, C., and van Goor, J.N., (eds.) Progress in standardization in 20
health care informatics, IOS Press, Amsterdam, 1993. 21
de Vries, H. J. Standardization: A Business Approach to the Role of National 22
Standardization Organizations, Kluwer Academic Publishers, Boston, 2003.23
Dechow, N., and Mouritsen, J. Enterprise resource planning systems, management control 24
and the quest for integration. Accounting, Organizations and Society, in press. 25
44
Ellingsen, G. Global Reach, Local Use: Design and Use of Electronic Patient Record 1
Systems in Large Hospitals, PhD Thesis, NTNU, Trondheim, 2002.2
Ellingsen, G and Monteiro, E. “A Patchwork Planet. Integration and Cooperation in 3
Hospitals,” Computer Supported Cooperative Work, (12:1), 2003a, pp. 71 – 95.4
Ellingsen, G and Monteiro, E. “Big is Beautiful. Electronic Patient Records in Norway 1980 –5
2000,” Methods of Information in Medicine, (42), 2003b.6
Fomin, V., Keil, T., Lyytinen, K. ”Theorising about Standardization: Integrating Fragments of 7
Process Theory in Light of Telecommunication Standardization Wars,” Sprouts: 8
Working Papers on Information Environments, Systems and Organizations, 2003.9
Forster, P. W., and King, J. L. "Information Infrastructure Standards in Heterogeneous 10
Sectors: Lessons from the Worldwide Air Cargo Community," in Standards for 11
Information Infrastructures, B. Kahin and J. Abbate (eds.) MIT Press, Cambridge, 12
MA, 1995, pp. 148-177.13
Giddens, A. Modernity and Self Identity: Self and Society in the Late-Modern Age, Polity14
Press, Cambridge, UK, 1991.15
Graham, I., Spinardi, G., Williams, R., and Webster, J. “The Dynamics of EDI Standards 16
Development,” Technology Analysis & Strategic Management (7:1), 1995, pp. 3-20. 17
Graham, I., Lobet-Maris, C., and Charles, D. “EDI Impact: Social & Economic Impact of 18
Electronic Data Interchange” (EDI. TEDIS project C9, report prepared for the 19
European Commission, http://www.ed.ac.uk/ ehjar36/tedis.html), 1996.20
Gregory, J. “Sorcerer's Apprentice: Creating the Electronic Health Record: Re-inventing 21
Medical Records and Patient Car,e, PhD Dissertation, Department of 22
Communication, University of California-San Diego, La Jolla, CA, 2000.23
Gregory, J. "When Logics of Illness, Patient Care Interactions, and Health Records Collide," 24
Paper presented at Public Proofs: Science, Technology and Democracy, 4S-EASST 25
45
Conference, Paris, France, 25-28 August, 2004. 1
Hanseth, O. “Gateways – Just as Important as Standards,” Knowledge, Technology and2
Policy, (14:3), 2001, pp. 71-89.3
Hanseth, O., and Braa, K. “Hunting for the Treasure at the End of the Rainbow. 4
Standardizing Corporate IT Infrastructure,” Computer Supported Cooperative Work 5
(CSCW),(10:3-4), 2001, pp. 261-292.6
Hanseth, O., and onteiro, E. “Inscribing Behaviour in Information Infrastructure Standards,”, 7
Accounting Management & Information Technology (7:4), 1997, pp 183-211.8
Hanseth, O., Ciborra, C., Braa, K. “The Control Devolution. ERP and the Side Effects of 9
Globalization,” The Data Base for Advances in Information Systems. Special issue: 10
Critical Analysis of ERP systems: The Macro Level. (32: 4), 2001, pp. 34-46.11
Hanseth, O., Monteiro, E., Hatling, M. ”Developing Information Infrastructure: The Tension 12
Between Standardization and Flexibility,” Science, Technology, & Human Values, 13
(21: 4), 1996, pp. 407-426. 14
Hawkins, R. "The Rise of Consortia in the Information and Communication Technology 15
Industries: Emerging Implications for Policy." Telecommunications Policy, 1999, pp. 16
159-173..17
Hughes, T.P. Networks of Power: Electrification in Western Society, 1880-1930, John 18
Hopkins University Press, Baltimore, MN, 1983. 19
Keen, P.G.W. Shaping the Future: Business Design through Information Technology, 20
Harvard Business School Press, Boston, MA, 1991.21
Klein, H.K., and Myers, M.D. “A Set of Principles for Conducting and Evaluating Interpretive 22
Field Studies in Information Systems,” MIS Quarterly, vol.23, no.1, 1999, pp.67-93.23
Latour, B., and Woolgar, S. Laboratory Life: The Construction of Scientific Facts, Princeton 24
University Press, Princeton, NJ, 1986.25
46
Latour, B. Science in Action, Harvard University Press, Cambridge, MA, 1988.1
Latour, B. “Is Re-Modernization Occurring – And If So, How to Prove It? A Commentary on 2
Ulrich Beck,” Theory, Culture & Society (20:2), 2003, pp. 35 – 38.3
Law, J. “Technology and Heterogeneous Engineering: The Case of the Portuguese 4
Expansion,” in The Social Construction of Technical Systems: New Directions in the 5
Sociology and History of Technology. W. E. Bijker, T P. Hughes and T. Pinch (eds.), 6
MIT Press, Cambridge, MA, 1987, pp. 111-134..7
Law, J. Organizing Modernity, Blackwell, Oxford, 1994.8
Law, J. “Landbroke Grove, Or How to Think about Failing Systems,” Manuscript, 9
http://www.comp.lancs.ac.uk/sociology/papers/law-ladbroke-grove-failing-10
systems.pdf, (last accessed November 2004), December, 2003a.11
Law, J. “Traduction/Trahison: Notes on ANT,” Manuscript,12
http://www.comp.lancs.ac.uk/sociology/papers/law-traduction-trahison.pdf, (last 13
accessed November 2004), November, 2003b.14
Law, J., and Bijker, W.E. “Postscript: Technology, Stability and Social Theory,” in Shaping15
Technology/Building Society, W.E. Bijker and J. Law (eds.), MIT Press, Cambridge, 16
MA, 1992, pp. 290-308.17
Law, J., and Mol, A. Complexities. Social Studies of Knowledge Practices, Duke University 18
Press, Durham, NC, and London, 2002.19
Law, J., Urry, J. “Enacting the Social”, Manuscript,20
http://www.comp.lancs.ac.uk/sociology/papers/law-urry-enacting-the-social.pdf, (last 21
accessed November 2004), 2003.22
Mahoney, J. (2000) Path-dependence in historical sociology. Theory and Society, Vol. 29. p-23
507-548.24
McDonalds, C.J. “ANSI's Health Informatics Planning Panel (HISPP) - The Purpose and 25
47
Progress,” in De Moor, G.D.E., McDonald, C.J., and Noothoven van Goor, J.(eds.). 1
Progress in Standardization in Health Care Informatic,. IOS Press, Amsterdam, 1993.2
Mol, A., and Berg, M. “Introduction,” in Differences in Medicine: Unravelling Practices, 3
Techniques and Bodies, M. Berg and A. Mol (eds.), Duke University Press, Durham, 4
NC, and London, 1998, pp. 1-12.5
National Health Service NHS. The Shared Care Record Review Document. Draft version, 6
July 5th 2005, URL: http://www.connectingforhealth.nhs.uk/crdb/docs/scrrdocument. 7
[accessed 2005 July 28]8
Perrow, C. Normal Accidents, Princeton University Press, Princeton, NJ, USA, 1999.9
Rolland, K. Re-inventing Information Infrastructure in Situated Practices of Use, PhD Thesis, 10
University of Oslo, 2003.11
Schneberger, S. L., and McLean, E. R. “The Complexity Cross: Implications for Practice”. 12
Communications of the ACM, (46 :9), September 2003, pp. 216-225.13
Shapiro, C.l, and Varian, H. R. Information Rules: a Strategic Guide to the Network 14
Econom,: Harvard Business School Press, Boston, MA, 1999.15
Shapiro, S., Richards, B., Rinow, M., and Schoechle, T. "Hybrid Standards Setting Solutions 16
for Today's Convergent Telecommunication Market," in 2nd Conference on 17
Standardization and Innovation in Information Technology. IEEE Boulder, CO, US,18
2001, p p. 348-35119
Star, S.L., and Griesemer, J.R. “Institutional Ecology, Translations and Boundary Objects: 20
Amateurs and Professionals in Berkeley’s Museum of Vertebrate Zoology, 1907-21
1939,” Social studies of Science (19), 1989, pp.387-420.22
Star, S. L., and Ruhleder, K. “The Ecology of Infrastructure: Design and Access for Large 23
Information Spaces,” Information Systems Research (7), 1996, pp. 111-133.24
Statens Helsetilsyn. Pasientjournalen: Innhold, gruppering og arkivering av 25
48
pasientdokumentasjon i somatiske sykehus, Document number IK-2451, 1993.1
The White House. Transforming Health Care: The President's Health Information 2
Technology Plan. 3
http://www.whitehouse.gov/infocus/technology/economic_policy200404/chap3.html4
[accessed 2005 July 28]5
Timmermans, S., and Berg, M. “Standardization in Action: Achieving Local Universality 6
through Medical Protocols,” Social Studies of Science (27), 1997, pp. 273-305.7
Timmermans, S., and Berg, M. The Gold Standard: The Challenge of Evidence-Based 8
Medicine and Standardization in Health Care, Temple University Press, Philadelphia, 9
PA, 2003.10
Urry, J. Global Complexity, Polity Press, Cambridge, UK, 2003.11
Vercoulen, F., and van Weberg, M. "Standard Selection Modes in Dynamic, Complex 12
Industries: Creating Hybrids between Market Selection and Negotiated Selection of 13
Standards," Maastricht: NIBOR, Netherlands Institute of Business Organization and 14
Strategy Research, 1998.15
Walsham, G. Interpreting Information Systems in Organizations, Wiley, 1993.16
Walsham, G. “Interpretive Case Study in IS Research: Nature and Method”, European 17
Journal of Information Systems (4), 1995, pp.74-8118
19
Author Biographies20
21Ole Hanseth, PhD, is Professor in the Department of Informatics, University of Oslo. His 22research focuses mainly on the interplay between social and technical issues in the 23development and use of large-scale networking applications. He is also Visiting Professor at 24London School of Economics, Department of Information Systems.25
26Edoardo Jacucci, MSc, is a PhD candidate at the University of Oslo, in the Department of 27Informatics at the Information Systems Group, Oslo, Norway. His research focuses on socio-28technical conceptualizations of standards and on processes of standard making in the health 29
49
care sector in Norway and in South Africa. He received his MSc degree in Information 1Systems Engineering at the Politecnico di Milano, Italy.2
3Miria Grisot, MA, is a PhD candidate at the University of Oslo, in the Department of 4Informatics at the Information Systems Group, Oslo, Norway. Her research interest is on 5issues of coordination related to the implementation of IT in hospitals. She holds a MA in 6Political Science from the University of Bologna, Italy.7
8Margunn Aanestad, PhD, holds a post Doctoral position in the Department of Informatics, 9University of Oslo. She worked within health care and telecommunications before her 10doctoral study of surgical telemedicine. Her research interests are broadly related to IT in 11health care.12
13