flexible generification: ictstandardization strategies and service … · actors, including...

EMPIRICAL RESEARCH

Flexible generification: ICT standardizationstrategies and service innovation in health care

Ole Hanseth1 andBendik Bygstad1

1University of Oslo, Oslo, Norway

Correspondence: Ole Hanseth, University ofOslo, Boks 1072 Blindern, Oslo 0316, Norway.Tel: +4795908509;

Received: 13 October 2013Revised: 24 April 20142nd Revision: 13 August 20143rd Revision: 27 November 2014Accepted: 22 December 2014

AbstractStandards have played an important but often unrecognized role in the develop-ment of modern organizations. This role is accentuated by today’s growth of largebusiness and government infrastructures, in the turbulent processes of globaliza-tion. In this paper, we investigate the relationships – and tensions – betweenstandardization strategies and service innovation in the health-care sector. Ourempirical material is seven longitudinal case studies in the Norwegian health-caresector, collected and analysed over a period of 20 years. We identify three genericstandardization strategies; anticipatory standardization, integrated solutions andflexible generification. We argue that the first two strategies do not support serviceinnovation while the strategy of flexible generification does so. We consider ourresults important for the evolution of the future ICT-enabled service economy.European Journal of Information Systems (2015) 24(6), 645–663.doi:10.1057/ejis.2015.1; published online 14 April 2015

Keywords: information infrastructure; standardization; service innovation; health care

IntroductionInnovation of new ICT-based services has transformed several industries,such as the financial sector, the music industry and the media. The serviceinnovations have taken many forms and the role of ICT has varied, fromintermediation in distribution, automation in inter-bank financials and co-creation of value in retail banking (Tidd & Bessant, 2009). A significant partof service innovations has emerged in large networks rather than in singleorganizations, making innovation in business ecologies (Busquets, 2010;Eaton et al, 2011) or information infrastructures (Henfridsson & Bygstad,2013) a key topic.These successes have raised high expectations to ICT-based innovation in

the health sector, which experiences strong growth in both mature andemerging economies. The potential of ICT-based service innovations inhealth is large. Such services are, for instance, expected to improve thelogistics and care of patients, supporting an increasing range of diagnosticand surgical instruments, and the prescription and distribution of medicines(Hovenga et al, 2010). ICT is widely held to have the inherent capacity ofenabling radically differentways of working and organizing than the existingones. Overall, it is expected to enable the sharing of vast amounts of medicalinformation between the different actors of health: the hospitals, doctors,labs, pharmacies, authorities and patients. The diversity is almost over-whelming, with thousands of different diseases with their individual treat-ments, a high degree of specialization and a heterogeneous organizationalecology of different actors and interests. ICT solutions must deal with thesecomplexities in specialized systems, but at the same time ensure that the

European Journal of Information Systems (2015) 24, 645–663© 2015 Operational Research Society Ltd. All rights reserved 0960-085X/15

www.palgrave-journals.com/ejis/

http://dx.doi.org/10.1057/ejis.2015.1http://www.palgrave-journals.com/ejis

overall logistics and patient-centred care are supported,while also addressing privacy concerns. Accordingly, webelieve e-health represents a good example of the growingcomplexities related to ICT in many sectors, whichNorthrop et al (2006) describe as the emergence of whatthey call ‘ultra large scale ICT solutions’.With this background, it is clear that successful innova-

tion in e-health is quite demanding. Addressing thechallenges related to the above-mentioned growing com-plexity usually requires standards, in order to enable com-munication across organizations, and sharing databetween departments and professions. For example, if thehospital, the General Practitioner (GP), the lab and thepharmacy shall share patient data, their IT systems mustrelate to defined standards. Standards are socio-technicalcomposites; they will often include formats of records,protocols (i.e., sequences of actions) and application pro-gram interfaces. Standards are defined by a diverse set ofactors, including international bodies, such as Interna-tional Standardization Organization (ISO), industry con-sortia, such as the Open Group, and individual vendors,such as Microsoft. This variety of standards and standardi-zation processes is reflected in de Vries’ (1999) definitionof standardization, which he arrived at after reviewing anddiscussing a huge number of such definitions:

… the activity of establishing and recording a limitedset of solutions to actual or potential matching pro-blems directed at benefits for the party or partiesinvolved balancing their needs and intending andexpecting that these solutions will be repeatedly orcontinuously used during a certain period by a sub-stantial number of the parties for whom they aremeant. (p. 155)

In line with this, he defines a standard as,

limited set[s] of solutions to actual or potential match-ing problems directed at benefits for the party orparties involved, balancing their needs and intendingand expecting that these solutions will be repeatedlyor continuously used during a certain period by asubstantial number of the parties for whom they aremeant. (de Vries, 1999, p. 13)

This paper is based on these definitions.It is well-documented that standards constitute an

important factor for macro-economic growth in advancedeconomies, and also play an important role for innovationof services in high-tech sectors such as engineering, tele-com and ICT (Blind, 2013). Standards, however, are adouble-edged sword; while they enable communicationand co-operation they may also constitute barriers toinnovation (Hanseth et al, 1996). For instance, if a stan-dard is technically complex, it adds an extra burden toinnovation, while a standard that is too narrowly definedmay effectively stop innovation. On the other hand, if astandard is simple, effective and easy to use, it may spur anumber of innovations, as the HTTP and IP standardsenabled the innovative growth of the Internet (Hanseth &Lyytinen, 2010). It is therefore crucial to understand how

the development, implementation and diffusion of stan-dards can enable and support service innovation, insteadof hindering it. We suggest the term standardization strate-gies, as a lens to study this phenomenon. What is lackingin the current literature is (a) a conscious conceptualiza-tion of standardization strategies as a phenomenon andtheir role in innovation, (b) a research-based typology ofsuch strategies and (c) an empirically based assessment ofwhich strategies that have proven to be efficient in variouscontext. In this paper we make a systematic effort tocontribute to this knowledge gap, focusing on the healthsector. In order to do so, we build on the lessons learnedfrom 20 years of investigation.Traditionally, standards are developed within the frame-

work of a standardization body like the ISO. Such bodieshave detailed rules specifying how the activities should beorganized, who are allowed to participate, voting rules,criteria a standard’s specification must fulfil to beapproved. The standards are specified based on anticipa-tion about what the users’ future needs will be. Thestandard is then, hopefully, implemented in solutions thatare adopted by the anticipated users. This is the waystandards have been settled within, for instance, telecom-munications, and this strategy was also adopted by the ITindustry when standards emerged as an important issuewithin this sector. However, the formal approach of moststandardization bodies has often been considered too slowand bureaucratic for the IT industry. Accordingly, variousconsortium models have been preferred in many cases,such as The Open Group (theopengroup.org). Anotherstandardization strategy is the one followed by the Inter-net community, where standards are developed andsettled in a much more bottom-up approach. A standardis not finally approved as such until, first, the need for it isproved by real use of solutions implementing the standard,and, second, several independent implementations of thestandard have proved to interoperate.The growing number of digital systems in health care

makes standards crucial. Accordingly, continuous ICT-enabled service innovation in health care requires standar-dization strategies that support this. There are great poli-tical expectations about the contributions from ICT forservice innovation and effectiveness in the health-caresector. Moreover, it is a sector with a plethora of standards,which presents a number of large cases to learn from andtheorize on. Our key term, standardization strategies,denotes the strategies or approaches chosen for developingnew standards, that is, how the development of newstandards are organized and which steps are taken orprocedures followed in the definition and implementationof a new standard. In addition, the concept of standardiza-tion strategy also includes how actors relate to standards ininnovation projects more broadly, that is, which existingstandards they choose, how they implement them, howthey are used in inter-organizational integration and howthey are changed over time.Our research question is then: Which standardization

strategies support service innovation in the health-care sector?

Flexible generification Ole Hanseth and Bendik Bygstad646

European Journal of Information Systems

We investigate which strategies are most successful interms of leading to ICT solutions that eventually arewidely adopted. At the same time we aim at understandinghow adoption and use of the solutions based on thesestandards enable and contribute to service innovation.This means that the success of a standardization strategyis ultimately measured in terms of the benefits that thenew services, which the standards are a part of andenabling, deliver to their adopters.We identify three different strategies for developing

standards, which we name (i) anticipatory standardiza-tion, (ii) integrated solutions and (iii) flexible generifica-tion. Of these three the last one is most successful, in thesense that it best supports service innovation. While weprimarily focus on the relations between standardizationstrategies and service innovation, standardization strate-gies have, of course, a broader impact, not the least uponindividual actors or stakeholders possibilities to influencethe definition and evolution of a standard.We proceed in the next section by presenting our

theoretical approach of information infrastructures, andassessing some key research on ICT standards and serviceinnovation. Then, our research approach is describedbefore we present the main body of the paper, a collectionof cases from the Norwegian health sector researched overa period of 20 years. Finally, we discuss our findings andpropose conclusions. Our contribution sheds new lightover the role of ICT standards in service innovation. First,we conceptualize the phenomenon of standardizationstrategies as the process of developing and diffusing stan-dards. Second, we identified three types of such strategies,which we called anticipatory standardization, integratedsolutions and flexible generification. Third, we investi-gated in depth the results of the three strategies in theNorwegian health sector and found that only the strategyof flexible generification supports service innovation inthe health sector.

Related researchIn this section we first briefly review the research onservice innovation and the role of ICT within this. Next,we present our theoretical approach, the information infra-structure perspective. Then, we discuss the role of stan-dards, focusing on the tension between standardizationand flexibility and the concepts of flexible standards andgenerification.

Service innovation and ICTTidd & Hull (2003) described service innovation as theorganizational response to technological opportunitiesand market imperatives. The full diversity of services andthe complexity of their inner attributes make it difficult todetermine a single definition of service provision. Ratherthan proposing a complete definition, one way to clarifythe nature of services can be to look at their distinguishingcharacteristics as compared with tangible goods (Nardelli,2012). Although the differentiation is much more blurred

in the actual practice than in theory, services tend toinvolve customer participation in the service processand to be: (a) simultaneously produced and consumed;(b) perishable; (c) intangible; and (d) heterogeneous(Fitzsimmons & Fitzsimmons, 2000). Each service is char-acterized by a unique combination of these attributes andrelative degrees. According to den Hertog (2000), fourmain dimensions describe a new service: (1) new serviceconcept; (2) new client interface; (3) new service deliverysystem; (4) technological options. Any service innovationinvolves a certain blend of these dimensions.The role of ICT in modern service innovation is central,

and has been documented in a number of spectacularinnovations, such as Internet banking and the musicindustry (Tidd & Hull, 2003), and in companies such asGoogle, Amazon and Apple. These developments havehighlighted that the arena for ICT-based service inno-vation has changed from single organizations to net-works and ecosystems. In a recent contribution Lusch &Nambisan (2014) proposed a three-part framework forresearch on ICT-based service innovation in digitalinfrastructures.Service ecosystems are emergent, relatively self-contained,

self-adjusting systems of loosely coupled actors whoexchange service and co-create value. Researchers need tounderstand how the different elements of digital infra-structures interact in their evolution (see, for example,Henfridsson & Bygstad, 2013). Researchers also need tounderstand how digital infrastructures balance the needfor structural integrity vs the need for structural flexibility.Service platforms are key parts of ecosystems; they are

modular structures that facilitate the interaction of actorsand resources. They enhance innovation through liquefy-ing resources and increasing resource density. Inno-vation is a process of combining resources (Schumpeter,2012), and platforms serve both to make these resourcesavailable in a structured way, and to provide an arena forexperimentation.Value co-creation is the processes that underlie resource

integration, and incorporate different actors in the serviceecosystem. Value is co-created by the service provider andthe customer through resource integration (Lusch &Nambisan, 2014). The term value co-creation illustratesthat a service should not only be seen as an output from avendor, but also how it is experienced and used.These issues are highly relevant for research on service

innovation in the e-health. The health sector currentlyinvests large amounts in ICT solutions, with great expecta-tions to both efficiency and innovations (Hovenga et al,2010). Some researchers have argued that these expecta-tions are unreasonable, because the scale (the number ofactors), the scope (the breadth of application types) andthe depth (the heterogeneity of data and data types) arevastly more demanding than the well-known successesfrom other sectors (Sauer & Willcocks, 2007). The historyof health informatics is characterized by large promisesthat failed to materialize, for many reasons (Hovenga et al,2010). One reason for these disappointments is that

Flexible generification Ole Hanseth and Bendik Bygstad 647


service innovation initiatives in e-health have oftenfocused on systems rather than ecosystems. In order toanalyse the importance and role of standards in thedevelopment of these ecosystems, we conceptualize themas information infrastructures.

Information infrastructures and standardsICT solutions supporting new services are shared by bothservice providers and service consumers. Often they areshared by most members of a business sector (like electro-nic payment services, airline booking and health care).This makes the solutions and their development complexbecause the solutions include a large number of technolo-gical components at the same time as a large number ofdevelopment and user organizations are involved. Muchresearch into the challenges related to the developmentand use of such ICT solutions are conducted by concep-tualizing them as information infrastructures (Star &Ruhleder, 1996; Hanseth & Lyytinen, 2010). An informa-tion infrastructure is defined as ‘a shared, evolving, open,standardized, and heterogeneous installed base’ (Hanseth,2002).

● Shared: An information infrastructure is shared by themembers of a community, including vendors, users andstaff.

● Evolving: It is not ‘designed’, but evolves continually, asgrowth and innovation expands it.

● Open: In principle, there is no limit on the number ofusers.

● Standardized: Infrastructures rest on standards, whichallow for scaling and interoperability.

● Heterogeneous: Infrastructures consist of different ele-ments; technology, users, organizations and so on.

● Installed base: Such structures are seldom created fromscratch; rather they grow from existing practices andinfrastructures.

An information infrastructure is a socio-technical sys-tem. In the health-care sector, for example, such structuresmay encompass several patient record (EPR) systems,hundreds of medical units and tens of thousands of users.Standards are core elements of information infrastructures.A large part of the research on information infrastructurehas focused on, and disclosed, a very dense and complexweb of relations between technical and social issues andelements of the standards (Star & Ruhleder, 1996; Hanseth& Monteiro, 1997; Brunsson et al, 2012).

Standards and standardization strategiesResearch on ICT standards is in its infancy, but its volumeas well as assumed importance is growing (e.g., illustratedby the Special Issue of MIS Quarterly on the topic pub-lished in 2006 (Lyytinen & King, 2006) and OrganizationScience in 2012 (Brunsson et al, 2012)). This reflects thegrowth in number of and importance attributed to ICTstandards (Romer, 1990; Schmidt & Werle, 1998;Brunsson & Jacobsson, 2002; Lyytinen & King, 2006;

Brunsson et al, 2012), but it also reflects a rapid growthin the variety of standards. Barry (2001), for instance,argues that the European Union should primarily be seenas a standardization effort aiming at developing thestandards needed for European integration. This integra-tion requires traditional technical standards but alsostandards like the Euro currency as well as standards forchocolate and the shape of cucumbers. The latter kindof standards will, however, be closely related to andembedded into ICT standards implemented by the vastrange of ICT solutions required to make the envisionedEuropean integration real. As the number of standards aregrowing, so are also the connections and interdependen-cies between them, which generates a need for focusingon ecologies of standards rather than individual ones(Nickerson & zur Muehlen, 2006).The growth in research on standards also reflects a

change in requirements of standards because of a morecomplex and rapidly changing world caused by, forinstance, globalization processes. This growing complexityspeed of change again leads, according to Brunsson et al(2012), to a need for more dynamic and flexible standardsas well as more research on the dynamics of standards. Thisgenerates new requirements to standardization processesand their organization. Updegrove (2007), for instance,argues that the exiting standardization system has its rootsin the industrial era of the 19th century and was designedto address the needs of such an economy and is ill-suitedto meet the demands of the 21st century. According toUpdegrove (2007), the need to meet such changes isreflected in the rapid growth of consortia as an alter-native approach to standards making compared with thetraditional formal bodies. Brunsson & Jacobsson (2002)summarize these trends into what they describe as theemergence of a whole new ‘world of standards’.

The dynamics of standards Traditionally standards havebeen seen as stable after their definition and, accordingly,their dynamics has been described only as a sequence ofstages: definition, implementation, diffusion and use.This stage model has been the foundation of most formalstandardization activities, like in telecommunication andalso widely adopted in areas like health care. Assumptionsabout the stability of standards have led to a top-downspecification-driven strategy for standards developmentwhere the standards are developed and settled by standar-dization committees where all relevant stakeholders parti-cipate. This model is in contrast with the one used inthe development of Internet standards that is based on amore bottom-up, experimental and evolutionary approach(Hanseth et al, 1996). The Internet model is experimentalalong two dimensions: the development of individualstandards and the development of the Internet as a whole.Abbate (1999) describes how the Internet is developedlayer by layer from the bottom. When the standards atone level stabilize, this layer serves as a platform forexperimental development of services and their standards



at the next level. More specifically, Internet standards spe-cifications go through stages of development, testing andacceptance. These stages are formally labelled ‘maturitylevels’ called ‘Proposed Standard’, ‘Draft Standard’ and‘(Full) Standard’, respectively (RFC, 1996).A proposed standard specification should have resolved

known design choices, is believed to be well-understood,has received significant community review and appears toenjoy enough community interest to be considered valu-able. However, further experience might result in a changeor even retraction of the specification before it advances.A specification from which at least two independent andinteroperable implementations from different code baseshave been developed, and for which sufficient successfuloperational experience has been obtained, may be elevatedto the ‘Draft Standard’ level. Finally, a specification forwhich significant implementation and successful opera-tional experience has been obtained may be elevated tothe Internet standard level. In addition, the InternetSociety also registers ‘Experimental’ standards that aretypically specified as a part of some research or develop-ment effort.Fomin et al (2003) propose a dynamic process model of

standardization, called the Design-Negotiation-Sensemak-ing model (D-N-S) model, which sees development anddiffusion of standards as integrated and overlapping. Theirmodel integrates separate lines of enquiry to standardiza-tion activities including Simon’s theory of artefact design(D), Weick’s concept of sense-making (S) and Latour’sconcept of negotiation in socio-technical networks (N),and organizes them into a hierarchically organized web ofstandardization events. Botzem & Dobusch (2012) focusalso on the long-term evolution of standards and propose aprocess model where the evolution of a standard is seen aswhat they call a recursive relationship between input andoutput legitimacy, that is, the legitimacy a standardobtains through participation in the definition of a stan-dard, and the legitimacy a standard obtains based on itsadoption. Hanseth et al (1996), based on a comparativestudy of the ISO/OSI and the Internet approaches tostandardization, see the tension between the need forstability and standardization on the one hand and theneed for flexibility and change on the other. This tensionemerges from a combination of mutual support and con-tradiction. On the one hand, one component’s stabiliza-tion through its standardization makes the overall systemmore flexible in terms of enabling the development, andfurther on standardization, of new and additional compo-nents extending the system as a whole. On the other hand,the development of additional components produces newrequirements to the already-standardized components,that is, generates a need to change them. At the same time,a complex technological system needs to change overtime to be sustainable. However, all components cannotchange all the time while keeping the system working.Some components need to be kept stable when others arechanging while also some components need to change ifothers are to be kept stable.

Pollock et al (2007) argue that successful commercialsoftware has proved to be effective standards. The past twodecades have witnessed a great shift in how organizationsacquire software systems; from developing proprietary andtailored solutions to buying standard software packages,such as SAP. Mainstream information systems research hasemphasized the benefits of this approach; large softwarepackages are relatively cheap, and they contribute to‘streamlining’ (i.e., standardize) the processes of the orga-nization. Other researchers have investigated implementa-tion of such solutions and often found that substantiallocal appropriation is necessary (Avgerou, 2002).Further, Pollock et al (2007) argue that researchers tend

to over-emphasize the gap between localized practices andstandardized, generic solutions. They found that, overtime, there is a subtle and complex interplay betweensuppliers, consultants and users that manages the balancebetween standards and organizational diversity. Thisincludes, for example, flexible configuration, the construc-tion of templates, the smoothing of differences and thegeneralization of requirements. This process is institutio-nalized and managed by the ‘software community’, andthat the standard evolves in a way which Pollock et al(2007) call generification of software packages, that is, thatsoftware packages can serve an increasingly larger numberof organizations. This process then is a dynamic standardi-zation process where a specific solution is made moregeneral to serve the needs of a larger user community, thatis, to work as an increasingly more general standard.A rich picture of standards dynamics, and the issues

involved, in the ‘new world of standards’ is described anddiscussed in Egyedi & Blind (2008) by means of a series ofcase studies. They illustrate that there are significantvarieties regarding scope and stability among standards.When standards change, problems may emerge related tothe accessibility of archival data and incompatibilitybetween components implementing the standard. Stan-dards change because the context they need to fit intochanges. However, how they change is influenced by thenature of competition between different standards andtechnologies. Furthermore, a standard may change indifferent ways. Egyedi & Blind (2008) identify three suchways which they name grafting (i.e., a new standard isdefined based on another (predecessor) with the intentionto improve the latter’s functionality and/or usefulness inother respects (e.g., ease or cost of implementation) whilepreserving compatibility), extension (i.e., adding featuresto an existing standard) and revolution (i.e., replacing anexisting standard with a radically different one), respec-tively. They also describe some strategies to cope withincompatibilities between different versions of standards.One such strategy is buying larger collections of softwarefrom one and the same vendor.

The flexibility of standards The more dynamic world ofstandards generates new and stronger requirementsregarding standards’ flexibility. van den Ende et al (2012)



argue that the more flexible a standard is, the easier it is toadopt and the more successful it is. They draw this con-clusion based on an analysis of three standards battles: Blu-ray vs HD-DVD, USB vs Firewire and WiFi vs HomeRF.A standard’s flexibility depends on its specific features.

Hanseth et al (1996) make a distinction between use andchange flexibility. Use flexibility means that a standard canbe used in many different ways and in different areaswithout being changed, while change flexibility meansthat the standard is easy to change. They further point outthat these two kinds of flexibility are related and that astandard’s total flexibility is the sum of these two. Exten-sive use flexibility implies that a standard does not need tobe changed that often, and, oppositely, limited use flex-ibility means that a standard needs to be changed moreoften and, accordingly, the standard needs to be easier tochange. Hanseth et al (1996) point to one aspect of astandard making it flexible to change: simplicity. Further,they argue that a standard’s simplicity depends on theapproach to defining a standard. They argue that a top-down specification-driven approach will often lead tomore complex, and accordingly less flexible, standardsthan a more bottom-up and evolutionary process. Thisargument is supported by the analysis of the ISO/OSI andInternet standardization approaches.Building upon Hanseth et al (1996), Braa et al (2007)

present a more elaborated framework for developing infra-structures and their standards, which they call the ‘flexiblestandards strategy’. The framework is based on an exten-sive action research programme developing informationinfrastructures for health care in a huge number of coun-tries in Africa and Asia. Using complexity theories tointerpret the outcomes of the action research, they pro-pose a strategy with two main components. First, creatingan attractor that emerges as a new standard and thatevolves into a system of standards. Second, suggesting thatthe individual standards must be crafted in a manner thatallows the whole complex system of standards to beadaptive to the local context. Furthermore, the strategy isbased on two somewhat paradoxical principles that theycall the principle of flexible standards, and the principle ofintegrated independence. They illustrate how this approachis supported by obtaining ‘rich information from minimaldata’, how ‘radical change’ can be achieved through takingsmall steps, and how gateways and a focus on datastandards (as opposed to technical standards) are theimportant components when aiming at scaling nationalhealth information systems in developing countries.The flexibility of a larger system made up of standar-

dized components does not only depend on the flexibilityof the individual standards, but also on the architecture ofthe whole system. An architecture based on the location offunctions close to the application that uses the function,the so-called end-to-end architecture, is one example ofproviding flexibility by systems design (Saltzer et al, 1984).The point of this architecture is that functionality incommunication networks only can be appropriatelyimplemented if based on knowledge that only exists close

to the applications standing at the endpoints of a commu-nication system. Thus, the network should not controlhow it grows, the applications should. Lessig (2001)exemplifies this argument by illustrating the Internet as anetwork where intelligence is in the fringes. Since thenetwork is not optimized for any application but open forand inviting the unexpected and surprising, innovationscan flourish without changes in standards. While stan-dards nurture and sometimes are the very preconditionsfor innovation, the interrelationship between innovationand standards is intricate. Standards may, for instance,result in future innovations being hampered by previousinnovations that now are de facto standards in a market(Dunphy et al, 1996). Because of an increasing installedbase, not only does the cost of switching and changingstandards become higher but innovations are required toconform to existing standards. Research has also foundthat there is a subtle and bidirectional relationshipbetween standards and service innovation in networks;standards provide a platform on which services may beinnovated on, but the innovation process also drives theneed for new standards (Tidd & Hull, 2003; EuropeanCommission, 2011; Wakke et al, 2012).

Summing upAs this review has shown, the knowledge on the impor-tance of, and the dynamics of standards, has increasedsignificantly over the past decade. In particular, in analys-ing the spectacular success of the Internet researchers hashighlighted the centrality of standards, and the organicdevelopment of them. This research has changed thestandardization field from a technical speciality to a muchbroader discourse on evolution of the world’s socio-tech-nical infrastructures.What is lacking in the current literature is (a) a conscious

conceptualization of standardization strategies as a phe-nomenon and their role in innovation, (b) a research-based typology of such strategies and (c) an empiricallybased assessment of which strategies that have proven tobe efficient in various context. In this paper we make asystematic effort to contribute to this knowledge gap,focusing on the health sector. In order to do so, we buildon the lessons learned from 20 years of investigation.

MethodOur research has been motivated by a strong interest tounderstand the development of large information infra-structures in health as they evolve at different levels.Studying these large socio-technical structures over timeis challenging because of the complexity of the domain;the number of actors and initiatives is high, and theprojects often last for several years. The significance ofstandards has been prominent because they played acrucial role at different levels.Our research approach has been a multilevel, longitudi-

nal case study (Pettigrew, 1985; Miles & Huberman, 1994;Hitt et al, 2007). Specific projects have been studied in



detail over time and documented extensively. At higherlevels regional and national initiatives (with internationallinks) have been followed, and we have been particularlyinterested to investigate the dynamics between differentlevels.

Case selectionWe have followed the development of the Norwegianhealth systems over a period of 20 years, more or lesscontinuously from 1992 to 2013. Norway is a Scandinaviancountry of 5 million inhabitants, living in a large (385,000km2) and rugged (mountains and fjords) territory. Theministry of Health has the political responsibility, whilethe Directorate of Health co-ordinates e-health initiatives.The primary health service is the responsibility of 428municipalities, while secondary services are organized in4 Health Regions with 70 hospitals.The cases were chosen based on a combination of

systematic and pragmatic reasons. First, we have followedthe central government-sponsored initiatives, startingwith the standardization initiatives in the early 1990s,and continuing with some large and long-lasting projects.These projects reflect the strategic thinking, in particularon the issue of standardization, of the central healthauthorities over the studied period. Second, during ourresearch we identified three smaller projects that weredifferent; they were local or private initiatives, and theyhad alternative approaches to standardization. We choseto follow these projects over time, since they offered anopportunity to compare and contrast the dominatingstandardization paradigm with alternative strategies.

Data collectionThe most important source has been interviews withinformants in different roles; doctors and nurses, linemanagers, IT professionals, staff users and high-levelbureaucrats. Interviews focused on three topics:

● The interaction between government agencies, ICTspecialists and the medical professions in understandingneeds and developing standards and solutions

● The actual use of ICT solutions by the different organi-zations and users

● The overall interplay of technology and innovation;how ICT solutions enable (or hinder) the develop-ment of new work processes and innovation of newservices

Hundreds of reports have been collected and analysed.At many occasions, solutions have been demonstrated,and we have observed systems in use in many situations.Table 1 shows an overview of the cases. The seven projectshave been followed over a long period of time, and severalinformants have been interviewed several times. In orderto capture informants’ reflections over the cases, results ofthe projects have also been discussed with informantsmany years after the projects finished.

Data analysisFirst, each case was analysed separately, focusing on therole of standards in the innovation process. We haveanalysed the development of solutions, the innovationprocesses and the use of standards, and the developmentof new standards. For each we identified key elements, andtheir interaction, in the innovation process, and the out-come of the initiative (Langley, 1999).The full portfolio of cases was analysed in two dimen-

sions (Pettigrew, 1985).A comprehensive analysis was conducted, focusing on the

links between international and national standards andthe outcome of the different cases. A central part of theanalysis was the role of standards in designing the overallsolution. For example, comparing ePrescription 1 andePrescription 2 revealed significant differences regardingthe role of standards, in two projects that had fairly similargoals and objectives. These differences served as input toidentify the three standardization strategies.Then a temporal analysis was done, focusing on the

development over time. This analysis documented thetrajectories of the projects, but also of the various dis-courses in the sector. For example, the forward chaining ofevents served to explain intentions of stakeholders, whilebackwards chaining of events served to explain outcomes.Overall, the temporal analysis helped to understand thedynamics of the standardization strategies. Combiningthese analyses resulted in the identification of three dis-tinct strategies (shown in Table 2), which are described indetail. We theorized our findings (Langley, 1999) to astandardization process framework, shown in Table 3.

Validation of resultsValidation of results focused on the relationship betweenstandardization strategies and the identified outcomes.We assessed our results carefully, asking whether theidentified outcomes could be explained by other factorsthan the standardization processes. For instance, projectswere different in size and context, and we assessedwhether these differences in project attributes had influ-enced outcomes. We also examined whether technicalaspects of the different solutions could explain the differ-ences in outcomes.

Table 1 Cases and data collection

Case Type of project Period ofproject

Period ofdata

collection

1. CEN TC/251, etc. Standardizationinitiatives

1992–1996 1989–1996

2. ePrescription 1 Pilot project 1993–1996 1990–19963. The Elin project Regional project 2004–Present 2004–20094. ePrescription 2 Large national project 2004–Present 2008–20115. Fürst’s Med Lab Commercial project 1987–Present 1984–1996,

2009–20106. NNHN Regional project 1997–2003 2005–20107. DIPS Interactor Commercial project 2006–Present 2006–2011



Our examination revealed that although the projectswere quite different, the overall pattern was clear and well-supported by the evidence. In addition we asked for infor-mants’ feedback at case level and policy levels by discussingpreliminary results and issues with key informants andpolicymakers. Their input provided corrections and amend-ments, but also served as input to more analysis.

Standardizing information exchange in healthcare in NorwayThis section presents the activities related to the defini-tion/specification, implementation and use of standardsfor information exchange between health-care institutionsin Norway since the first of these activities started around1987 and up until now (October 2013). We identify threedifferent strategies for developing standards, as illustratedin Table 2.In the following, we describe the content of the strate-

gies; the kinds of standards developed; their implementa-tion in ICT solutions and infrastructures; the use of thesesolutions and infrastructures; and, finally the degree ofservice innovation and the organizational benefits gained.One of the standardization strategies was explicitly

chosen after extensive discussions around 1990. A broadconsensus was reached about the need for and importanceof standards and how to develop them. What we here callanticipatory standardization has ever since been the offi-cial and dominant strategy. However, in our empiricalmaterial we identify two other strategies. These were not,however, explicitly formulated as strategies, nor were theydiscussed as such. They were mere emergent strategies(Mintzberg, 1978), and they were more ‘strategies-in use’than ‘espoused strategies’.

When the standardization activities started in Norwayaround 1990, this was a part of a broad international trend,and a range of ISOs was established. Later many moreorganizations have emerged related to health care. For anoverview of the current situation, see ITU (2012). Thestandardization activities presented in this paper havebeen integrated with some of these international activities,in particular those taking place within CEN TC/251. Over-all standards cover a broad range of different areas relatedto health care like standards for transmission of data fromsensors, video compression, security, patient records, epi-demiological data and so on. Moreover, activities related tothe integration of social media and other health-caresystems have recently emerged. Currently, standards forintegrating mobile phone apps and new sensor technolo-gies to other health-care systems seem to be a hot topic.Some standards focus on a narrow field, like video com-pression, while other standards, like integrated healthenterprise, specified a framework for how to combine abroad range of individual standards in the establishmentof large-scale information infrastructures.Different types of standards have different characteristics

and their development, implementation and adoption raisedifferent challenges. This is certainly the case for purelytechnical standards, like standards for integrating newsensors to mobile phones, and standards enabling andsupporting new working practices and ways of organizing.The standards focused in this paper are of the latter kind.

Strategy 1: Anticipatory standardizationThe development of solutions for electronic informationexchange between health-care institutions in Norway

Table 2 Standardization strategies

Standardization strategy Description Cases

1. Anticipatory standardization Top-down standard process, worked out as detailed compromises 1. CEN TC/251, KITH2. ePrescription 1

2. Integrated solutions User-driven projects, with standards as part of requirements specifications 3. The Elin project4. ePrescription 2

3. Flexible generification Work processes and actual use determine standards, which are adapted pragmatically 5. Fürst6. NNHN7. DIPS Interactor

Table 3 Standardization process framework

Standardization strategy Process steps Outcome for serviceinnovation

1 2 3

1. Anticipatorystandardization

Top-down definition ofstandards

Diffusion of standards Implementation of standards insolutions

Hinders innovation

2. Integrated solutions Specification of solution andstandards

Development of standardizedsolutions

Implementation of solutions Slows downinnovation

3. Flexiblegenerification

Innovating work processes Development of solutions Standardizing solutions Supports innovation



started when a private laboratory, Dr. Fürst’s MedicineLaboratory (Fürst) in Oslo, developed a system for trans-mission of lab report to GPs in 1987. The system was verysimple and was developed in only 3 weeks by one person.The interest of Fürst was simply to make a profit byattracting new customers. Each GP received on averageapproximately 20 reports a day, which took quite sometime to register manually in their medical record system. Itwas assumed that the system would help GPs save muchtime otherwise spent on manual registering and that theGPs would find this attractive.The system proved to be a commercial success and

brought Fürst lots of GPs as new customers. This impliedless business for the other labs. Within a few years, mostnon-private labs (in hospitals) developed or bought sys-tems with similar functionality in order to be competitive.These systems were designed more or less as blueprints ofFürst’s system. There were, however, some minor, butimportant differences that we will return to.

CEN TC/251, KITH Alongside the growing number of labsadopting systems for the exchange of reports, an increas-ing number of actors envisioned a wider range of applica-tions of similar technology in other areas. These actorsbelonged to the health-care sector as well as possible ven-dors of such technology. They all perceived it as importantthat the technologies could be shared among as manygroups as possible in order to reduce costs and enableinterconnection of a wide range of institutions. All of themalso agreed that standards were crucial to achieving this.Similar developments were also taking place in othercountries.In 1990, the Commission of the European Community

delegated to CEN (Comité Europeen de Normalization, theEuropean branch of ISO) to take responsibility for workingout European standards within the health-care domain inorder to facilitate the economic benefits of a Europeaninner market. CEN established a so-called technical com-mittee (TC/251) on 23 March 1990 dedicated to thedevelopment of standards within health-care informatics.Being a formal standardization body, CEN TC/251’s stan-dardization strategy had to be the one of anticipatorystandardization.In 1989, the Ministry of Health in Norway decided that

standards should be developed, and after some initialactivities a standardization programme was set up in1991. The same year the Ministry also established KITH(Competence Centre for IT in Health), which was dele-gated the responsibility for standardization within ICT inhealth care and coordination of the standardization pro-gramme. KITH decided that the Norwegian standardiza-tion activities should be tightly integrated with those ofCEN TC/251. The focus of the programme was initially thestandardization of various EDIFACT messages and theirso-called implementation guides. Early versions of theEDIFACT message for lab reports were implemented insome of the solutions that were established in the early

nineties while others were introduced in a series of pilotprojects. We will here describe the activity related to onesuch message standard; the message to be used for electro-nic transmission of prescriptions.

The ePrescription project 1 The choice of a standardiza-tion model in this project was not strictly given from theoutset, but adopting EDIFACT as the basis for electronicprescriptions seemed inevitable even though alternativeswere proposed. A representative of one of the vendors ofelectronic medical record systems (Profdoc) suggestedearly on that bar codes should be used instead of electronicmessages. Another alternative to EDIFACT was suggestedby the health insurance authorities. They proposed anarchitecture where prescriptions were stored in a databaseinstead of being transmitted directly to the pharmacies.The important difference with this solution comparedwith a pure EDIFACT one, is that the pharmacies shouldretrieve the prescriptions only when the patient arrivedat the pharmacies. This difference entailed that thehealth insurance authorities no longer would pay for pre-scriptions that were never picked up. According to thehealth insurance authorities, this represented a substantialloss. The database solution would also offer the patientsthe freedom to choose which pharmacy to get thedrugs from, which was particularly important for reiter-ated prescriptions. In reality, however, the prescriptionproject never seriously considered deviating from anEDIFACT message-based solution in line with predomi-nant conceptions on politically correct standardizationstrategies.The message specification worked out in the pre-project

and the implementation guide was circulated to theparticipants involved in the Norwegian efforts for com-ments before proceeding with the main project. Reac-tions varied greatly. The comments from the vendors ofelectronic medical record systems were particularlyimportant, as they were vital to making an integration ofelectronic prescriptions and the GPs’ existing systemsfeasible. The two largest vendors expressed quite differentattitudes. One embraced the idea, as they already hadsome experience with similar work in Sweden on electro-nic prescriptions. They expected to be able to integrate aprescription module with their medical record systemrelatively quickly, thus giving them a leading edge oncompetitors. The other principal vendor of medicalrecord systems was quite hostile in their comments. Theyquestioned the very idea of electronic transmission ofprescriptions. They demanded that the scenarios shouldbe spelled out in more detail in order to make the benefitsmore visible. Some GPs commented that additionalinformation going beyond what was strictly a part of theprescription also had to be included in order to makethe systems implementing the standard possible to use.This included, for instance, information about whetherthe patient herself would pick up the drugs at thepharmacy or if others would do it.



On the basis of the feedback, new versions of theproposed standard were worked out and again sent out forcomments. However, the project maintained their focuson the specification of an electronic EDIFACT version ofthe paper-based prescriptions. All comments, both fromGPs and suppliers, which did not fit into this scheme, wereneglected (Pedersen, 1996).After producing a more comprehensive requirements

specification and a beta release, a pilot project in theBergen area started in 1994. The project was marred withproblems. Financing was not ensured, and was a recurringissue. An advanced solution had been specified, requiringseamless integration with other systems (such as welfaresystems, EPRs and medical registers), which were not yetavailable. The EPR vendors did not prioritize the solution,because of other development pressures and poor finan-cing. In the pilot version, there were so many errors thatthe doctors had to fax the prescription in addition to theelectronic transmission. In 1996, the project ran out ofsteam (and money), and a project report summarized theremaining problems: the need for a central medical regis-ter, data security concerns, paying for running costs anduser support.

Results of Strategy 1 The strategy adopted in the begin-ning of the nineties was the result of quite extensive dis-cussions about the needs for standards and how theyshould be developed and settled. There was a strong con-sensus that the traditional anticipatory standardizationstrategy was the right one. Partly independent of this spe-cification-driven approach, the standardization activitiesfocused on defining electronic versions of existing paper-based communication entities like lab orders and reports,prescriptions and so on. There was no attention paid tousers’ work practices, or how electronic informationexchange could help to develop improved services beyondspeeding up existing paper-based practices.Over the past 20 years a number of (draft) standards

have been specified. Some of the solutions for exchange oflab reports established in the early nineties implementeddraft versions of standard EDIFACT messages, whichhelped GPs remove a significant amount of manual dataentry work. With this exception, the implementation anddiffusion of the standards defined have been very slow.This has increasingly been viewed as a major problem innational strategic plans for ICT in health care (HOD, 1996,2008). These strategic plans have correspondingly repeatedthat the implementation of standards needs to speed up,and more pressure needs to be put on the various actors todo so. Weaknesses of Strategy 1 will be highlighted throughthe presentation and discussions of the other strategiesbelow.

Strategy 2: Integrated solutionsThe Elin project represented a new standardization strat-egy. The focus changed from anticipatory standardizationto ‘user-driven development of integrated ICT solutions

supporting communication and collaboration’. Thischange means that the focus shifted from the specificationof messages representing paper forms to the user require-ments and functionality needed to support collaborationinvolving GPs. This functionality was specified throughmore active user involvement, and then integratedsolutions satisfying these requirements were designed.Standards were developed as a part of the design of thesolutions.

The Elin project The Elin project was triggered by oppor-tunities created by the governmental programme BIT(Business sector-oriented IT projects). This programmeaimed at supporting the development of ICT solutions forspecific business sectors based on ‘committed collabora-tion’ among relevant actors within the business sector andsoftware suppliers.The aim of the Elin project was given by the framework

set by the BIT programme and the project manager’sexperience. He had been actively involved in the develop-ment of solutions for information exchange (both as anend user and as a user representative) since the earlynineties, in particular in the establishment of the NorthernNorwegian Health Network in 1997. The aim was todevelop requirements specifications as a basis for user-friendly standardized solutions for electronic health care-related communications for GPs. The vision was phrasedas ‘better communication and collaboration, and notjust development of technical solutions for messageexchange’. This included the development of solutionsfor exchange of admission and discharge letters, lab ordersand reports, illness and doctor’s declarations, prescriptionsand communication with patients. Also in the Elin projectthere was a strong focus on the main information objectsto be exchanged, like lab orders and reports, admission anddischarge letter, prescriptions and so on. However, incontrast to the previous standardization activities, theseobjects were viewed and understood more as integratedparts of the work practices in which they appeared.The project was split into three phases. In the first, the

focus was on exchange of discharge letters between GPsand hospital departments and outpatient clinics. In thesecond phase the focus was on exchange of dischargeletters between medical specialists’ offices and GPs,exchange of orders and reports between radiology labs toGPs, and information exchange with patients. The thirdphase focused on improving and piloting the technicalsolutions.The BIT programme had developed a framework for

their projects that was considered quite successful. Thisframework included specifications for how the projectsshould be organized (steering group, project groups etc.), aset of contracts between the participating organizations,criteria for selection of pilot organizations and a processmodel splitting the project into a number of phases: a pre-project for planning the project, a main project developinguser requirements and developing the solution, then



piloting, testing and finally diffusion. This framework wasfurther elaborated and adapted to the context of healthcare during the Elin project and given the name Elinmethodology.KITH had already proposed standards for the informa-

tion objects to be exchanged in the Elin solutions. How-ever, these did not fit well with the requirementsdeveloped in the project because KITH did not have asufficient understanding of the work and communicationprocesses involving orders and reports. Several importantobjects, such as response to admission letters (informingthe admitting physician what will happen to the patient)were not defined, and existing ones were inappropriatebecause they did not satisfy user requirements. Anotherchallenge was the fact that the existing standards were allbased on asynchronous (i.e., email-based) communica-tion. Although this corresponded to the logic of thepaper-based communication patterns, for various (andoften very strange and surprising) reasons, the exchangeof the messages took too long time. Accordingly, newstandards and messages had to be specified based on aweb services model.The project has played a major role in the development

of user requirements for ICT solutions supporting commu-nication between GPs and other health-care institutionsthat are well-aligned with user needs and requirementsfrom health-care authorities. The requirements have beenimplemented in solutions that have diffused to someextent, but the implementation and spreading of solutions

have definitively been much slower than expected. TheElin project was, according to Christensen & Grimsmo(2005), however, quite successful in establishing strong,enthusiastic, collaborative networks of users and suppliers.

ePrescription 2 In 2004, the Ministry of Health initiatedyet another pilot study on electronic prescriptions. Thebackground was a report in 2001 from the Office of theAuditor General that raised concerns on the accountabilityof prescription refunds from the Welfare AdministrationAgency (RTV). The following actors were included in thepilot study: Norwegian Pharmacist’s Union, NationalInsurance Administration, Norwegian Medical Association(representing physicians) and Norwegian MedicinesAgency. The Directorate of Health managed the project.The ePrescription project was established with direct

funding from the parliament of about €40 million fromthe Norwegian Parliament during the 6-year period from2005 to 2010. By the end of 2010, about €60 million hadbeen spent on the project. During 2006 several detailedrequirements specifications and architectural documentswere written, specifying an ambitious, fully integratedsolution. The Prescription Exchange was designed to han-dle 300million transactions per year. This reflected that, inthe designed solution, each prescription would generateapproximately 10 transactions, from a national volume ofaround 27 million prescriptions per year. As shown inFigure 1, the architectural solution was based on 31different (standardized) messages (each arrow in the figure

ePrescriptionsExchange

My Presciptions

EPJ-Systems

Pharmacy-system

Prescription

Prescription information

Hand-over message

Deleted prescription

ePrescriptions information

Prescription information

Hand-over message

Request for expedition

FEST(Gvt Medicine

Agency)

Application(Gvt Medicine

Agency)

Refunds and control(NAV)

ApplicationNAV

Refundrequest

Notificationof

hand-over

Prescription and expedition information

Recall

Reply onRefund request

Reply onapplication

Request for assessment byGvt Medicine Agency

Consent information

GPinformation

Information on medicins in use

Reference number

Reply from Medicine Agency

Prescription and expedition information

Application toMedicine Agency

Figure 1 The architecture and message structure of ePrescription.



represents a standardized message sent between the mod-ules) being sent to the Prescription Exchange, whichwould perform various controls before distributing themto other actors.The specifications made by The Directorate of Health

emphasized that the various actors were responsible for‘their’ modules, with a central database, the PrescriptionExchange, as the key. The project was organized in sub-projects reflecting each institution that was included inthe service, and five subprojects were established.In May 2008 the first pilot implementation was inaugu-

rated by the Minister of Health. It was carried out in avillage in the Eastern part of the country and included theGPs and the local pharmacy. It turned out to be a minordisaster, and after 4 months a crisis was declared. The mainreason for the problems was not the ePrescription solu-tion, but that the new version of the EPR system wasunstable. Somewhat unreasonably, the ePrescription pro-ject got the blame in an angry press. The main technicalsolution was tested and accepted during 2009 while wait-ing for the vendors to complete and test their newversions. A new pilot was set up in May 2010 – whichturned out to be successful – and contracts for large-scaleoperations were signed.At the same time other challenges surfaced. While

primary health care (the GP level, administrated by muni-cipalities) issued 70% of the prescription, the rest wasissued by hospitals. These are organized in four healthregions, as separate state own enterprises. In autumn 2009,it became clear that the IT managers in the health regionshad not prepared sufficiently for integrating hospital EPRs(which are different from the GPs) with the ePrescriptionsolution. Moreover, they raised comprehensive objectionsto the architecture of the solution. During some heatedmeetings in the winter 2009–2010 a kind of compromisewas reached: the health regions would follow their ownframework for integrating various old and new systems,while making an effort to implement a short-term solutionfor ePrescription.The solution for primary care was successfully imple-

mented in the whole country in stages, during 2011–2013.It was, in fact, the first successful national e-health infra-structure. However, it still did not include the secondaryhealth sector, because of integration problems and securityissues. The adoption of the ePrescription solution inhospitals is found to require a modified and less sophisti-cated security solution. However, unfortunately, this mod-ified security solution implied that significant changes alsohad to be made to the modules running in the otherinstitutions. When ePrescription will be deployed inhospitals is still uncertain, 9 years after the project wasinitiated.

Results of Strategy 2 The Elin and ePrescription projectsestablished a new standardization approach that we herecall ‘integrated solutions’. The main difference betweenthis and the ‘anticipatory standardization’ approach is the

strong focus on user requirements in the ‘integrated solu-tions’ approach. Both approaches, however, follow a top-down and specification-driven strategy.On the basis of the ‘integrated solutions’ approach a

broad range of standardized messages have been defined:fourteen in the Elin project and 31 in the ePrescriptionproject. These projects clearly demonstrate importantweaknesses in the original strategy with its strong focuson defining electronic versions of the existing paper docu-ments. The contrast between the focus on one singleEDIFACT message representing the prescription in theproject in the nineties and the 31 messages defined in thecurrent project should be a clear illustration of this. Theseprojects also revealed that most messages defined byfollowing the established approach did not satisfy theusers’ requirements either and had to be modified.However, this strategy has not been an unambiguous

success either. Some solutions have been implemented,but their diffusion and use is modest. For example, theePrescription project, has still, 9 years after the pre-projectstarted, a long way to go before the solution is fullyadopted in line with the aims of the project, in spite ofgenerous funding from the Norwegian government.While the focus on users’ work practices and needs

represented a step in the right direction, the top-downand specification-driven approach to systems develop-ment generates a considerable complexity, both technicalcomplexity of the envisioned solution, but also organiza-tional complexity in terms of the number of independentactors (organizations) involved that need to reach agree-ment on a broad range of issues and coordinate theiractivities. In addition, technical and organizational issuesare connected in complex ways that are hard to under-stand and manage.

Strategy 3: Flexible generificationCompared with the strategy established by the Elin pro-ject, the third strategy has even more focus on users’practices and needs, a stronger focus on developing work-ing solutions and a correspondingly lower focus on stan-dardization as such. This strategy is close to the onefollowed by Fürst when they so successfully introducedservices for electronic exchange of lab reports in Norway in1987. This strategy has to a large extent successfully beenfollowed by Fürst ever since and has also been adopted by afew other initiatives more recently. We will here present,in addition to Fürst, two other initiatives we have identi-fied that have followed the strategy we have named‘flexible generification’.

Fürst Fürst’s solution was based on a slightly differentarchitecture than the solutions described above, whichwere based on an architecture where the applications wereextended with functionality to send and receive messages.Fürst’s solutions, in contrast, consisted of a client modulerunning on the GPs’ computers interacting with thepatient record systems and a server module running on the



same computer as the lab system and interacting with this.Lab reports were transferred from the server to the clientbased on a simple format. Lab reports were then trans-ferred to the patient record system via a file and based onthe same format. A slightly modified version of this formatwas used in most of the lab report transmission solutionsdeveloped around 1990 triggered by Fürst’s: the ‘$’ symbolwas used as a separator between individual reports anddata elements within each report. For this reason, themodified format was called the ‘$-format’. (See Appendixfor the specification of this format.)When Fürst’s solution for lab report transfer was success-

fully adopted by the lab’s customers, Fürst wanted toextend the scope of electronic services offered. The naturalnext step was electronic transmission of lab test orders.Fürst realized from the very beginning that this would be adifferent case. Receiving reports electronically was attrac-tive for GPs because they got faster access to the results,and they save time they otherwise have to spend onmanually entering the report data into the EPR. However,in the case of orders, Fürst’s primary motivation forimplementing an electronic solution was the reduction ofmanpower required internally in their lab.Fürst started the development of a pilot solution in 1992

together with one of the vendors of EPR systems. Thesolution was tested in a pilot implementation in a GPoffice in 1993. The experience gained by the pilot users didnot create much enthusiasm. The overall usability of thesolution was rather poor, and the GPs did not see anyimmediate benefits. Reasonably, Fürst concluded that asuccessful solution would have to offer the GPs some newand improved services as well.After some time, Fürst came up with the idea of offering

the GPs the possibility of ordering new tests of a specimenafter the results of those ordered first were available.Usually a GP orders several tests of the same specimen.Often, the combination of tests that is most relevantcannot be decided until the results of some of the analysisare seen. Accordingly, it would be useful to order sometests, look at the results and then decide on the additionalanalyses that are relevant. When both orders and resultsare transmitted electronically, this possibility may becomereality. This idea was later called ‘interactive ordering’.Fürst also wanted to use relevant standards where such

existed. Unfortunately, this solution could not be imple-mented based on the existing standardized EDIFACTmessages. Nor could it be implemented based on emailprotocols that were the standard carriers for electronic datainterchange-based communication (Hanseth & Monteiro,1997). However, such a solution could easily be imple-mented on top of a transaction-oriented on-line connec-tion, which Fürst’s used in its original solution. The onlyproblem was that it used dial-up connections, which wereconsidered too slow for the kind of interactivity requiredby Fürst’s envisioned solution. Accordingly, they decidedto postpone this until higher bandwidth networks weremore easily available and affordable. Around 2000, Fürstconsidered this to be the case, and they started to plan

implementation of the solution. At this time, they alsoplanned to pay the costs for their customers related tobroadband connections. However, when they were aboutto start implementation, the idea of a National NorwegianHealthcare Network had emerged. Fürst decided to supportthis initiative as much as possible. In 2003, they becamethe first users of this network and started actively offeringtheir solution to their customers.The ordering solution was built by extending (modestly)

the lab report solution. The same file format was used forintegration between the client module and the GPs’patient record systems. The solution has been enhancedover the years and now offers the users a broad range ofservices beyond the possibility of filling in and sending anorder. For instance, the users get an overview over whichanalysis theymay order, guides for which analysis to order,help functions for every button to be pushed, informationabout each analysis, 10 years of history for each patient,information about which orders new analysis may beadded to and so on (Fürst, 2014). The solution alsoincluded services for communication between GPs andspecialists working at Fürst. Users get access to informationand functions related to interactive ordering through thecommunication solution’s client module. This means thata vast range of information is exchanged between theclient and server modules in addition to the lab ordersand reports based on a broad range of protocols andformats while only orders and reports are exchangedbetween the client module and the GPs patient recordsystems. Fürst has spent about 1,5 man-years in total everyyear on developing, maintaining and improving thesolution.The interactive ordering solution has increasingly got a

reputation for being a useful tool and the growth in numberof users has accelerated, by November 2010 to about 3500,which accounts for more than 50% of Fürst’s customers.At any time during ordinary work hours 1100–1200 usersare logged in. Fürst’s solution is integrated with five differentEPR systems for GPs from three different vendors.

North Norwegian Healthcare Network (NNHN) TheNNHNwas established as a project in 1997. The aim was toset up an ICT network for the exchange of information likeorders, reports, admission and discharge letters amonghealth-care institutions in the Northern Norwegian healthregion. They applied to the Health Directorate (and Min-istry of Health) for funding. The directorate (and ministry)approved the idea and decided to fund the project. Theyalso decided that the other four health regions should dothe same.The project was staffed with people who had worked at

the IT department at the University of Tromsø. They wereall familiar with Internet technology and the Internet wayof building networks and services, that is, developingsimple working solutions first, and then defining thestandards when the solutions are working and proveduseful. They had in general a very pragmatic approach to



standards and standardization and the focus was on estab-lishing services that were widely adopted as fast as possi-ble. (The other four regional projects started insteadextensive activities aiming at specifying needs, etc.) Theproject was reorganized into an independent company,owned by the three counties of the Northern health regionin 1999. In 2003, when the Norwegian Healthcare Net-work was established, NNHN was merged into thisorganization.The services were established in close collaboration with

GPs and the hospitals. In addition to message exchange, anumber of web-based services were also established.NNHN decided that an important factor for successfulestablishment and operation of the network was to keepthe complexity of the software running on the GPscomputer at a minimum level. Instead, as much as possibleof the software functions should run at the hospitals’computers. This would increase the stability of the soft-ware in the GPs’ computers, that is, decrease the prob-ability of errors and the need for updating the software byadding new functions, for instance when amessage formatis changed. So if a lab system produced reports in astandard format, the report was converted to the formatread by the receiving GP’s ERP system at the hospital, andthen transferred in this format. NNNH’s solution was,then, based on an architecture similar to Fürst’s. NNHNalso did most of the work required to integrate thecommunication technology with the applications, that is,lab, radiology systems and patient record systems inhospitals, and patient record systems in GP offices. Mostof the data transfer and the integration with applicationswere based on the $-format or modifications of this. The$-format was used consistent with its original specificationin the case of lab orders and reports. However, it wasslightly modified so that it could also be used for admis-sion and discharge letters and radiology orders and reports.This modified version of the $-format was then used as astandard for integration between NNHN’s communicationsystem and most patient record systems used by GPs andlab, radiology and patient record systems in hospitals inNorthern Norway. In a few cases other formats were used.Overall the NNHN effort was very successful. All hospi-

tals and GP offices in the region were connected to thenetwork, and almost all documents covered by the serviceswere exchanged electronically. This was in strong contrastto the slow uptake of similar services in the other regions.

WELL/DIPS Interactor: Interactive orders and referrals In2001, and related to the NNHN activities, the UniversityHospital of North Norway (UNN) studied the use ofresources and error rates related to their laboratory activ-ities. (This case is more extensive described in Johannessen& Ellingsen (2009).) This study revealed that the paperorders from primary care often contained errors, lackedclinical information or had a mismatch between the paperorder and the sample tube. In addition, manual and repe-titive work in receiving the samples was considered a waste

of resources. UNN believed electronic ordering would helpimprove the situation regarding both problems. They hadworked together with the local IT vendor WELL for sometime as a part of other NNHN activities and discussed howa service for electronic ordering could be designed. (WELLwas bought by DIPS, the market leader of ElectronicPatient Record systems for hospitals in Norway, in 2009.)Both were aware of Fürst’s solutions for interactive order-ing. WELL was interested in developing a similar solution.A difference was that it should be a generic interactiveordering product for the health-care market outsideNorthern Norway that could be used by many labs andalso support other kinds of ordering like referrals and radi-ology orders.In 2006 they established a joint project. The system for

electronic ordering of laboratory services was built in aniterative way and in close collaboration with the users atUNN and in general practice. Four months into theproject, WELL presented a simple but working solutionthat satisfied the minimum requirements. It enabled thesubmission of an electronic order and that was more or lessa digitized version of the existing paper-based requisition.This system was based on existing file formats developedby Fürst and other NNHN activities, and required little orno effort from the EPR vendor. However, based on the real-life use and testing of this solution the product wascontinuously changed and adapted to the needs of theusers. The result of this incremental process was theproduct named Interactor. In collaboration with AkershusUniversity Hospital and UNN, WELL extended the soft-ware to support interactive referrals. Interactor was in 2010used by 9 hospitals and approximately 60 GPs, while apilot for interactive referrals included six GPs. In 2013, acontract was signed for implementing Interactor for inter-active referrals between all hospitals and GPs within thegeographical area covered by the Western Norway Regio-nal Health Authority.The types of analyses a lab can perform are quite

dynamic and may change continuously. This implies thatthe parts of the system where the GPs make their ordersneed to be modified every time a lab makes changes intheir repertoire of analyses. In addition, each lab offers aunique repertoire of analysis. Accordingly, the systemssupporting electronic ordering need to know the repertoireof analysis offered by each single lab, and they need to beupdated every time a lab changes its repertoire. WELL andUNN, therefore, agreed that electronic ordering had to bebased on a model where the labs specify their repertoireelectronically in a software module on the health network.The updated repertoire is downloaded from this module tothe client that is used when orders are specified. Themodule for referrals is based on the same ideas and thesame technology for integration as for interactive labordering.The exchange of orders and referrals was based on

formal standards in the same way as NNHN. However,much of the information exchanged between GPs andhospitals had to be specified by WELL as no formal



standards existed, for instance, for the specifications ofcontent of the analysis repertoire, requests for the latestversions and so on. As for the Fürst and NNHN solutions,the interface between the Interactor modules and the GPand hospital systems was most crucial, and based on themodified $-format.

Results of Strategy 3 The activities presented in this sec-tion focused on developing useful and well-working solu-tions following an experimental and evolutionaryapproach. We see this approach as a combination of the‘flexible standards’ and ‘generification’ strategies pre-sented in the related research section. Accordingly, we callthis strategy ‘flexible generification’. This strategy sharesthe strong focus on user needs with the ‘integrated solu-tions’ strategy, but is fundamentally different in the sensethat ‘flexible generification’ is a bottom-up and evolu-tionary approach while the other is top-down and specifi-cation driven.This ‘flexible generification’ strategy has delivered a

wide range of successful solutions – solutions that offerednew and improved services. The evolutionary approachhas been important for developing simple solutions thatcould be developed for reasonable costs within reasonabletimeframes. This approach has also allowed early userfeedback based on the use of running systems, which hasbeen crucial for improving the systems to fit user needs.Moreover, this mere experimental approach has also gener-ated new ideas about how the technology can be designedto support new and improved health-care services and notjust speed up existing paper-based practices.These activities have also contributed to standardiza-

tion. All three cases show a pragmatic approach to stan-dards. They implemented the relevant standards that wereavailable and modified these when needed and definedtheir own messages and formats where there were nostandards. When defining their own formats, they did soas much as possible by using and modifying existing ones.A typical example of this is the role of the $-format. Thisformat was originally defined by Telenor, by slightlymodifying the format Fürst used in its solution from1987, when they developed the lab report transmissionsolution for UNN around 1990. This format was used inmost of the lab report transmission solutions developedthe following years and was then established as a sort ofnational de facto standard. NNHN then used a modestmodification of the $-format for transmission of mostpatient information exchanged between GPs and hospitalsin the Northern Norwegian health region and related tothe Interactor product for implementing an interactivereferral service in one additional health region in Norway.The interfaces between the communication services’ GPclients and the GPs’ EPR systems and other applicationshave evolved into de facto standards in a similar way.Moreover, the same happened to other message defini-tions, like the more formally approved EDIFACT stan-dards. This illustrates a bricolage-like process where

standards are selected and modified to fit users’ needs(Ciborra, 1992). Through such processes the standards areimproved and made more generic to serve the needs, thatis, standards have been developed through a generificationprocess. Further, this generification of the de facto stan-dards was quite simple just because the standards wereflexible standards. This flexibility is illustrated by the factthat the so-called $-format was specified on one single A4page. In contrast, the specification of the first EDIFACTmessage defined by CEN TC/251 for lab reports was on499 pages!The experimental approach taken by these efforts that

had the declared aim of developing new and innovativeservices also revealed that a much broader range of formatsand protocols were needed for establishing such newservices. In the solutions developed in the efforts pre-sented here, these formats and protocols have been usedfor communication between the client and server modulesof these solutions. These solutions have in all cases beendeveloped by one organization and each client or servermodule has only been communicating with its ‘twin’module. So far no third-party server or client module hasto our knowledge been developed and used to commu-nicate with any of the client or server modules developedin these efforts. However, when the interactive orderingand referral solutions diffused more widely and stabilized,it is reasonable to believe that the protocols and formatsused will also need to be standardized.The efforts following the strategy presented in this

section also differ from the two others in a more general,but indeed crucial way: How they cope with complexity.The solutions developed in all three efforts are more or lesscompletely designed and developed by one single organi-zation. Combined with the pragmatic approach to stan-dardization, the focus on simple and working solutions,and an evolutionary development strategy, has been lead-ing to technical solutions of lower complexity. However,even more important, the outcome has also been that theorganizational complexity has been much lower: fewerorganizations needed to be involved and agree on detailedtechnical specifications.

DiscussionWe have in this paper presented the history of develop-ment, implementation, diffusion and use of ICT standardsfor information exchange between health-care institutionsin Norway since this activity started in 1987 and up tilltoday (October 2013). We have identified three differentstrategies for developing standards. Our focus is on howeach of these strategies enables and supports solutions thatbest contribute to the overall improvement of the health-care sector through the development of new and improvedmedical services.The general picture in the field is that the implementa-

tion and diffusion of standards have been very slow. Thishas been the pattern in most national strategies for ICT inhealth care. However, we believe that in order to make the



future of ICT standards brighter than their past, we need acritical examination of the strategies followed and theachievements. We sum up our argument in Table 3.The first strategy identified, which we call anticipatory

standardization, is the official and traditional one. To ourknowledge, this strategy has not been seriously challengedofficially by any actor within the field. The two otherstrategies identified are not recognized as such within thefield; they are emergent strategies. By contrasting thesestrategies and their achievements, we believe importantlessons may be learned both by practitioners andresearchers.The first and official strategy, anticipatory standardiza-

tion, has delivered a number of standards specifications.However, the standards, with their rather extreme focusonly on replacing paper forms with similar digital informa-tion objects, turned out to be unattractive for applicationvendors, as well as user organizations. One importantreason for this was the fact that they could not beimplemented in useful systems without substantial mod-ifications and specifications of additional messages andprotocols. This limitation was to a significant degree over-come in the second, the integrated solutions strategy.With astronger focus on users’ working practices and needs, thisstrategy has delivered more appropriate and complete setof specifications. However, the implementation of thespecifications has been a very slow process, mainly becauseof the complexity of the solutions specified and theorganizational complexity of the coordinated implemen-tation process the specifications require. When the stan-dards were successfully adopted, the existing paper-basedservices are improved. However, the benefits are defini-tively limited as they still mimic the paper-based processes.Moreover, there is a high risk that the solutions based onthese complex standards soon will emerge as legacy sys-tems resisting change efforts and accordingly represent amajor problem when one tries to improve processes inways enabled by ICT solutions. That is, they may inhibitrather than enable or stimulate future service innovations.The modest results of these strategies are also admittedin two recent reports from central authorities. In a proposi-tion from the Ministry of Health to the Norwegian parlia-ment in 2012, the way the established e-health servicesreproduce existing practices is characterized as putting‘electricity on paper’ (MOH, 2012). In an audit of thestatus of electronic message exchange within health care,the Office of the Auditor General of Norway even con-cluded that after a more than 20 years long effort, manynecessary standards are still lacking, many of the standardsdefined are not implemented in solutions and many of theimplemented solutions are not adopted (Riksrevisjonen,2014).We see the third standardization strategy, flexible gener-

ification, as being by far the most successful one in terms ofdelivering working solutions that also enable the innova-tion of new services that go beyond existing paper-basedpractices. Moreover, the differences betwee

flexible generification: ictstandardization strategies and service … · actors, including...

Documents