1

Stabilization and Destabilization Processes at Work in Digital

Infrastructures: The Co-Functioning of Architecture and

Governance

Joan Rodon, ESADE Business School, Barcelona

Ole Hanseth, University of Oslo

Abstract

Digital infrastructures (DIs) are characterized by fluidity and openness to number and types

of users and limitless possibilities for re-combinations of digital artifacts, while at the same

time having closed and static structures that give them continuity. This paper studies this state

of tension of DIs by focusing on the interactions among the existing and new socio-technical

components as they connect and disconnect across time and space. In those interactions

components exercise their capacities to affect and be affected by other components and this

leads to tensions. This paper captures this permanent state of tension of DIs with the notion of

stabilization. Against this backdrop, socio-technical components of DIs become involved in

processes that extend the existence of the DI through time (stabilization processes) and

processes that allow the DI to change and exist outside the current scales (destabilization

processes). To illustrate this notion, we apply to a longitudinal case study of a DI sponsored

by a regional healthcare service administration of Spain. We demonstrate how the dynamics

of this DI are characterized by cascades of processes of destabilization followed by re-

stabilization and vice versa. Further, our account reveals that DI evolution occurs at different

levels of scale and that stabilization processes at one scale trigger destabilization processes at

other scales. The paper adds to the literature on DI by providing a more integrated model

seeing DI evolution as the interaction between processes of stabilization and destabilization

2

which are conditioned by the co-functioning of the architecture and governance structure of

the DI.

Keywords: digital infrastructure, stabilization, evolution, architecture, governance

1. Introduction

In many domains we observe a proliferation of digital infrastructures (DIs) that ubiquitously

become the basis for the operation of individual organizations and entire sectors, and the

emergence of digital innovations (Yoo et al. 2010). Such DIs are complex socio-technical

systems that grow over long time scales by integrating and extending existing installed bases

(Grisot et al. 2014; Hanseth and Lyytinen 2010). DIs are not built from scratch but grow

“conservatively through mutation and hybridization, rather than outright break with the past”

(Blanchette 2012; p. 33). They are “always an unfinished work in progress” (Edwards et al.

2009, p.365). Against this backdrop, prior studies emphasize that a challenge for the

management of DIs is how to make them evolvable (Gawer 2014).

DIs change over time in ways not captured by existing models of the (long term) evolution of

ICT solutions (Reimers et al. 2014). Accordingly, Reimers et al. (2014) suggest that a

promising route to study evolution consists of conceptualizing it through tensions between

components that destabilize the DI. In particular, a first stream of research has empirically

illustrated that the management of DIs involves recurrently dealing with tensions that result

from the complex interactions between socio-technical components (Tilson et al. 2010).

Another stream of literature studies how configurations of two core components of DIs

(namely, architecture and governance structure) condition the evolution of DIs (Henfridsson

and Bygstad 2013). In this paper we will follow Reimers et al.’s (2014) request for research

into new models capturing the evolution of DIs by integrating both streams of research. The

two research questions addressed in this paper, therefore, are: 1) how do the recurrent

3

tensions of DIs determine their evolution?; and 2) how do the interactions between the

architecture and governance condition those tensions?

To address these two research questions we conducted a longitudinal, in-depth qualitative

case study of a DI sponsored by a regional healthcare service administration of Spain

covering a 14-year period from 2000 to 2013. This paper adopts an ontological approach that

describes the DI as in-process, becoming, and ever-changing. Our findings conceive the

evolution of DIs as a set of interacting processes of destabilization –that disrupt existing

relationships among components of the DI and/or create new ones– and re-stabilization –that

bring more homogeneity and persistence to the DI. Those processes in turn are shaped by the

interactions between, or co-functioning of the DIs’ architectures and governance structures.

The remainder of the paper is structured as follows. First we review existing research on the

evolution of DIs followed by our analytical framework: Assemblage Theory. In section four

we are introducing our research setting and overall research approach. The paper then

presents the narrative and analysis for the case study. The paper concludes with a discussion

of the findings and reflection on the contributions of our paper.

2. Research on the evolution of DIs

Literature on the evolution of DIs has focused on two main research problems. First, there is

a stream of research that has shown that evolution is conditioned by two constitutive

components of DIs, namely, architecture and governance (Ciborra 2000; Hanseth and

Bygstad 2014; Henfridsson and Bygstad 2013; Tiwana et al. 2010). Tiwana et al. (2010)

advocate for thinking in terms of co-design and co-evolution of DI architecture and

governance. The alignment between DI architecture (in terms of decomposition, modularity,

and design rules) and its governance structure (in terms of decision rights, control

mechanisms, and ownership) shapes the evolvability of DIs at diverse timescales (Tiwana et

4

al. 2010). While governance is understood as the lever to reduce behavioral complexity (i.e.,

the difficulty to predict the aggregate behavior of the DI’s components), architecture is a

lever that tackles structural complexity (i.e., the complexity derived from the multiple and

unclear connections between the components). Henfridsson and Bygstad (2013) review 41

cases of DIs and identify three self-reinforcing mechanisms (innovation, adoption, and

scaling) that explain evolutionary outcomes. They find that the decentralized control and a

loosely coupled architecture triggered all three mechanisms, while the adoption and scaling

mechanisms were also triggered in cases based on a centralized control and loosely coupled

architecture.

Research on the complexity of DIs has also noted that the traditional approaches for

controlling and managing the development and implementation activities, which are based on

hierarchical organizing and decision rights with project management controlling the whole

activity on the top, and a tightly coupled architecture have proved inadequate due to the

number and heterogeneity of actors (Ciborra 2000; Constantinides and Barrett 2014; Hanseth

and Ciborra 2007). van Schewick (2010) and Hanseth and Lyytinen (2010) show how the

combination of a layering and end-to-end architecture and a governance structure based on a

loosely organized network of actors have contributed to the successful evolution of the

Internet. Finally, in their research about ten DI initiatives in the Norwegian health care sector,

Hanseth and Bygstad (2014) find that an approach that combined a centralized architecture

and a centralized control structure was most successful for establishing new and innovative

DIs, while a decentralized architecture and decentralized control were crucial for further

development of innovations on the DI after it was successfully established.

A second stream of literature has revealed that DIs are characterized by tensions that need to

be balanced (Edwards et al. 2007; Ribes and Finholt 2009). For instance, balancing the

retention of important elements over extended periods of time with the transformation of DI

5

with respect to other important elements (Baldwin and Woodard 2009; Henfridsson and

Bygstad 2013; Reimers et al. 2014); the stability brought by the installed base to enroll new

actors and services with the flexibility to leverage unbounded growth of actors and services

(Hanseth et al. 1996; Tilson et al. 2010; Tiwana 2014; Wareham et al. 2014); the autonomy

of independent actors to seek generativity through distributed control with the centralized

control that enables that generativity to be played out in their own interest (Nielsen and

Aanestad 2006; Tilson et al. 2010; Tiwana et al. 2010; Wareham et al. 2014); the logic of

generative and democratic innovations and the logic of infrastructural control (Eaton et al.

2015); the collective identification of actors to reduce undesirable variance toward

contributions to the social good of the ecosystem with individual identifications that increase

desirable variance to encourage explorative and entrepreneurial responses (Wareham et al.

2014); the top-down demands for integration with the persistent, bottom-up reliance on the

installed base of systems and practices (Hepso et al. 2009); and the sensitiveness to local

contexts with the need to standardize across contexts (Rolland and Monteiro 2002; Silsand

and Ellingsen 2014). These studies usually conceptualize these tensions as a duality; that is,

these tensions are seen as interdependent, complementary, mutually enabling, and constituent

of one another (Tilson et al. 2010; Wareham et al. 2014). For instance, flexibility and

variability are achieved by means of the stability granted by standards as the latter enable

novel recombinations of digital components of DIs. Likewise, “stability can be bolstered only

by allowing flexibility…. [variation] at the edge and across layers bolsters the stability of

infrastructures” (Tilson et al. 2010, p.754).

Tilson et al. (2010) capture this duality with a conceptual model which depicts evolution as

dependent “on the definition and placement of control points… as well as on the ways they

are challenged by the dynamics of generativity” (Tilson et al. 2010, p. 755). Later, Tilson et

al. (2012) provide a validation support for this model with the cases of Apple’s iOS and

6

Google’s Android. Reimers et al. (2014) study the balance of persistence and transformation

of DIs by focusing on changes occurring at the level of individual practices and of

constellation of practices. Reimers et al. (2014) conduct a case study of the evolution of

electronic ordering systems in the Australian pharmaceutical distribution industry over a

period of 30 years, and show how those systems gradually moved from closed and

proprietary, to quasi-open systems as a result of the appearance of new practices and the

mutual adjustment of practices. Overall, both models conceive DIs as being open and fluid

due to the tensions among the socio-technical components of DIs, while at the same time

having closed and static structures that give them continuity. However, we still have a gap in

our understanding of the relationship between the tensions to which DIs are subjected and

that shape their evolution. This is the void this paper fills. Before doing so, we draw upon

assemblage theory, as developed by DeLanda (2006), to make two related ontological

considerations of DIs which prior literature has not made explicit.

3. Assemblage Theory

The first consideration regards the nature of relations among components of DIs. DIs cannot

be solely defined by relations of interiority, meaning that a DI is a cohesive whole premised

on the aggregation of multiple socio-technical components each filling a specific purpose

within the DI and not having existence outside the DI. In fact, an analytical focus on the

relations of interiority limits the possibility of accounting for the dynamics of DIs in relation

to their outside environment and for the roles and functions that components can play outside

the DI to which they belong. The necessary relations that form a DI are partly constituted by

the existing practices and actors, technological capabilities, architecture, and so forth, as well

as by the relations that these components have with other entities that are exterior to or in the

surroundings of the DI. Accordingly, DIs are not seamless totalities but wholes whose

components are characterized by their relations of exteriority (DeLanda 2006), meaning that

7

the boundaries of the DI are open as new socio-technical components connect and disconnect.

That is, the relations between components are only contingently obligatory. A component

develops a function based on the relations with other components but it has existence outside

those relations. For instance, a mobile DI such as iOS or Android compromises multiple

heterogeneous socio-technical parts –e.g. telecom carriers, OEM, content providers, APIs,

SDKs, app developers, apps, legal contracts, app users, and so forth– that might participate in

other DIs –e.g. OEMs might manufacture devices for diverse DIs, app developers might build

an app for several DIs. The autonomous parts together form more or less permanent DIs. By

looking at a DI as involving relations of exteriority allows us to foreground ongoing

processes of composition and decomposition. Moreover, this focus on the relations of

exteriority shifts the attention from the inert properties of component parts to their capacities

to interact with (or to affect and be affected by) other components. And this leads us to the

second ontological consideration.

Any component of a DI is equipped with properties and capacities. While the properties are

intrinsic to the component, the capacities are relational and exercised in the interactions

between components (DeLanda 2006). Properties are always there, but capacities need

something –a catalyst– to be triggered. For instance, the fact that a digital product is modular

(a property) does not mean that it will become generative (a capacity) per se. Generativity,

which refers to an “overall capacity to produce unprompted change driven by large, varied,

and uncoordinated audiences” (Zittrain 2006, p.198), is dependent for instance on how the

modular product interacts or is combined with other digital products, which are sometimes

layered upon one another, and how those digital products are contextually appropriated and

used (Yoo 2013; Yoo et al. 2012). To illustrate, Google Maps (which is modular) becomes

generative as Nikon’s myPicturetown app integrates with it so videos and photos can be

mapped to the locations where they are shot. That is to say, the generative capacity of a

8

digital product does not simply depend on its modularity (a property) but also on the

interactions of multiple components that exercise their capacities and which may be either

internal or external to the digital product. Allowing for possibility of non-linear complex

interactions between components is relevant when studying the emergence and evolution of

DIs (Hanseth and Lyytinen 2010), but if the components are fused into wholes with only

relations of interiority, that possibility disappears. Hence, the fact that the multiple socio-

technical components of DIs exercise their capacities as they connect and disconnect across

space and time makes DIs be in a state of permanent tension.

In order to study the relation between tensions and DI evolution, this paper uses the notion of

stabilization, which we adapt from DeLanda (2006). By stabilization we broadly refer to the

processes that increase the internal homogeneity of the DI giving it an identifiable boundary

and produce more or less permanent articulations between the socio-technical components of

DIs that ultimately extend their existence through time. This occurs e.g. through the

standardization of the practices, meanings, and roles of actors, the establishment of

architectural control points that centralize the control structure, or the sorting processes that

exclude certain roles. For instance, the tight centralized control that Apple has on the iOS

ecosystem (Tilson et al. 2012) is an example of a highly stabilized DI. Such a centralized

control standardizes and excludes some uses and business models for app developers.

Alongside stabilization, any shift in the relations among components of a DI can also create

tensions that trigger destabilization processes. Destabilization processes work in the opposite

direction disrupting the order of the DI by means of increasing a DI’s heterogeneity and

fragmentation, promoting the geographical dispersion of components, and dissolving the

borders between components of the DI. For instance, the sponsor of a DI might decide to

open it by creating an API or an SDK that encourages 3rd-party developers to innovate in new

applications and services that ultimately blur the boundaries of the sector. Social networking

9

technologies like twitter, Facebook or Whatsapp are another example of destabilizing forces

as they blur the spatial boundaries of social interaction by eliminating the need for physical

co-presence and increasing geographical dispersion. Destabilization processes, which usually

remain relative meaning that they retain the possibility for re-stabilization, allow DIs to

change and exist outside the scales at which they were formed.

Of course any component of a DI can participate in both processes of stabilization and

destabilization by exercising different capacities and inducing changes in the relations

between components (DeLanda 2006). For instance, incremental and additive changes to an

API of a DI are made to favor the extension and continuity of the installed base (stabilization)

while at the same time attract new components –e.g., new apps, new users– that destabilize

the DI.

In short, we suggest that DIs are always in tension responding to continuous destabilization

processes counter-balanced by re-stabilization processes. Destabilization at a certain scale of

a DI tends to be followed by re-stabilization at the same or another spatiotemporal scale and

vice versa. In that respect, the processes of destabilization and re-stabilization allow us to

view tensions – e.g., continuity and change, control and autonomy–, which characterize

evolution, as related.

4. Method

In order to illustrate how DIs recurrently go through processes of stabilization and

destabilization, and how those processes are conditioned by the co-functioning between

architecture and governance, we conduct a longitudinal, in-depth case study (Yin 2009) about

an electronic prescription digital infrastructure (EPDI) for the public health service in the

autonomous region of Catalonia, Spain. The empirical case covers 14 years (from 2000-

10

2013). Before we present the details of the data collection and analysis, we provide some

background of the Spanish model of pharmacies of which the EPDI became a component.

4.1. The Spanish model of pharmacies and the infrastructure in-place

The model of pharmacies in Spain compromises multiple components operating at different

scales. At the lower scale, there is the pharmacist; a health agent who exercises its

professional practice in community pharmacies or hospital pharmacies by dispensing of

drugs, production of patient-specific preparations, and other pharmaceutical care tasks (e.g.

health promotion, tracking patients’ medication record, checking drug interactions, etc.). In

order to practice pharmacists must be registered in the College of Pharmacy of the province

where they practice.

Community pharmacies are private healthcare establishments of public interest. Pharmacies

are the only health establishments authorized to dispense prescription-only medicines and

over-the-counter medicines to the general public. Medicines in Spain are publicly funded.

Until 2012 medicines were provided to pensioners for free; working age people paid 40% and

those suffering from chronic illnesses paid 10% of the cost of medicines. Pharmacies are

privately owned, and only pharmacists are allowed to own a community pharmacy. One

pharmacist or a group of pharmacists can own only one pharmacy. Pharmacy chains are not

allowed forms of ownership. The establishment of pharmacies is regulated responding to

demographic and geographic criteria in order to guarantee a homogeneous access of the

services to citizens. Regulations are defined at the national and the autonomous community

levels. While the central government is in charge of the general coordination of

pharmaceutical care and of matter related to pharmaceuticals such as registration, each

autonomous community organizes the planning of the pharmacy system.

In the autonomous community of Catalonia, the main actors that constitute the field are: the

Catalan Health Service (CHS), the Catalan Council of Pharmacists (CCP), the four Colleges

11

of Pharmacists (which coalesce into the CCP), the community pharmacies, pharmacists, and

trade associations of pharmacies. The CHS a public body that guarantees the right of health

care protection for all citizens in Catalonia. The CCP is a corporate and public legal entity

that represents the interests of all pharmacists in Catalonia, as well as the interests of

community pharmacy owners and ensures that regulations are respected.

An important stabilizing component of the model of pharmacies is the agreement between the

CHS and the CCP for the provision of services. The agreement establishes and regulates a

relational framework between the CHS and the CCP with regard to the conditions by which

pharmacists provide pharmaceutical care, invoice according to the contract economic

regulations, temporary fund the dispensed drugs and health products, continuously deliver

health care information to the CHS, do health promotion and disease prevention, and perform

pharmaceutical surveillance and security alert management of drugs and health products to

the population served by the CHS.

A core practice of pharmacists is the dispensing of drugs which interacts with other practices

(e.g. prescribing, invoicing) and actors (e.g. physicians, patients, CCP, CHS) and involves

flows of information, patients, money, and so on. The functioning of the prescribing,

dispensing and invoicing before the digitalization was as follows (see Figure 1). Once the

physician had decided the drug treatment for a patient the latter was given a paper

prescription. Physicians used clinical workstations to generate the prescriptions and printed

them. The patient took the prescription and her individual medical card to the community

pharmacy, where the drug was dispensed. Then pharmacists stored and signed those paper

based prescriptions. Pharmacists used a pharmacy management system (PMS) for tasks such

as the management of sales, inventory, or purchasing orders. Periodically, pharmacies

grouped the paper-based prescriptions they had dispensed in a given period of time and sent

them to the CCP. The CCP then checked all those prescriptions, scanned them, forwarded the

12

scanned and paper prescriptions to the CHS, and handled the invoicing for pharmacies. In

particular, the CCP submitted a single invoice to the CHS. So, the CCP, not pharmacists, was

the one in charge of invoicing the CHS. The CHS reimbursed that invoice to the CCP who

checked for errors and finally paid pharmacies according to the signed prescriptions they had

previously sent.

In this scenario, the information about patients’ drug use was fragmented in the diverse

systems of the multiple health providers and pharmacies either electronically or in paper.

That meant that the CHS had only a retrospective and partial view of patients’ treatments. It

could find out about those issues only when pharmacies invoiced but not when physicians

prescribed or pharmacists dispensed. The EPDI would target this fragmentation and

heterogeneity of systems.

Figure 1: Flows involved in the paper-based prescribing, dispensing and invoicing.

4.2. Data Collection and Analysis

Our empirical data is based on a 5-year data collection effort (mid-2008 to end-2013)

covering the period 2000-2013. Thus we focused on retrospective and real time data events.

We gathered these data from a variety of sources, including interviews, archival records, and

ethnographic observations to triangulate the data obtained (Yin, 2009). Between 2008 and

2013 we conducted 20 in-depth semi-structured interviews with key informants involved in

13

the EPDI from its genesis in 2000 until 2013. Those interviews were conducted in three main

periods: May – August 2008, January – May 2010, and February – May 2013. We identified

interviewees by applying the snowball sampling technique (Miles and Huberman 1994); that

is, identifying subjects for inclusion in our sample by referral from other subjects. All the

interviews were recorded and immediately transcribed after they concluded.

In the first period (May – August 2008), by then the EPDI was being piloted, we started

interviewing the project leader (a CHS executive), and then we moved to representatives of

two of the main user representatives of the EPDI: healthcare providers and pharmacies. At

that time, two healthcare providers participated in the pilot and we interviewed subjects from

both providers that were actively involved in the project: IT staff and the CIOs, primary care

physicians, primary care pharmacy coordinators of the healthcare providers, and community

pharmacists. Likewise, on the side of community pharmacies we interviewed members of the

CCP who were involved in the project, and the pharmacists that were part of the pilot. We

also interviewed external consultants that had participated in the project from its early stages.

In this first period questions were more open, and sought to understand interviewees’

opinions concerning the project, the role they played, the decisions they had made concerning

the EPDI’s design and the project’s organization, the events they thought were most critical

since it started, and the expected outcomes.

In the second period (January – May 2010), by then the EPDI was in the last phase of its roll-

out, we put the focus on the side of the pharmacists and the CHS. We conducted interviews

with the current vice-president of the CCP and the former one, who had been in charge of the

project, pharmacists that were using EPDI, and the CHS executive who was the project

leader. Three of the informants had already been interviewed in the first period. In these

interviews we sought to check the informants’ opinions during the first period of data

collection (3 interviewees were the same), evaluate the design, pilot and roll-out decisions

14

and the current outcomes of the project, and understand how the EPDI was influencing the

work and role of pharmacists, the CCP, and the CHS.

Roll-out was officially completed on June 2010. Three years later (February – May 2013) we

started the third round of interviews. We interviewed the general IT coordinator of the

Department of Health, IT manager of the CCP, pharmacists, and a provider of pharmacy

management system (PMS). All the informants had been in their positions since 2007 when

the EPDI was piloted. In these interviewees we sought to understand the changes and

adaptations to the EPDI, and its impact on pharmacists, the CCP, the CHS and PMS vendors.

Besides interviews we attended half-day workshops about the EPDI that were organized by

the CHS on 2008, 2009, and 2010. All the agents involved in its design, implementation and

use participated in those workshops. During these workshops we had multiple informal

conversations with 16 participants. Those conversations enabled us to collect more data and

check our understanding and assumptions about the EPDI. Another very relevant source of

data was archival records: presentations, workshops recorded in video, mailing lists, meeting

minutes, internal reports, and press articles. Overall we gathered more than 500 archival

records talking about the EPDI in Catalonia from 2000 to 2013. The collection and analysis

of these archival records started in May 2008 and lasted until 2013. Those archival records

became the main source of evidences while in other case cases were used as a complementary

to the evidences obtained in the interviews and the workshops. Finally, from 2010 one of the

authors periodically did on-site observations on how pharmacists used the EPDI, the changes

that they made on their practices, and their perceptions on the EPDI.

This corpus served as the basis for our data analysis in this paper. The data analysis

proceeded in four phases. In the first phase, we started our data analysis by constructing an

initial timeline of key events from 2000 to 2013 (see Appendix 1). We then created a thick

descriptive narrative of the case (Langley 1999). That narrative presented the formation and

15

evolution of EPDI and the flow of activity particularly from the perspective of two of the core

actors in project: the CHS, and the CCP. In the second phase, we identified four major

tensions that were salient during the genesis, design, roll-out, operation and evolution of the

EPDI. Next, we reanalyzed that narrative around those four tensions and coded them aiming

to identify destabilization and re-stabilization processes as well as the components involved

in those processes. In the fourth phase, we examined how those components were related

with the EPDI’s architecture and governance regime, and in turn, how they conditioned the

destabilization and re-stabilization processes.

In the following section, we present three tensions and their associated destabilization and re-

stabilization processes.

5. Findings

We will here present our findings organized along three tensions: between centralized control

and local autonomy, between stability and change of pharmacies’ practices, and between the

evolution of the EPDI and the flow of financial resources. Each of these tensions can be seen

as interactions between different assemblages: between central government and the pharmacy

assemblage, between the digital technology and pharmacies’ practices, and between the

overall EPDI assemblage and external ones. Further, each of the tensions was dominant in

different phases of the evolution of the EPDI; accordingly the presentation of these tensions

also represents to a large extent the evolution of the EPDI throughout its history.

16

Tension 1: Balancing centralized control and autonomy of actors at national and

regional scales

In 1999, the Spanish Ministry of Science and Technology started a project called PISTA1 that

aimed to promote the use of telecommunication networks to improve the delivery of services

to citizens, and to leverage the potential of data analysis. One of the initiatives of the PISTA

project aimed at establishing a national Spanish electronic prescribing solution. This initiative

would target the fragmentation and heterogeneity of prescribing and dispensing systems that

were in-place. Several stakeholders were invited to the project: governments of the

autonomous regions2, the representatives of the diverse professionals involved in prescribing

and dispensing –i.e., the Colleges of Physicians and the Spanish Council of Pharmacists–,

and patient associations. On April 2001 a first draft of a requirements analysis was delivered.

Following a top-down approach, the PISTA project delivered in 2002 a conceptual model for

a common Spanish electronic prescription DI and a roadmap for its implementation. It was a

one-size-fits-all model based on a centralized architecture that the Spanish Council of

Pharmacists objected to from the very beginning. They argued that the main goal of the

central government was simply to control the pharmacists’ practice and to reduce public

expenditure on drugs, rather than the use of IT to develop the pharmacists’ professional

practice and improve the quality of their services (Cordobés, 2002). For instance, pharmacists

complained that the government’s model did not allow them to do pharmaceutical care

because pharmacists did not have access to the database of pharmacological treatments, and

the central database only stored those drugs that were funded by the National Health Service.

Further, the model did not account for the interaction between physicians and pharmacists, it

1 PISTA stands for Promoción e Identificación de Servicios Emergentes de Comunicaciones Avanzadas (Promotion and Identification of Advanced Communication Services) 2 The governments of the autonomous regions involved were Catalonia, Madrid, Basque Country, Canary Islands, and Asturias. Other regions such as Andalusia and Valencia were not involved but they were already developing their own electronic prescription infrastructures called Siglo XXI and Gaia respectively.

17

did not include the invoicing process, and the pharmacist was assigned the role of a drug

dispenser only. PIST arrived, then, at a design of a national solution. I.e. the design was

stabilized even though there was no strong consensus.

Finally, because Spain has a decentralized health care system in which the autonomous

regions are in charge of organizing the planning of the pharmacy system, the diverse

autonomous communities started their own electronic prescription projects from 2004. From

the outset, these regional projects adopted the design guidelines proposed in the PISTA

project. In the case of Catalonia, by mid-2004 the Catalan Health Service (CHS) set the

foundations for building an electronic prescription digital infrastructure (EPDI) that sought to

improve the efficiency of the Catalan health system by streamlining patients’ access,

containing drug expenditures, and reducing prescription and dispensation errors due to lack

of coordination between the agents involved. To achieve those goals the CHS needed a real-

time, holistic view of patients’ treatments. This involved doing changes to existing practices.

For instance, physicians would not make individual prescriptions anymore but medication

plans that would last up to one year; that in turn, would eliminate the need for co-presence of

patients and physicians in the prescribing process and would reduce the number of patient

appointments with primary care. Patients would pick up medicines at any pharmacy

according to a concrete temporal window thus avoiding that patients accumulated more drugs

than necessary. Pharmacists would remotely access the content of prescriptions thus avoiding

misinterpretation of prescriptions. The CHS would have the information about acts of

prescribing and dispensing in real-time and would have the capacity to influence both acts for

instance by forcing the prescription of generics.

From the outset of the project, the CHS as sponsor of the project was at the center of its

governance structure, and it initially set two central requirements for the EPDI. First, all the

data —i.e., prescriptions, dispensations, invoices, patients, drugs, health providers,

18

physicians, pharmacies, pharmacists— should be integrated and accessible online by the

diverse stakeholders —physicians, pharmacists and the CHS. Second, the processes of

prescribing and dispensing should run in real time; that is, any drug could be dispensed at any

community pharmacy in Catalonia immediately after it was prescribed, regardless of the

location of the prescribing physician. To fulfill these requirements, the CHS proposed, in line

with the PISTA guideleines, a CHS-centered architecture consisting of a central system

owned and managed by the CHS (called SIRE) that contained an integrated database with all

the data (see Figure 2). On the one side, the health providers3 would have to interconnect

their systems with the SIRE. On the other side, pharmacists were expected to connect directly

to CHS’ system, all using the same application for the dispensing and invoicing processes.

Figure 2: CHS-centered architecture for EPDI.

This CHS-centered architecture outlined a scenario in which the EPDI was portrayed as a tool

to control the administrative processes of pharmacists. Second, while on the side of

physicians the CHS-centered architecture preserved the interaction model between health

providers and the CHS by allowing health providers to keep their internal systems and

practices and only forcing them to create a communications module with the central server of

the CHS, it disrupted the existing order of the field of pharmacy as it bypassed the traditional 3 Health providers are those organizations that provide health services to the people insured by CHS. These services are hired by CHS and managed through contracts. There are two main kinds of health services: primary healthcare services and specialized healthcare services. At the time of the project, there were 36 health providers.

19

central position of the CCP in the invoicing process (Figure 1). Pharmacists perceived the

bypassing of the CCP as a weakening of their role and position in front of the CHS. It opened

the door for the CHS to set bilateral agreements with pharmacists in the future. Moreover,

that design made the deregulation of the field of pharmacy easier as the CHS would have

more control. Deregulation would be a key destabilizing force that would threaten the role

and identity of pharmacists as it would open the door to new entrants (e.g., pharmacy chains)

that would not be so much under the influence of the CCP –i.e., the CCP would be detached

from the day-to-day practices of pharmacists. Finally, the fact that the architecture excluded

the CCP also affected its revenue model. Besides the membership fees from pharmacists, an

important source of the revenue for CCP had been the invoicing process as pharmacies paid

for the processing of the paper-based prescriptions.

Based on these weaknesses of the architecture, the CCP proposed changes, turning the CHS-

centered architecture into a dual one (see Figure 3) consisting of a private network that would

interconnect all the pharmacies plus a central server, called SIFARE, that replicated the data

of the CHS’s SIFARE server that pharmacists needed —i.e., prescriptions, dispensations and

data catalogues. The CCP would own the private network and SIFARE. Pharmacies would

not have a direct access to SIRE but instead to SIFARE through the private network and the

SIFARE would synchronize in real time with the SIRE. This dual architecture mirrored the

existing structure of the model of pharmacies, and was also aligned with the discourse of the

CCP that linked the architecture with the professional development of pharmacists that relied

on shared services and strategies. The CHS reaction to this alternative architecture was

initially very negative. They were seriously afraid that this architecture represented serious

risks regarding the achievement of their two main objectives. However, after some

negotiations between the CHS and CCP, the dual architecture was approved and the

governance structure for the project was redefined to involve two main committees: a

20

steering committee, and an executive committee later called follow-up committee, in which

diverse members of the CHS, CCP, health providers and other stakeholders were present4.

Moreover, on September 2005 the CHS and the CCP signed an amendment to the

pharmaceutical agreement which established the clauses for the development of the pilot for

the EPDI, and made explicit the role of the CCP (ANNEX, 2005). That is, the amendment to

the pharmaceutical agreement reified and consolidated the architecture and roles of actors

previously defined. At this point in time, the idea of a National DI resulting from the

interoperability of diverse regional DIs that followed the design guidelines defined in the

PISTA project was fading away.

Figure 3: Dual architecture for EPDI.

To summarize – the CCP challenged and managed to destabilize the design proposed by CHS

which was based on the guidelines from the PISTA project. A main reason for this was that a

solution based on this design had capacities to interact with the pharmaceutical sector in and

could destabilize existing practices of the CCP and pharmacies and the relations between the

CHS and the pharmacies in ways the CCP found unacceptable. The CCP and the CHS

managed to reach agreement about a new architecture and governance structure. This dual

architecture and governance structure have stayed stable, at least up to now.

4 This organizing structure was still running in 2013. The steering committee meets every quarter, and the follow-up committee meets monthly. Likewise, working groups are created when new domains of study are required (e.g. prescribing and dispensing by active ingredient, prescribing and dispensing of narcotics, professional filters, communication to population, analysis of legal requirements).

21

Tension 2: Balancing stability and change of pharmacists’ practices

On the side of pharmacists, the CCP decided that SIFARE should be as transparent as

possible for community pharmacies so that they would not be forced to discontinue the use of

the existing pharmacy management systems or use an additional system for dispensing (as

would have happened with the initial design proposed by the CHS, see Figure 2).

Accordingly, in 2005, the CCP created an advisory committee for technology and

communications which brought together the CCP and the pharmacy management systems

(PMS) vendors. The members of the advisory committee met every quarter to discuss about

the EPDI’s status and agree on new requirements and services and the pace for implementing

them. Under this governance structure the CCP revamped a recognition program for PMS

vendors. The program defined a minimum set of functional and technical requirements that

PMS should fulfill, and homogenized the behavior of PMS vendors and pharmacists in terms

of access to and use of EPDI. Moreover, PMS vendors should be able to integrate their

solutions with SIFARE in a way that minimized the changes to the practices of pharmacists.

The CCP developed a set of web services for SIFARE and an API (see Figure 4). Those

vendors who passed the recognition program got an API from the CCP to interconnect their

PMS solutions with SIFARE5. The API enabled the dispensing and invoicing practices to run

partially at the pharmacies and at the CCP. In particular, the API kept some degree of

freedom for pharmacists about how they should perform their work (destabilizing capacity)

while at the same time all the pharmacies sent the same information to the CCP’s server for

checking purposes (stabilizing capacity).

5 Out of the more than 35 PMS vendors that operated in 2004, only 9 got the recognition and remained in the market. Five PMS vendors got the recognition in 2005, one in 2007 and three in 2008.

22

Figure 4: EPDI’s architecture finally implemented.

These architectural components created new dependencies among the actors – CHS, the CCP,

PMS vendors and pharmacies. The CHS developed and maintained a set of SIRE web

services to be used by the CCP for dispensing and invoicing. The CCP created the SIFARE

web services and an API for PMS vendors (see Appendix 2 for more details on the relation

between SIFARE Web services and the API). With that API, PMS vendors had to adapt their

applications to prescribe electronically, and install and configure the new version of the PMS

in pharmacies.

Other components that contributed to the stabilization of the EPDI were the security model

and the private network for pharmacists. The EPDI’s security model was defined by the

Catalan Certification Agency (CATCert).6 In this model, the CCP acted as a Registration

Authority ensuring that any digital certificate would be bound to the pharmacist to whom it

would be assigned in a way that assured non-repudiation. That is, this security model

conferred the CHS and CCP more control over pharmacists’ practices thus reducing

undesirable actions from the latter. Regarding the communications, pharmacies would be

connected through a virtual private network (VPN). The CCP signed an agreement with a 6 The CATCert is a governmental agency that was set up in 2002 in order to implement and roll-out the digital signature in all the Catalan governmental institutions and provide services to those organizations ensuring that the electronic transactions fulfill the legal guarantees.

23

telecom provider. That agreement homogenized the service and price conditions for all the

community pharmacies, regardless of their location or size. Each pharmacy would have an

asymmetric digital subscriber line and a backup integrated services digital network line to

connect to the central server of the CCP—the SIFARE7. On May 2008 the pilot was

satisfactorily completed. Overall, the pilot had involved 70 physicians, 56 pharmacies and

17,000 patients, and 314,500 prescriptions had been dispensed.

The EPDI continuously went through destabilization and re-stabilization processes

throughout its roll-out. Those processes were triggered by accelerations and expansions, and

decelerations and compressions of flows of actors (e.g., new physicians and pharmacists

using the EPDI, PMS vendors passing the recognition program, patients migrated into the

EPDI); information services (e.g., medication plans, dispensations, invoices, digital signature,

web apps, CHS-independent services); technological resources (e.g., processing capacity of

the web, application and database servers, of pharmacists’ computers, bandwidth of the VPN,

software bugs); and skills and capacities (e.g. skills and capacities of pharmacists to adopt a

new service, and of PMS vendors to adapt their solutions to a new API). For instance, as new

physicians started using the EPDI, new patients were migrated to the EPDI, and new

pharmacists were trained to start using the EPDI. This in turn, considerably increased the

number of transactions and at some points collapsed the capacity of the technological

infrastructure which had to be adjusted. That is, destabilizing events marked the transition to

a re-stabilizing work such as an increase of the processing capacity of the application servers,

database servers, and pharmacists’ computers, and the bandwidth of the VPN; the setting up

of a technical office and help-desk service to support pharmacists in resolving technical

problems and performing a baseline audit to check whether pharmacies were ready for

electronic prescription; the creation of an e-newsletter to inform pharmacists about the roll-

7 From 2012 some pharmacies started setting up 3G back-up connections.

24

out and the launch of new services; and the upgrading and corrective maintenance of the

software applications running on SIRE, SIFARE and PMS.

The rollout was completed during the third quarter of 2010. At that time all the pharmacies

were using the EPDI. On August 2010 the electronic prescriptions dispensed accounted for

50% of all the prescriptions being billed. On November 2011 an average of 385,222

prescriptions were dispensed electronically daily. This accounted for more than 75% of all

the prescriptions being billed. The CHS estimated that the EPDI had saved around 5,100,000

(patient) visits to primary care centers for collecting prescriptions. During 2011 the CHS

decided to extend the rollout of the EPDI to the specialized care, and in 2012 to the geriatric

residences, and home care. None of these rollouts entailed making significant changes to

SIFARE, PMS or the pharmacists’ routines. Early 2013, 83% of prescriptions dispensed in

pharmacies were electronic.

Tensions between the stability and change of pharmacists' practices remained as the CCP

launched new services independently of the CHS (see years 2011 to 2013 in appendix 1). The

rationale for those services was consistent with the role of the EPDI as an opportunity to re-

professionalize their practice (as seen in previous tension). To do so, the CCP leveraged its

ownership over certain elements of the architecture (the SIFARE and the VPN). On the one

hand, the CCP developed web apps for pharmacists (e.g. tools to support the invoicing,

management of alerts, management of users, management of digital signature, etc.). On the

other hand, the CCP launched new services to pharmacists through PMS vendors by

extending the API for PMS. The idea was that the PMS should be the entry door for

pharmacies to those services. For instance, at the end of 2011 the CCP announced the

“paperless pharmacy” project which entailed leveraging the SIFARE and the VPN to digitize

some paper-based procedures (e.g. recipe and narcotics books). Until then although most of

the PMS electronically stored those documents, pharmacists still had to periodically print

25

those documents and carry them physically to the Department of Health. With this project

pharmacists would keep using their PMS but that information would not only be locally

stored at the PMS but also at SIFARE. Then pharmacists would electronically sign and

submit that documentation that was stored at SIFARE to the Department of Health. This

“paperless pharmacy” project homogenized and reorganized some of the activities and flows

of data of pharmacists (e.g., the case of recipe and narcotics books) aiming to generate

efficiencies. Those data would not be stored in the pharmacists’ local computers or on paper

anymore, but centralized in the CCP’s server. To achieve its purpose, the CCP required the

cooperation and involvement of PMS vendors who had to adapt their solutions to the new

services. Accordingly, in 2013, the CCP worked on a new recognition program for PMS

vendors more oriented to technical aspects and tied to professional services. For instance, in

the case of pharmaceutical care, the CCP had already defined a protocol, and wanted that all

PMS vendors implemented that protocol, not its own one each. This new recognition program

was a component that further helped strengthen CCP’s ties with PMS vendors and this in turn

increased the degree of conformity of PMS. The new services also had the capacity to

reinforce the role of the CCP as a service provider for pharmacists.

Tension 3. Balancing the expansion of the EPDI and the compression of flows of

financial resources

A recurrent destabilizing force of the EPDI was its cost; in other words, the need for a steady

flow of financial resources to fund the infrastructure. On the side of the CCP, there was a

need to invest in the technological infrastructure comprising SIFARE and the VPN. The idea

was that the reduction of cost related to the processing paper-based dispensations for

invoicing (e.g. scanning and checking of dispensations) would cover the costs of the new

technological infrastructure of the CCP. But in any case the CCP searched for alternative

26

sources of funding. For instance, in 2008 they received funding for the VPN from the Center

of Innovation and Development of the Catalan Government.

On the side of individual pharmacies, they had to invest in the connectivity services to the

VPN, upgrading their PMS, the digital signature systems, swipe cards, and swipe card

readers. In 2008 the CCP reached an agreement with a Spanish bank. For those pharmacies

having an account in the bank with a minimum balance, the bank would pay part of their cost

of the connectivity and provide them with their digital certificates and swipe card readers for

free8. In 2010 the CCP received some financial support from the Department of Health of

Catalonia that was distributed among pharmacies. Moreover, a condition for PMS vendors

passing the recognition program was that they had to cover the costs of adapting their PMS to

interconnect with SIFARE as well as the costs of upgrading the PMS for their customers.

The economic crisis in Spain, which started in 2008, became another major destabilizing

force for the EPDI, particularly from 2010. As a result of the pressure from the EU to reduce

the deficit, the Spanish Ministry of Health and Social Security decreased the prices of public-

funded drugs and the margins of pharmacies. Second, the central and the regional

governments approved new taxes in 2012 by which citizens had to partially pay the drugs in

the pharmacy. Those taxes stimulated a fall in drug consumption and the pharmaceutical bill

dropped (see Appendix 4). Third, the financial tensions between the Spanish government and

the regional governments had also a destabilizing effect on Catalan pharmacies. The

autonomous regions, Catalonia among them, lost direct access to financial markets and the

Spanish government became the only source of funding for the regions. The Spanish

government leveraged that new scenario to put pressure on the autonomous regions in order

to reduce the deficit. Then from 2010 the CHS started committing repeated defaults to

8 In 2012 the funding agreement with that bank was still in force and the number of pharmacies benefiting from it had remained constant (around 1,300). As an example the total contribution of the bank was 373,680€ and 520,132€ on 2008 and 2012 respectively.

27

pharmacists. In short, as a result of the economic crisis there was a deceleration of the flows

of invoices, reimbursements, and funding that destabilized the EPDI.

The Catalan pharmacists faced those destabilizing events by leveraging again its ownership

and control over certain components of the EPDI. One initiative was the “paperless

pharmacy” project (already described in the previous section), which reduced the cost of

paper-based processing costs of pharmacists. Another relevant re-stabilizing event move was

the setting up by the CCP of a company called TICFarma in 2011. TICFarma offered

telecommunication services to the same pharmacies and pharmacists. With TICFarma, the

pharmacists (as a collective) increased its bargaining power in front of telecom providers.

TICFarma managed to: (1) reduce the connectivity costs for pharmacists, and (2) launch new

telecommunication services for pharmacists. Moreover, the CCP used TICFarma’s profits to

pay the cost of the technological infrastructure consisting of the SIFARE and the VPN.

Through TICFarma the CCP also reinforced its role as a service provider for pharmacists.

Likewise, the CCP mobilized a collective action of pharmacists to search for alternative ways

to fund the pharmacies and to put pressure on the Catalan government through measures such

as claiming the default interests judicially, organizing a campaign to collect signatures among

citizenship in defense of the pharmacies, and two general closures of the Catalan pharmacies

on October 25th 2012 and November 7th 2013. Overall, that reaction of the CCP increased the

cohesiveness and resilience of the pharmacists as a collective, and in turn, it re-stabilized the

EPDI.

6. Discussion

6.1. Stabilization and destabilization processes at work

The previous section has empirically depicted three main tensions around the digitization of

the paper-based prescription infrastructure in Catalonia from 2000 to 2013. The first tension

28

illustrated the need to balance the interests of the National government towards an integrated

National DI with those of the regional governments and other lower-scale collectives (e.g. the

pharmacists) on keeping control over the installed bases of systems and practices. Moreover,

the first tension also illustrated the balancing of the interests of the Catalan Health Service on

administrative and centralized control over the activity of pharmacists with the interests of

pharmacists on keeping their autonomy over their professional practice. The second tension

illustrated the need to balance the stability with the change of pharmacists’ and PMS vendors'

practices. The third tension illustrated the balance of pharmacists' needs to invest on a robust

infrastructure with their diminishing revenues from public-funded drugs.

These empirical results do not pretend to be an exhaustive list of tensions that extend the ones

reported by prior DI literature (Hanseth et al. 1996; Reimers et al. 2014; Tilson et al. 2010),

or to theorize about the mechanisms that balance those tensions (Wareham et al. 2014).

Rather, our results illustrate the idea that DIs are always in tension responding to continuous

processes of stabilization and destabilization. In broad terms, the digitization of the

prescribing, dispensing and invoicing processes involved a progressive disruption of the

existing ordering of the paper-based prescription infrastructure (i.e., freeing up existing

relations between the diverse components: physicians, patients, pharmacists, invoices,

documents, roles, legacy systems, and so forth), and subsequently rearranging and

recombining some of those components, as well as assembling new ones. Our narrative of the

digitization speaks of continuous processes of destabilization and re-stabilization which

provided the DI with both change and endurance, respectively. Our account has highlighted

the balancing acts of the CCP to overcome the tensions.

Stabilization and destabilization processes are related in different way. For instance, in our

case there are many examples of sequences of processes where destabilization is followed by

re-stabilization and also examples where a stabilization process triggers a destabilization

29

process. A typical example of the latter is that when one version of the EPDI stabilizes, this

triggers an adoption process which destabilizes the existing practices in pharmacies (which

again is followed by re-stabilization). In such cascades of processes there are also feedback

loops and self-reinforcing processes (Hanseth and Lyytinen 2010; Henfridsson and Bygstad

2013). The adoption of the EPDI contributes both to its stabilization and destabilization.

Adoption of the EPDI implies that it grows in number of computers and software modules

and this increases the costs of switching to a new version, i.e. it stabilizes the EPDI. At the

same time, growth in number of users of the EPDI generates more traffic which again implies

that after some time the EPDI must be upgraded to handle this growth, i.e. destabilizes the

EPDI. Further the stabilization of the EPDI triggered the generation of ideas about new

services that could be added to the EPDI. When such add-ons are adopted they again

contribute to both the stabilization and destabilization of the existing EPDI. We also see that

in principle independent processes interact and strengthen or weaken each other. For instance,

the financial crises triggered a number of different processes at various levels, all contributing

to destabilizing the EPDI in terms of constraining the funding of its operations.

We also see different kinds of destabilizing process. Some processes have the character of

more sudden incidents which immediately cause a breakdown of an assemblage (the EPDI,

pharmacies’ practices, etc.) and its restructuring and re-stabilization. This happened, for

instance, when serious software errors were detected. Other destabilizing processes have the

character of accumulation of destabilizing elements, but where change of the assemblage is

not happening until the number of destabilizing elements has reached a certain threshold.

This is typically the case with the adoption process where each new user of the EPDI

represents a destabilizing element, but where each element has no direct effect until the

threshold is reached. For instance, as the number of physicians and prescriptions going

through EPDI increased, the processing capacity of the servers and the bandwidth of the VPN

30

had to be scaled; otherwise, there would be more disruptions in the prescribing and

dispensing applications. At a certain threshold of disruptions (in terms of number or

duration), physicians and pharmacists perceived that they could not exercise their

professional practice, and accordingly, they temporally reversed the use of the EPDI. As

more physicians discontinued the use of the EPDI, the less value pharmacists saw in using the

EPDI.

6.2. Interactions between assemblages

The stabilizing and destabilizing processes through which the EPDI evolves are the outcome

from the interactions of the multiple heterogeneous components (or assemblages) –e.g., CHS,

CCP, PMS vendors, pharmacies, services, APIs, transactions, norms, economic crisis, laws,

banks, and so forth– characterized not by their properties but rather by capacities to interact,

i.e. relations of exteriority. In those interactions components exercised their capacity to affect

and be affected. Moreover, by conceiving the DI as characterized by relations of exteriority

we have shown that the DI is subject to tensions from its surroundings. That is, the DI is not a

static object; rather it is in-process, becoming, and ever-changing. The DI is not a whole with

clear boundaries, but an open-ended grouping of heterogeneous components (Hanseth &

Lyytinen 2010) that over time have established more or less durable relationships with each

other and the whole. Components enter into the whole via contingent relations. In other

words, the components of a DI are open to various “plugins” that can change at any time thus

rendering the degree of stabilization of a DI precarious and hence its evolution unknowable in

advance. In short, we consider that the notions of relations of exteriority and (de-

)stabilization offer a conceptual tool for grasping the dynamics and tensions characterizing

DIs and following them through space and time.

Another characteristic of our narrative about DI evolution is that it describes things and

events occurring at different levels of scale (pharmacy, collective of pharmacists represented

31

by the CCP, regional DI, and national DI) and the bottom-up movements of scales that lead to

the emergence and evolution of the DI. For instance, on the side of pharmacists (at a lower

scale), the multiple installed bases of pharmacies were articulated together through

components such as SIFARE, API, VPN, recognition program, agreement with a bank, and

so forth. Similarly, on the side health providers, the multiplicity of physicians' installed bases

was assembled. Then both assemblages (pharmacies and health providers) were articulated

together through SIRE, web services, regulations, security model, and so forth, and

synthesized into the EPDI. However, this bottom-up assembling process has not yet occurred

at the national scale since the multiple regional DIs are still not interoperable. Due to

differences among regions in terms of availability of resources, installed base of systems and

practices, and political interests and priorities, the several regional projects that were

underway followed their own design guidelines. Moreover, for these regional projects

interoperability across regions was not a priority. This definitively destabilized the vision of

the central government about the standardization and interoperability of the prescribing and

dispensing processes across Spain. That is, while the initial outcome of the PISTA project

destabilized the regional installed bases, the subsequent stabilization of those regional DIs9

ended up destabilizing the National DI. Yet at the European scale there has been an initiative

called epSOS10 which aimed to interconnect diverse National Contact Points supporting

electronic prescribing and dispensing. In the case of Spain, the regions of Andalusia,

Catalonia and Balearic Islands participated in the project. So in Spain, epSOS provided the

opportunity for interoperability across regional DIs without having to go through a National

DI. That is, some of the regional DIs were articulated into a European scale bypassing the

national scale. Overall, this study provides empirical evidence to support the idea that the 9 By the end of 2014 (15 years after the PISTA project was initiated) twelve out of seventeen regional DIs were stable (in the sense that in those regions 75% of the prescriptions were done electronically, and 100% of pharmacists used the DI for dispensing and invoicing). 10 Smart Open Services for European Citizens (www.epsos.eu) project run from July 1st 2008 to July 31st 2014.

32

existence of a DI at one scale depends on the stability of the DI at a lower scale and that any

DI emerges and evolves through bottom-up movements.

6.3 The co-functioning of architecture and governance

This section discusses how the co-evolution and co-functioning of the architecture and the

governance regime conditioned the processes of destabilization and re-stabilization processes

(see Table 1 for a summary of the properties of the architecture and governance structure).

As shown in the first tension, the architecture and governance regime were not static but

changed during the design of the EPDI (between 2004 and 2005). There was a process of

disaggregation in which the initial integral centralized architecture and the top-down

governance structure shifted to a dual one and middle-out approach respectively. The integral

CHS-centered architecture and the top-down governance structure were important conditions

in the simultaneous stabilization of the DI at the regional scale and destabilization of the DI at

the scale of pharmacists. With this integral CHS-centered architecture pharmacists could be

monitored, analyzed, and profiled by analytics in which the central database controlled by the

CHS would govern the constitution the subject (the pharmacist) as a crucial instrument of

control at a distance. So the control over the profession would reside away from pharmacists

in the central database and the analytics would be performed only by the CHS (government).

Moreover, the integral CHS-centered architecture involved a change in the relation between

CCP and pharmacists; the CCP would be detached from the day-to-day practices of

pharmacists such as dispensing and invoicing.Therefore, this configuration of architecture

and governance represented a powerful destabilization of pharmacists’ practices.

Then as a result of the re-stabilizing work of the CCP, that configuration of architecture and

governance shifted to a dual architecture (figure 3) comprising two central nodes (SIRE

controlled by the CHS, and SIFARE controlled by the CCP) and a middle-out approach

comprising a steering committee and follow-up committees that went toward bringing closer

33

the needs of health providers, government, pharmacists, Colleges of Physicians, and the CCP.

In the dual architecture both nodes (SIRE and SIFARE) stored very similar data and run

similar applications. Data would remain centralized residing in the servers of both the CHS

and the CCP thus conferring real-time visibility over the data to the CHS. At the same time,

the fact that data and applications would reside in the CCP’s server (SIFARE) also meant that

the CCP would be able to implement services that could enhance their professional

development.

Moreover, the middle-out approach to governance and the dual architecture altered the

organization of interdependencies. In particular, the CHS devolved the responsibility over the

design, development and operation of diverse parts of the EPDI to the CCP. That created two

centers of activity (governance’s property): the CHS and the CCP. The CCP would be in

charge of building the private network for pharmacies, developing the SIFARE, and

promoting EPDI’s use among pharmacists. Because the CCP would mobilize resources,

material (e.g. money) and expressive ones (e.g. legitimacy) for pharmacists, it was perceived

as a catalyst for the success of the project.

Overall, the configuration comprising the dual architecture and the middle-out approach to

governance supported the re-stabilization of the DI at the regional scale and the scale of

pharmacists. Yet a side effect of this configuration was the potential increase of the

heterogeneity within pharmacists, which would further destabilize the DI on their side.

Tension 2 shows how this side effect was minimized with a decoupling of one of the central

nodes (SIFARE) from pharmacists’ PMS. This decoupling increased the degree of

modularization of the architecture on the side of pharmacists. The functionalities of SIFARE

were decomposed into loosely coupled components and the interface specifications for how

PMS should interact with SIFARE were codified through web services. This way the CCP

was able to incorporate gateways (e.g., recognition program, API) that facilitated the

34

integration of PMS vendors. Hence, those gateways respected and kept up the multiple

practices and systems of pharmacists, as well as accommodated the number and frequency of

new services sponsored by the CHS to the interests and capacities of pharmacists and PMS

vendors. In that respect, the architecture became modular in production (Baldwin and Clark

2000), adding stability to the installed base of pharmacists while at the same time giving them

space for certain autonomy and new innovations. This shift to a loosely coupled architecture

between the central node (SIFARE) and pharmacists (PMS) was also accompanied by new

components that further decentralized the governance structure (e.g. advisory committee

between the CCP and PMS vendors, and the adjustments to the recognition program). This

governance structure enabled the CCP to keep the control over pharmacists and PMS vendors

(stabilizing the EPDI) while at the same time delegated some roles and responsibilities to

PMS vendors and by doing so, stimulated that some decisions could be taken in a more

decentralized manner (destabilizing the EPDI). Moreover, this configuration of architecture

and governance allowed the EPDI to grow autonomously on the side of pharmacists without

the need for a CHS-centered control.

Finally, tension 3 shows how the destabilization and re-stabilization processes associated

with the expansions and compressions of flows of financial resources were conditioned by a

combination of components of the dual architecture and the middle-out governance structure.

The CCP further leveraged its control over some architectural components (SIFARE and the

VPN) and its position as representative of pharmacists, all of which were obligatory passage

points (Callon 1986) for the CHS, pharmacists, PMS vendors, and telecom providers, to

undertake a series of actions. One of those actions was the creation of TICFarma, which

emerged at the nexus of a series of destabilizing flows related with the economic crisis. For

instance, delays by the CHS in the reimbursement of the pharmacists’ invoices (a slowdown

of the speed of flows) combined with a lack of alternatives for funding pharmacies (Spanish

35

banks limited the credit to companies), and with the reduced margins of drugs that led some

pharmacies to limit the diversity of drugs that they dispensed11 (volume reduction of flows of

public-funded drugs and invoices). Before TICFarma, the infrastructure built by the CCP (for

pharmacists) had been mainly financially supported by the same pharmacists through a

percentage of the invoices. With TICFarma that infrastructure would be also funded by the

telecom providers and the potential new services that could be developed (e.g. the paperless

pharmacy project). Hence, TICFarma helped to dampen the decelerations of those flows (e.g.,

payments, funding) by ensuring the continuity of use of the EPDI, and ultimately re-

stabilized the EPDI.

In short, we have shown how the interactions among the properties of the architecture (being

dual and modular) and the middle-out governance structure endowed the DI with the capacity

to evolve in a way that balanced the homogenization of practices with the stimulation of new

innovations. That is, the architecture and governance of the EPDI created a space in which

diverse flows of actors, information services, technological resources, financial resources,

and skills passing through the DI could more easily circulate.

Table 1: Properties of architecture and governance

Tension Architecture Governance

Centralized control and autonomy of actors at national at regional levels

From an integral centralized tight-coupling architecture comprising a single data base, and a single application for dispensing and invoicing,

to a dual architecture at the core that decoupled the two central nodes: SIRE and SIFARE).

From a top-down approach and centralized control structure (national and regional),

to a middle-out approach that brings together the needs of regional government (CHS), health providers, pharmacists (and the CCP), Components of the governance between CHS and CCP: 1) Amendment to the pharmaceutical agreement; 2) Steering & follow-up committee for EPDI; 3) Decree regulating the EPDI.

11 Some pharmacies stop dispensing some drugs (particularly the most expensive drugs) arguing that they could not afford funding themselves the drugs.

36

Stability and change of pharmacists

Modular architecture decoupling the CCP’s central node (SIFARE) from pharmacists’ systems (PMS). The CCP used gateways (not standards) such as recognition program, or API exposed in DLL to minimize disruption of installed base.

Platform-architecture on the side of pharmacists that allowed new innovations from the CCP (new data services, paperless pharmacy project) and from 3rd parties (PMS vendors).

New components of the governance between CHS and CCP: 1) Act for the roll-out; and 2) Security model (encryption of communications between SIRE and SIFARE; CCP as a Registration Authority).

Components of the governance between the CCP and pharmacists’ side: 1) Advisory committee between CCP and PMS vendors; 2) Recognition program for PMS vendors; 3) Agreement with telecom provider; and 4) Security model (digital certificate for pharmacists).

Extension of the recognition program for PMS vendors.

Expansion and compression of flows of financial resources

Dual architecture, and in particular the architectural control points: SIFARE and VPN.

Components of the governance between the CCP and pharmacist: 1) Advisory committee between CCP and PMS vendors; and 2) Redefinition of the recognition program for PMS vendors.

Foundation of the TICFarma company.

6.4. Towards a contingency theory of DI evolution

Our findings support previous research (reviewed in section 2) showing that DIs evolve

according to the way relevant tensions are managed and that the evolution of DIs is shaped by

the architecture and governance of DIs. But our findings extend previous research by

providing a more integrated model that depicts DI evolution as the interaction between

processes of stabilization and destabilization that are conditioned by the DI’s specific

configuration of architecture/governance.

We consider the EPDI’s specific configuration of architecture/governance as a fortunate

choice. It has been a key factor in the successful establishment and evolution of the DI. This

is of special significance because both the architecture and the governance structure are far

from optimal according to, if not directly contradicting, conventional Software Engineering

37

and IS development and implementation wisdom. The dual architecture introduced what was

seen as unnecessary redundancy and complexity. At the same time, the governance structure

where two governing units (CHS and CCP) worked to a large extent independently and

autonomously could be seen as representing a high risk for the fragmentation of the solution

to be developed. However, this configuration of architecture and governance structure was

well aligned with the structures of the user community which is characterized by two main

groups: pharmacies/pharmacists and health care providers/physicians. This configuration

supported the evolution of the DI into two loosely connected platforms which made it fairly

easy for the CHS and CCP to reach agreement about what and how data should be shared.

Moreover, this configuration enabled the CHS and the CCP develop additional services on

top of their platforms for their respective user communities. This conforms to the findings of

Constantinides and Barrett (2014) who suggest that a polycentric approach to DI governance

may support successful infrastructure development and scalability. We consider that the

configuration of architecture/governance identified in our study can be successfully replicated

in the development of shared DIs for user communities that are constituted by clearly

identifiable user groups.

In line with Henfridsson and Bygstad (2013) we suggest that we need to move towards

contingency theories (or mid-range theories) about the relations between specific

configurations of architecture and governance and how they fit specific kinds of DIs and

business sectors. We may move towards such a theory by combining our findings regarding

the performance of the configuration of architecture/governance in our case with others’

findings. This includes for instance, the Internet’s combination of end-2end architecture and a

governance structure composed of a huge loosely connected networks of individual

developers and development organizations drawing heavily upon the Internet itself for

coordination and information sharing (Hanseth and Lyytinen 2010; van Schewick 2010). The

38

smartphone centered ecologies based on platforms and “appstores” where individual actors

(e.g. Apple, Google) control the platform (Eaton et al. 2015; Tilson et al. 2012). Also

Henfridsson and Bygstad (2013) found that DIs characterized by loosely coupled architecture

and a governance regime based on decentralized control and where the three self-reinforcing

mechanisms (innovation, adoption, and scaling) were present, evolved successfully.

Moreover, they also found three successful cases characterized by a tightly coupled

architecture and a centralized control governance structure, where only adoption and scaling

mechanisms were present. Henningson and Hanseth (2011) studied the European Union

eCustoms initiative aimed to computerize and harmonize the diverse existing national DIs

and integrating them into a pan-European eCustoms DI. By drawing upon assemblage theory,

they depict the evolution of the eCustoms DI as dialectic between stabilizing and

destabilizing processes, and show how the loose coupling between the different national DIs,

combined with the tight-coupling of national DIs, and a governance structure characterized

by a classical hierarchical structure at the national level rather than at the European level,

shaped the evolution of the diverse DIs. The authors argue that while national DIs evolved

quite successfully, there was a fragmentation of the European eCustoms DI. In that sense, the

authors argue that the “eCustoms initiative have failed because they stabilized what should

have been destabilized (the exiting national DIs) and destabilized what should have been

stabilized (i.e. limit the growth and integration of additional systems)” (p. 17). Finally, in

their research on DI initiatives in the Norwegian healthcare sector, Hanseth and Bygstad

(2012, 2014) found that a configuration based on centralized architecture and centralized

control was most successful for establishing new and innovative DIs, while a decentralized

architecture and decentralized control were crucial for further development of and

innovations on the DI after it was successfully established.

39

The different DIs mentioned here are developed to support very different kinds of services

and user groups. Some of the initiatives aim at stimulating maximum innovations while

others aim at restructuring and harmonizing existing DIs. These are important parameters

when determining which stabilizing and destabilizing processes one want to generate in order

to make a DI evolve in desired directions and then which architecture governance

configurations can help achieve that.

7. Conclusion

This paper has explored the evolutionary dynamics of DIs. We have shown that DIs are

characterized by fluidity and openness to number and types of users and limitless possibilities

for re-combinations of digital artifacts, while at the same time having closed and static

structures that give them continuity. By drawing upon assemblage theory, this paper has

articulated a two layer model that conceives the evolution of DIs as a set of interacting

stabilizing and destabilizing processes which are conditioned by the co-functioning of the

DIs’ architecture and governance structure. Our account builds on and integrates the ongoing

debates in the literature which on the one hand study the influence of architecture and

governance in the design and evolution of DIs (Constantinides and Barrett 2014; Hanseth and

Bygstad 2014; Henfridsson and Bygstad 2013; Tiwana 2014), and on the other hand, which

depict DIs as characterized by tensions that need to be balanced (Hanseth et al. 1996; Ribes

and Finholt 2009; Tilson et al. 2010; Wareham et al. 2014). We suggest that further research

could investigate different dynamics around the processes of stabilization and destabilization

that characterize DI evolution, as well as other configurations of architecture and governance

in other settings that will extend our results.

References

ANNEX. 2005. “Annex al Concert d’Atenciَ Farmacèutica entre el Servei Català de la Salut i el Consell de Col.legis Farmacèutics de Catalunya, per establir les clàusules de

40

desenvolupament de la prova pilot sobre la recepta electrٍnica en l'àmbit de determinades àreas b,” , pp. 1–6.

Baldwin, C. V., and Clark, K. B. 2000. Design Rules: The Power of Modularity. Vol. 1. Cambridge, Mass: MIT Press

Baldwin, C. Y., and Woodard, J. 2009. “The architecture of platforms: a unified view,” in Platforms, Markets and Innovation, A. Gawer (ed.), Edward Elgar, pp. 19–44.

Blanchette, J. 2012. “Viewpoint: Computing as if Infrastructure Mattered,” Communications of the ACM (55:10), pp. 32–34.

Ciborra, C., Braa, K., Cordella, A., Dahlbom, B., Failla, A., Hanseth, O., Hepsø, V., Ljungberg, J., Monteiro, E. and Simon, K2000. From Control to Drift: The Dynamics of Corporate Information Infrastructures, Oxford: Oxford University Press.

Constantinides, P. and Barrett, M. 2014. “Information Infrastructure Development and Governance as Collective Action”, Information Systems Research (39:1), pp. 1-17. http://dx.doi.org/10.1287/isre.2014.0542

Cordobés, A. 2002. “La receta electronica (II). Proyectos de las comunidades aut�nomas,” Offarm (21:10), pp. 148–156.

Cordobés, A. 2002. “Receta electr�nica (I). Proyecto Pista y repercusiones sobre la oficina de farmacia,” Offarm (21:8), pp. 142–150.

DECRET. 2007. “DECRET 159/2007,” Diari Oficial de la Generalitat de Catalunya, (Vol. 25622) , p. 25620.

DeLanda, M. 2006. A New Philosophy of Society: Assemblage Theory and Social Complexity, Bloomsbury Publishing Plc.

Eaton, B., Elaluf-Calderwood, S., Sørensen, C., and Yoo, Y. 2015. “Distributed Tuning of Boundary Resources The Case of Apple’s iOS Service System,” MIS Quartely (39:1), pp. 217-244.

Edwards, P. N., Jackson, S. J., Bowker, G., and Knobel, C. P. 2009. “Introduction: An Agenda for Infrastructure Studies,” Journal of the Association for Information Systems (10:5), pp. 364–374.

Gawer, A. 2014. “Bridging Differing Perspectives on Technological Platforms: Toward an Integrative Framework,” Research Policy (43:7), pp. 1239–1249.

Grisot, M., Hanseth, O. & Thorseng, A.A., 2014. “Innovation of, in, on infrastructures: Articulating the role of architecture in information infrastructure evolution,” Journal of the Association of Information Systems (15:4), p.197-219.

Hanseth, O., Monteiro, E., and Hatling, M. 1996. “Developing Information Infrastructure: The Tension Between Standardization and Flexibility,” Science, Technology, and Human Values (11:4), pp. 407-426.

Hanseth, O., and Ciborra, C. 2007. Risk, Complexity and ICT, Edward Elgar Publishing. Hanseth, O., and Lyytinen, K. 2010. “Design theory for dynamic complexity in information

infrastructures: the case of building internet,” Journal of Information Technology (25:1), pp. 1–19.

Hanseth, O., Bygstad, B., Ellingsen, G., Johannesen, L.K. and Larsen, E. 2012. “ICT Standardization Strategies and Service Innovation in Health Care,” in International Conference on Information Systems, Orlando.

41

Hanseth, O., and Bygstad, B. 2014. “Generative Information Infrastructure Architectures. A Longitudinal Study of eHealth Infrastructures in Norway,” Unpublished manuscript.

Henfridsson, O., and Bygstad, B. 2013. “The generative mechanisms of digital infrastructure evolution,” MIS Quarterly (37:3), pp. 907–931.

Henningsson, S., and Hanseth, O. 2011. “The Essential Dynamics of Information Infrastructures,” in International Conference on Information Systems, Shangai.

Hepso, V. ;Monteiro, E. . R. H. 2009. “Ecologies of e-Infrastructures,” Journal of the Association for Information Systems (10:Special Issue), pp. 430–446.

Langley, A. 1999. “Strategies for Theorizing from Process Data,” Academy of Management Review (24:4), pp. 691–710.

Miles, M. B., and Huberman, A. M. 1994. Qualitative Analysis: An Expanded Sourcebook, London: Sage Publications.

Nielsen, P., and Aanestad, M. 2006. “Control Devolution as Information Infrastructure Design Strategy: A Case Study of a Content Service Platform for Mobile Phones in Norway,” Journal of Information Technology (21), pp. 185-194.

ORDRE. 2008. “ORDRE SLT/72/2008,” Diari Oficial de la Generalitat de Catalunya, (Vol. 16457) , p. 16454.

Reimers, K., Johnston, R., and Klein, S. 2014. “An empirical evaluation of existing IS change theories for the case of IOIS evolution,” European Journal of Information Systems (23:4), pp. 373–399.

Ribes, D., and Finholt, T. A. 2009. “The Long Now of Technology Infrastructure: Articulating Tensions in Development,” Journal of the Association for Information Systems (10:5), pp. 375–398.

Rolland, K. H., and Monteiro, E. 2002. “Balancing the Local and the Global in Infrastructural Information Systems.,” Information Society (18:2), pp. 87–100.

Silsand, L., and Ellingsen, G. 2014. “Generification by Translation: Designing Generic Systems in Context of the Local,” Journal of the Association for Information Systems Research (15:Special Issue), pp. 177–196.

Tilson, D., Lyytinen, K., and Sorensen, C. 2010. “Research commentary -- Digital infrastructures: The missing IS research agenda,” Information Systems Research (21:4), pp. 748–759.

Tilson, D., Sorensen, C., and Lyytinen, K. 2012. “Change and control paradoxes in mobile infrastructure innovation,” in 45th Hawaii International Conference on System Sciences, pp. 1324–1333.

Tiwana, A. 2010. “Platform Evolution: Coevolution of Platform Architecture, Governance, and Environmental Dynamics,” Information Systems Research (21:4), pp. 675–687.

Tiwana. 2014. Platform Ecosystems: Aligning Architecture, Governance, and Strategy, Morgan Kauffmann.

Ure, J., Procter, R., Lin, Y.-W., Hartswood, M., Anderson, S., Lloyd, S., Wardlaw, J., Gonzalez-Velez, H., and Ho, K. 2009. “The Development of Data Infrastructures for eHealth: A Socio-Technical Perspective,” Journal of the Association for Information Systems (10:5), pp. 415–429.

Van Schewick, B. 2010. Internet architecture and innovation, MIT Press.

42

Wareham, J., Fox, P., and Cano, J.L. (2014). “Technology Ecosystem Governance,” Organization Science (25:4), pp. 1195-1215.

Yin, R. K. 2003. Case Study Research: Design and Methods, (Vol. 3rd) Thousand Oaks, CA: Sage.

Zittrain, J. L. 2008. The future of the Internet and how to stop it, New Haven & London: Yale University Press.

43

8. Appendix 1. Timeline of events

Next table presents a timeline for the events from 2000 until 2013.

Table 1: Timeline for the EPDI

2000

The Spanish Ministry of Science and Technology in collaboration with governments of the autonomous regions as well as the representatives of professional agents involved in the prescription and dispensation processes (physicians and pharmacists) starts working on the foundations for a common Spanish EPDI.

2001

The CCP leads a first pilot of EPDI, involving a hundred private physicians and 25 community pharmacies.

The CCP tries to extend this pilot to the public health but fails.

From October 1st, incorporation of the “individual medical card” to the dispensing process. This accelerated the process of computerization of Catalan pharmacies.

In parallel, the CCP, in collaboration with other professional colleges, sets up a company (FirmaProfessional) which issues digital certificates. Some pharmacies start using those certificates to send electronically their data for invoices or other kind of files to the CCP, or check online the data of patients’ individual medical cards.

2002 A first draft for a Spanish EPDI is presented. Since Spain has a decentralized health care system, the diverse autonomous regions are expected to start their own EPDI.

2004 The CHS sets the foundations for the building of an EPDI for the public health.

EPDI’s design is negotiated between the CHS and the CCP.

2005

During the first half of the year the EPDI’s design is approved.

The CHS defines the functional specifications for its system (called SIRE) and starts its development (which is outsourced to an IT services company, a joint venture of Telben and Accenture).

A governance structure for the project consisting of two committees (steering committee and executive committee) which involve all the actors (CHS, Health Providers, CCP, ….) is created. This governance structure is still valid in 2013. Obviously the role of the committees has changed throughout the phases of the project.

The CCP starts the development of its system (called SIFARE, which is outsourced to an IT services company, EDS which in May 2008 was bought by HP).

The CCP launches a recognition program for PMS vendors.

5 PMS vendors get the recognition.

The CCP issues an invitation to tender for the pharmaceutical virtual private network (VPN).

2006

The CHS completes a first version of SIRE (and its web services).

The CCP completes a first version of SIFARE (and the associated web service) and releases API 1 (for PMS vendors).

The CCP reaches an agreement with PMS vendors in order that the latter will assume the cost of upgrading their software for pharmacies.

Telefonica wins the tender to build the pharmaceutical VPN.

A pilot of the project starts and a first real prescription is done on April 2006.

44

The rollout of the VPN starts with the pharmacies that participate in the pilot.

Due to repeated technical problems and errors with SIRE, the CHS stops the pilot and starts a new version of SIRE to address those problems (the CCP hires a new IT services company, ATOS).

2007

1 additional PMS vendor gets the recognition.

The CCP gets a subsidy from the Center of Innovation and Development of the Catalan Government for the infrastructure

A new version of SIRE and SIFARE are completed on November.

At the end of the year the CHS resumes the pilot involving two health regions (in the provinces of Girona and Tarragona). The physicians of primary care services, pharmacies and patients are progressively added into the pilot, being physicians the ones who decide which patients must be prescribed electronically.

The rollout of the VPN continues among the pharmacies involved in the pilot.

2008

3 additional PMS vendors get the recognition.

On May the pilot is satisfactorily completed. The pilot has involved 70 physicians, 56 pharmacies and 17,000 patients, and 314,500 prescriptions have been dispensed.

The CHS starts the rollout of the EPDI during the 3rd quarter. The rollout consists of four phases each involving different health regions. The last phase (and health region) is that of Barcelona (that involves 2,200 of the 3,000 pharmacies in Catalonia).

The CCP reaches an agreement with a Spanish bank by which the bank will pay part of the cost of the connectivity for pharmacies as well as provide pharmacies with their digital certificates and swipe card readers for free.

On December the CCP creates an e-newsletter and a technical office to support pharmacies during the rollout.

2009

On January the CCP sets up a helpdesk to support pharmacists.

The CCP releases API 2 (for PMS vendors).

The last phase of the EPDI’s roll-out which involves the health region of Barcelona starts by mid-year

2010

On January the electronic prescriptions dispensed accounted for around 23% of all the prescriptions being billed.

The rollout of the VPN completes.

The Catalan Department of Health gives a financial aid to the CCP to support the connectivity of pharmacies

The CCP launches new web applications to support pharmacists (e.g. tools to support the invoicing, management of alerts, management of users, management of digital signature, etc)

The EPDI’s roll-out for all the primary care centers is completed during the third quarter of the year. All the more than 3,000 Catalan pharmacies are daily using the EPDI.

On August the electronic prescriptions dispensed accounted for 50% of all the prescriptions being billed.

2011

Due to some unsatisfactory response from the telecom provider (Telefonica), the CCP decides to extend the helpdesk services to include the monitoring of the infrastructure in order to detect failures before pharmacists realize. The aim is to minimize the impact of massive failure by acting in advance.

On November an average of 385,222 of electronic prescriptions are dispensed daily. The CHS

45

estimates that the EPDI had saved around 5,100,000 (patient) visits to primary care centers for collecting prescriptions.

The CHS does some pilots of EPDI in the specialized care and mental care.

The CCP sets up TicFarma (a business) to leverage the CCP’s ownership of the VPN, and hence its bargaining power in front of telecom providers to reduce the costs of telecommunication services (data and voice) for pharmacists. Apart from reducing the connectivity costs for pharmacists, with TicFarma the CCP also aims to launch new telecommunication services for pharmacists. With TicFarma the CCP seeks to transform all the pharmacies into a corporation which offers telecommunication services to the same pharmacies and pharmacists. Moreover, TicFarma’s profits are used to pay the cost of the CCP’s infrastructure consisting of the SIFARE and the VPN.

By the end of the year, the CCP announces the “paperless pharmacy” project. This project entails leveraging the SIFARE and the VPN to digitize some paper-based procedures (e.g. recipe and narcotics books) and interactions of pharmacies with the Department of Health. With this project the CCP will develop a set of web services and will release a new API for PMS vendors.

2012

The CHS starts the EPDI’s roll-out to specialized care. That rollout does not entail substantial changes to SIFARE or PMS.

The CHS integrates the paper-based prescriptions and dispensations into the EPDI

The Catalan government approves the “euro per prescription” tax by which patients will pay an additional euro for each prescription at the pharmacy. This new tax requires releasing new versions of the SIRE and SIFARE.

The CCP releases API 3 (for PMS vendors) to support the “euro per prescription” tax.

The Spanish government passes the co-payment act which entails that citizens will pay drugs based on their income. The calculation of the final amount and the payment will take place in the pharmacy when the patient picks the drug. This act entails making changes to SIRE and SIFARE.

The CCP releases API 4 (for PMS vendors) to support the so-payment act.

The CHS starts the roll-out in geriatric residences, and home care. Such a rollout does not entail any changes to SIFARE or PMS.

First closure of the Catalan pharmacists on October in protest for the repeated defaults of the CHS.

2013

The CCP redefines the recognition program for PMS vendors in order to include new professional services in line with the ones defined in the “paperless pharmacy” project.

The CCP releases API 5 (for PMS vendors) that supports the new recognition program.

The CHS works on new services that involve changes to SIRE, SIFARE and PMS –for instance, informing patients about the cost of treatments. Those changes are progressively incorporated in the new versions of SIFARE’s web services.

Second closure of the Catalan pharmacists on November in protest for the repeated defaults of the CHS.

46

9. Appendix 2. SIFARE Web services and APIs developed by the

CCP

A first version of the API was released in 2006 coinciding with the pilot. A second one was

released in 2009 before the massive roll-out in Barcelona. The third and fourth versions were

released in 2012 coinciding with two main regulatory changes, one from the Catalan

Government and another from the Spanish government; a fifth version which was to be

released by end 2013. The fifth API was launched in 2013 and included new professional

services (e.g. the case of the “paperless pharmacy” project).

List of SIFARE web services, their version and the corresponding API.

SIFARE Web Service API 1 (2006)

API 2 (2009)

API 3 (early 2012)

API 4 (oct.

2012)

API 5 (2013)

Calculate prescriptions that need to be dispensed v1 v3 v10 v12 v12

Look up next dispensation date v1 v2 v4 v5 v5

Insert dispensation v1 v1 v4 v6 v6

Insert paper-based dispensation v1 v2 v3

Delete dispensation v1 v1 v1 v2 v2

Delete paper-based dispensation v1 v2 v2

Look up dispensations made v1 v2 v4 v8 v8

Look up dispensations made by pharmacy v1 v1 v2 v4 v4

Look up paper-based dispensations made by pharmacy v1 v2 v2

Look up dispensations that need to be signed v1 v4 v7 v9 v10

Creation and signature of lots of prescriptions v1 v1 v1 v1 v1

Look up of contingent codes per pharmacy v1 v1 v2 v1 v1

Insert delayed dispensation v1 v1 v1 v3 v4

Block prescription v1 v1 v2 v2 v2

Create product reserve v1 v1 v2 v2 v2

Consult prescriptions v2 v2 v2

Consult generic alerts v1 v1 v1

47

Consult document associated to an alert v1 v1

Enable/Disable/Confirm reading of messages v1 v1 v1

Consult messages to pharmacists v2 v3 v3

Insert message v1 v1 v1

Consult required tax v2 v3 v4

Insert tax v1 v1 v1

Delete tax v1 v1 v1

Consult tax applied by pharmacy v1 v1 v1

Consult tax applied by citizen v1 v1 v1

Consult warning generated by the CHS v1 v1 v1

Consult document associated to a warning v1 v1 v1 v1 v1

Insert warning for the physician v1 v1 v1 v1 v1

48

10. Appendix 4. Evolution of pharmaceutical bill in Catalonia

Evolution of the pharmaceutical expenditure of the Catalan Health Service in the period 2004

to 2013.

Year   Pharmaceutical  expenditure  in  €  (1)  

Number  of  prescriptions

2004   1.135.994.555   88.349.827 2005   1.192.708.331   93.255.814 2006   1.240.916.916   95.908.647 2007   1.272.866.297   100.606.918 2008   1.364.946.036   107.207.365 2009   1.392.951.071   109.014.080 2010   1.386.379.728   112.320.893 2011   1.282.039.257   115.386.329 2012   1.136.101.523   110.619.114 2013   987.091.295   98.204.823 2014   1.009.847.263   100.053.114

     (1)  Aggregated  period  January  -­‐  September