Network effectiveness and the ABR Health
Network Holland West
Master’s Thesis
Msc Public Administration
Track: Public Management and Leadership June 2019
Student: Joran Rust
Student number: s1707086 Supervisor: dr. Jelmer Schalk
2 Table of Contents 1. Introduction ... 3 1.1 Introduction ... 3 1.2 Academic relevance ... 4 1.3 Practical relevance ... 5
1.4 Structure of the Research ... 6
2. Context ... 7
2.1 Policy on ABR health networks ... 7
2.2 Identifying the different actors in the health networks: ... 8
2.3 Differences between the health networks ... 9
3. Theoretic framework ... 11
3.1 Origin of the network approach ... 11
3.2 Performance of networks ... 12
3.3 Network governance and effectiveness ... 12
3.4 Roles in a network ... 19
4. Application of theory to the ABR health network Holland West ... 21
4.1 Health network Holland West ... 21
4.2 Contextualized Hypotheses ... 23
5. Research design and data collection ... 28
5.1Methodology ... 28
5.2 Case selection ... 29
5.3 Operationalization ... 30
5.4 Method of data collection ... 40
5.5 Method of data analysis ... 41
5.6 Validity and reliability ... 42
6. Analysis ... 45
6.1 Network Structure Analysis ... 45
6.2 Effectiveness of the health network Holland West ... 61
6.3 Possible points of improvement ... 75
7. Conclusion ... 79
7.1 Final conclusion ... 79
7.2 Implications for theory and practice ... 82
7. 3 Limitations and future research ... 83
8. References ... 85
Appendix I – The Health Network Goals ... 90
Appendix II – Zorgmarkten ABR health network Tasks ... 92
Appendix III – Topic List and Questionnaire ... 93
Appendix IV – Overview of the documents ... 98
Appendix VI – Coding Tree ... 101 Appendix VII – Trasncripts ... Fout! Bladwijzer niet gedefinieerd.
3
1. Introduction
1.1 Introduction
Antibiotics resistance and especially resistant micro-organisms (BRMO) are causing health problems around the world. This is because these resistant micro-organisms make it harder to treat diseases. Medicine which would normally be prescribed to cure a disease no longer works if microbes become resistant. At the moment the number of microbes that become resistant to the corresponding medicine is increasing. This resistance generates a threat to human life. Antimicrobial resistance leads to longer stays in hospitals, increased duration of illness, increased mortality rates and patients are less protected when they undergo surgery. The World Health Organization (WHO) calls upon countries to take antimicrobial resistance seriously and to act accordingly. It recognizes that antimicrobial resistance isn't located in one type of health organization or specific health sector. Therefore the WHO recommends an approach where all different health sectors and organizations work together and involve each other in the process of reducing antibiotics resistance (WHO, 2015).
Antimicrobial resistance poses such a threat to mankind that the World Health Organization (WHO) decided to create a Global Action Plan to reduce the effect of antibiotics resistance. The goal of the action plan is to improve awareness and understanding, strengthen knowledge, reduce the incidences of infections, optimize the use of medicine and to develop an economic case which stimulates investments in new medicines. Individual countries are stimulated by the WHO to generate their national action plan on antibiotics resistance which should be in line with the plan of the WHO (WHO, 2015). Besides the WHO other organizations also stress the importance of reducing antibiotics resistance. The European Union, for instance, created an action plan against antimicrobial resistance. The European plan was called the One Health Action Plan against Antimicrobial Resistance. It was created in 2011 and renewed in 2017 after review (European Commission, n.d).
In addition, scholars wrote about the threat of antimicrobial resistance. According to O'Neil (2014), antibiotics resistance is already claiming at least 50,000 lives every year in Europe and the United States, and on top of this hundreds of thousands are dying in other parts of the world. His review finds that antimicrobial resistance will be the number one cause of death if the current trend continues. This will result in antimicrobial resistance becoming more deadly than cancer (O'Neill, 2014, pp. 5-6).
4
Edith Schippers (2015), former Dutch Minister of Public Health, Welfare and Sports, wrote a letter to the Dutch Second-Chamber on Antibiotics resistance. She stated that compared to other countries the Netherlands is performing well on antibiotics resistance, but reports from the Inspection for Healthcare at the end of 2013 showed that there is still room and need for improvement. In the letter, she mentioned the threat of antibiotics resistance and stated that the cooperation on antibiotics resistance between health organizations lacked effectiveness. The movement of patients across different health organizations makes it hard to track their health status and their possible resistance to antibiotics. Therefore it is needed to produce a form of cooperation between the different health organizations, which allows them to function as one entity on the subject of antibiotics resistance. To facilitate this cooperation Schippers mentions that the different health organizations need to form networks (Schippers et al., 2015). This plea was later translated into 10 regional health networks with the goal to reduce antibiotics resistance in the Netherlands.
In this research, we try to determine the effectiveness of one of these health networks to provide best practices and possible points of improvement. We examine the health network from a network perspective where we examine the interactions of the entire health network and determine the presence of network characteristics. This research utilizes the case of the Antibiotics Resistance health network Holland West. Therefore the research question is as follows: How effective is the network governance approach of the antibiotics resistance health
network Holland West and can it be improved? 1.2 Academic relevance
This research has academic relevance in three ways. First, it establishes the importance of studying the performance of public organizations from a network perspective. Scholars originally focused on intra-organizational patterns to establish performance, but throughout the years a broader view was utilized. Scholars started to pay more attention to external processes as opposed to intra-organizational processes and patterns. In the 1960s scholars started to use an organizational perspective. They started to illustrate and describe the relevance of inter-organizational patterns for an organization. This perspective was mainly focused on the relationship between two organizations and not on a multitude of relationships between organizations in a network (Levin & White, 1961; Wiewel & Hunter, 1985). Almost thirty years after the inter-organizational perspective scholars found that it was important to look at entire networks as opposed to only two organizations (O'Toole, 1997; Agranoff & McGuire, 2001;
5
Provan & Milward, 2001). These scholars illustrated the importance of networks as the unit of analysis in public management research.
Second, measuring effectiveness. The relevance of measuring effectiveness has often been stressed. Different authors mentioned the importance of being able to measure and determine the effectiveness of public organizations. We can determine if an organization can deliver the required services or improve its service delivery if we can measure its effectiveness. Once the effectiveness has been measured managers will be able to determine which policies to implement and how to allocate resources to be more effective and improve their services (O'Toole, 1997; Turrini, 2010; Provan & Lemaire, 2012).
Third, we will combine literature about network governance to create a new method to measure and examine the effectiveness of networks. Previous articles on the effectiveness of networks conclude that there is not one way to measure network effectiveness and more research is needed (Provan & Milward, 1995; Provan et al., 2007; Provan & Kenis, 2007; Turrini et al., 2010). Besides being difficult to measure it’s also considered to be a normative endeavor due to the need to specify when a network is considered to be effective (Kenis & Provan, 2009). Therefore a literature gap remains for network effectiveness due to the ambiguity on the methods to measure effectiveness and the normative characteristic of network effectiveness. The case of health network Holland West provides academic relevance because it provides insights into a new type of network with specific goals and ambitions. By combining different theories on network governance we determine the applicability of these theories and create a new strategy in measuring network effectiveness. This might result in an improved method for measuring network effectiveness.
1.3 Practical relevance
When service-delivery networks are governed properly their effectiveness and efficiency will increase, which will result in the production of more public value (Provan & Milward, 2001, pp. 414-415). The ABR health networks are created to generate a better stream of information about antibiotics resistance amongst different health organizations in the region. According to Dutch law, they need to fulfill specific tasks and accomplish certain goals to be able to receive government grants. To accomplish this, the different organizations within the network have to work together to generate collective outcomes. The degree in which they will be able to do so depends on the effectiveness of the network. Therefore it is not surprising that the health networks start with a pilot period. During the pilot period, the health networks are supposed to
6
produce and exchange best practices to have the best possible set-up for a health network at a later stage (Schippers, 2016). By examining the network of Holland West we can determine whether there are best practices or points of improvement in the network. These points could then be shared with and amongst the other health networks to improve their effectiveness. Increased effectiveness of the health networks will result in improved capabilities to reduce the effects and possible spread of antimicrobial resistance in the Netherlands. Which in turn will result in fewer fatalities due to antibiotics resistance.
1.4 Structure of the Research
This thesis consists of 7 different chapters. The introduction provided us with the first chapter. Since the health network Holland West is a new network, the second chapter has been devoted to providing a detailed understanding of the origin and functions of the ABR health networks. This chapter will explain the origin, the actors, the tasks and the differences amongst the health networks in the Netherlands. The third chapter contains a literature review on network effectiveness and a theoretical framework that will be used as a basis for the research. We will review theories on network governance and network effectiveness to distill network governance structures and characteristics which are beneficial for the effectiveness of networks. In the fourth chapter, we provide some extra explanation of the health network Holland West. After describing the case we shall apply the theoretical framework to the case to create hypotheses which can later be operationalized into questions for the network participants. The fifth chapter will describe the methodology. We start by giving a short introduction to the methodology that has been applied for this research. The second part consists of the case selection followed by an operationalization, the methods for data collection, the methods for the data analyses and finally we will discuss the validity and reliability of the research. The sixth chapter describes and analyzes the results from the data sources. The seventh and final chapter contains the conclusion where we will draw some conclusions based on the research and provide points for further research.
7
2. Context
2.1 Policy on ABR health networks
The workgroup on health networks, in collaboration with Bestuurlijk Overleg from the Ministry of Public Health, Welfare and Sports (VWS), created a function profile for the health networks ABR with goals for the health networks and the required activities to achieve these goals. The profile is set up to stimulate coordination, uniformity, and collaboration on prevention and actions in the region. It also creates a connection with the national signaling and response structure. The function profile contains 11 goals and 16 actions for the health networks to fulfill. These are shown in Appendix I.
After publishing the function profile, the parties involved showed enthusiasm and wanted to start with setting up the health networks and reaching the described goals. As a base for these regional health networks, preexisting networks were used. The networks of choice were the Regionaal Overleg Acture Zorg (ROAZ) regions. These regions were tasked with providing, mapping, coordinating and preparing intensive care in their region in accordance with Dutch law (Wet Toelating Zorginstelling). They were picked for the health networks because the relevant health organizations were considered to be connected through these regions, making it easier to set-up the health networks. In each health network, a hospital is appointed with the task and responsibility to get all the relevant actors together in a Stuurgroep. These coordinating hospitals are all the University Medical Centers in the Netherlands, the Amphia hospital and the Isala hospital. Within the Stuurgroeps, all the relevant different organizations have to be present to get a close and complete chain (Schippers, 2016; Schippers, 2017).
Bruno Bruin, minister of medical care and sports, recently enacted legislation which provides the health network with a template to which they have to comply if they wish to receive government grants. This enactment describes the requirements for the health networks structure and twelve tasks they have to fulfill to get access to the grants. To be able to claim a grant, a health network also needs to have a certain structure. The health networks are obliged to have a Stuurgroep and a regional coordination team (RCT) and these entities need to be supported by a network coordinator. The Stuurgroep exists out of heads of different health organizations from the region and contains at least one or more professionals from the fields of public health, general practitioner care, hospital care, nursing homes, home care, disabled care, and pharmacy. Each Stuurgroep appoints a chair from its members. The regional coordination team exists out of a coordinator, an epidemiologist or data-analyst and a couple of field-experts from different
8
health organizations. The Stuurgroep and the regional coordination team are tasked with stimulating and aiding the regional actors in fulfilling their tasks. The network coordinator has to work at one of the coordinating hospitals. He or she is responsible for supporting the
Stuurgroep and the regional coordination team (Bruin, 2018). 2.2 Identifying the different actors in the health networks:
Two organizations coordinate and aide the ten health networks during the pilot stage. The Landelijk Netwerk Acute Zorg (LNAZ) and GGD GHOR are commissioned by the Ministry of Public Health, Welfare and Sports to keep an eye on the progress of the health networks and to match the goals of the health networks to the national policy on antibiotics resistance from the Ministry. This entails that they advise the health networks, collect challenges and best practices and bring together as many organizations as possible which could be beneficial for the health networks (Brugmans & Van der Zouwe, 2017).
Healthcare consultancy firm Zorgmarkten (2018) provided a clear overview of the different actors within the health networks. The regional health networks are top-down led by the Ministry and the governmental organizations RIVM (Rijksinstituut voor Volksgezondheid en Milieu). Inside the regional networks, they make a distinction between intramural organizations, extramural organizations, the GGD, and additional actors. Intramural they distinguish between hospitals, nursing homes, mental healthcare organizations with accommodations and residential groups with support or care. Extramural they distinguish between general practitioners, district nurses and home care. As additional actors, they identify labs and pharmacies. The labs are linked to hospitals because they get patients from the hospitals and the pharmacies are linked to general practitioners due to the referrals.
Based on the function profile the health networks ABR received eleven tasks they had to accomplish. The consultancy firm Zorgmarkten found the original eleven tasks from the function profile to be insufficient or too abstract. Therefore they decided to review the tasks and to complement them. This resulted in twelve tasks for the health networks ABR (Zorgmarkten, 2018) These are portrayed in Appendix II. These tasks where then further specified and made into legal requirements to gain access to grants for the ABR health networks. This was done by an enactment from the Minister of Medical Care on the 5th of February 2019 (Besluit vaststelling beleidsregels subsidiëring regionale zorgnetwerken abr, 2019).
9
2.3 Differences between the health networks
The original plan was to start with 5 pilots for the regional health networks ABR. Minister Schippers appointed Groningen/Friesland, Brabant, Rotterdam, Utrecht and Leiden as pilot regions because these regions were already working on collaboration initiatives for antibiotics resistance. After the publication of the function profile for the health networks other regions became enthusiastic, leading to the decision to start with 10 health networks. These 10 health networks are:
- Zorgnetwerk Noord Nederland - Zorgnetwerk Euregio-Zwolle
- Zorgnetwerk Noord-Holland-Oost / Flevoland - Zorgnetwerk Utrecht
- Gelders Antibioticaresistentie & Infectiepreventie Netwerk (GAIN) - Limburgs Infectiepreventie en antibioticaresistentie Netwerk (LINK) - Zorgnetwerk Noord-Brabant
- Zorgnetwerk Noord-Holland West - Zorgnetwerk Holland West
- Zorgnetwerk Zuidwest-Nederland (Vermeulen, 2017)
Although the tasks of these health networks ABR are quite similar due to the legislation and regulation to be able to apply for grants, they do have some differences. The main difference is experience with antibiotics resistance. At first, the minister wanted to start with five pilot health networks. She wanted to start with the regions Groningen/Friesland, Brabant, Rotterdam, Utrecht and Leiden because these regions were already working on antibiotics resistance. The region which was supposed to become the health network Noord Nederland was a special case. This region already cooperated on antimicrobial resistance through a monthly symposium:
REgionale Microbiologisch Infectieologisch Symposium (REMIS). Due to this collaboration,
professionals were able to share knowledge and they were better able to spot and identify microbial resistance at an earlier stage (Schippers et al., 2015; Schippers 2016; REMIS+, n.d.). Other cooperative networks were also created before the health networks ABR were founded. Different organizations in the regions of Limburg, Brabant and Zeeland were already collaborating with Flemish regions in the i-4-1-Health project (I-4-1-Health, n.d.). The health networks in Limburg and Brabant have an additional difference since these networks have the
10
same Kwartiermaker (ABR Zorgnetwerk Brabant, n.d.). The main difference between the now ten health networks is that some regions already had some experience with sharing information on antibiotics resistance or were already part of a network, while others were new to the subject of antibiotics resistance.
11
3. Theoretic framework
3.1 Origin of the network approach
Before scholars started to look at inter-organizational relationships, most of the studies were focused on the interactions of individuals within organizations. This produced a gap in the literature concerning inter-organizational relationships. Levin and White (1961) were one of the first authors who tried to fill this gap. They criticized the intra-organizational view and started to research inter-organizational relationships. They described inter-organizational relations as organizational exchange in an exchange system. Organizational exchange was defined as: "any voluntary activity between two organizations which has consequences, actual
or anticipated, for the realization of their respective goals or objectives" (p. 588). The
antecedent for this organizational exchange was the need for organizations to gain access to resources that wouldn't have been available to them without the inter-organizational relationship (Levine & White, 1961, pp. 585-588). The examination of inter-organizational relations has shown the relevance of inter-organizational patterns, but it is not a network approach. A network approach specifically focusses on the interactions between more than two actors, which wasn't the case with the inter-organizational approach.
O’Toole (1997) was one of the first authors to stress the importance of networks. According to O’Toole the importance and number of networks increased due to the increasing need to solve wicked policy problems. Wicked problems are problems that can’t be solved by isolating them and solving them one-by-one. He stated that new organizational forms like a network should be created to be able to solve these wicked problems. Other factors which increased the importance of networks are the limited possibilities of a single governmental organization, the political demand for inclusion and broader influence, the growing amount of committees and advisory commissions, and the need of managers to work in different fields at the same time to generate the required policy (O’Toole, 1997, pp. 45-47).
Agranoff and McGuire (2001) follow O'Toole by stating the need for an increased importance of networks. They stress that the importance of networks is due to the increasing need for knowledge. To be able to gain as much knowledge as possible, people have to interact with each other to be able to share and accumulate knowledge. They define networks as multiorganizational arrangements which are set up to solve problems that couldn't be solved by an organization on its own. Their research points out the distinction between network management and hierarchical management and states that network management requires
12
additional attention and research to be able to improve its conceptualization (Agranoff & McGuire, 2001, pp. 295-297).
3.2 Performance of networks
Andrews et al. (2010) mention performance and state that it is a very volatile concept in the field of public management. To clarify the concept of performance, they describe it as the production of outcomes or outputs from an organization. Performance is specified as a concept with different dimensions. These dimensions are efficiency, effectiveness, equity, and public satisfaction. Efficiency is the cost per product or service delivered, effectiveness is the number of goals that have been realized, equity entails the even distribution of outputs and outcomes amongst stakeholders. Improvements in one of these performance indicators might have a negative influence on the other. Therefore it remains a complicated task to manage the performance of public organizations.
Policy implementation is often hard because a lot of public policy programs require a joint effort to be successful. Most of the time, two or more actors are required to successfully implement a certain policy. Scholars often use the term governance to relate to this management of collective action. When different actors are working together, a network is born. An emphasis on multiorganizational networks makes it important for managers to pay attention to coordination between different organizations, besides the regular focus on internal responsibilities (Andrews et al., 2010, pp. 2-3). In this research, we have to look at the governance of networks since we try to find out how we can increase the effectiveness of the health networks. Therefore it is important to gain some insight into network governance and effectiveness.
3.3 Network governance and effectiveness
Provan and Milward (1995) were one of the first scholars to research network effectiveness. They recognized a gap in the literature on networks. Prior literature didn't focus on the nonstructural outcomes of whole networks. In prior research, on network issues, network outcomes and network effectiveness were for the most part ignored. Provan and Milward created a model for network effectiveness based on network structure and context. The variables in their model are centralized integration, external control, stability, and resource availability. Their results find that increased integration through centralization, direct external control, stability, and resource availability, lead to higher network effectiveness. However, they mentioned that their research had a lack of generalizability due to the limitations of the cases
13
they used. Moreover, they only measured it at one point in time which could be problematic for their conclusions (Provan & Milward, 1995). After the article of Provan & Milward other authors created new ways to measure network effectiveness (Provan & Kenis, 2007; Kenis & Provan, 2009; Turrini et al., 2010; Provan & Lemaire, 2012). Different authors use different variables to determine network effectiveness. Provan and Kennis (2007) use different models of network governance and Kenis & Provan (2009) use the form of network, type of inception and the developmental stage of the network. Turrini et al. (2010) created an integrated framework for network effectiveness based on earlier research on networks. They used network structural characteristics, network functioning characteristics and network contextual characteristics as variables for network effectiveness. Provan and Lemaire (2012) use a set of network characteristics to determine network effectiveness.
Provan and Lemaire (2012) state that networks often arise because individuals can gain from using a network. They state that a broad network facilitates access to resources, information, support and more. This access to a broad network is often referred to as someone's social capital. The same however can also be applied to organizations. Therefore it is important to study network governance for organizations as well.
Provan and Lemaire (2012) differentiate between two types of networks: egocentric and "whole" goal-directed networks. These network types are illustrated in figure 1 to clarify their structure. In egocentric networks, the focus is on single actors and the dyadic ties this actor has within his or her network and the benefits or costs these ties bring with them. Egocentric networks are often led by the self-interest of organizations, their interest to gain access to new resources, information, support, and other benefits. This form of network is mostly found when we look at private organizations. In contrast whole, goal-directed networks are mostly found within the public sector. This can be explained due to an important aspect of the goal-directed networks. As the name already implies the goal-directed network is centered around a certain goal that the entire network has in common. The goal-directed network approach, therefore, doesn't look at the ties of individual actors, but at all the ties amongst different actors within the network. Provan et al. (2007) describe a network as "a group of three or more organizations connected in ways that facilitate achievement of a common goal" (p. 482). Networks are often very important to solve complicated problems like healthcare, social policies, and emergency response.
14
Figure 1: Egocentric network (a) and goal-directed network (b)
Source: Provan, K. G., & Lemaire, R. H. (2012). Core concepts and key ideas for understanding public sector organizational networks: Using research to inform scholarship and practice. Public
Administration Review, 72(5), 638-648.
Provan and Lemaire (2012) mention a couple of important challenges to working in a whole network. These challenges are a varied commitment to network goals, culture clashes, loss of autonomy, coordination fatigue and costs, reduced accountability and management complexity. First, varied commitment to network goals. Networks are made up out of a lot of different organizational members with different interests. These interests might clash with the interest of the network causing friction as a result. Second, culture clash. The collaboration between different organizations is one of the main strengths of a network model, but this might also cause problems due to cultural differences between the organizations. Third, the loss of autonomy. Due to the interconnectedness of the different organizations within the whole network, organizations lose some of the autonomy they had before they entered the network. This causes the actions of one actor within the network to have an influence on the other actors within the network. Fourth, coordination fatigue and costs. In a network, a lot of different actors have to come to an agreement before an action can be taken. This makes decisions and activities more time consuming and costly. Fifth, reduced accountability. Within a network, it is hard to be certain which actor is accountable for actions. This increases the chance on ‘free riders', people who don't actively participate, but still receive the benefits. The results of a network are a team effort and therefore accountability is hard to establish. The final factor is managing complexity. A manager in an organization that is part of a network has to manage both the internal environment of the organization and the external environment of the network. This makes the management process more complex and time-consuming.
To have an effective network Provan and Lemaire (2012) state that five characteristics are required: involvement at multiple levels, network design, appropriate governance, legitimacy, and stability. Involvement at multiple levels entails that the network goals and
15
interests should be shared by multiple individuals throughout different levels of an organization. In other words, every organization should have multiple individuals connected to the network and believing in the goals of the network. In the best case scenario, an organization will have network participants from different departments and also from different hierarchical levels.
The second characteristic is network design, network design consists of the idea that networks should consist out of both, strong and weak ties. Every kind of network needs a different configuration, but both types of ties are needed to increase network effectiveness. Granovetter (1973) was the first to distinguish between weak and strong ties. He states that there is a paradox where weak ties are seen as producing alienation on the one hand while it provides opportunities to gather knowledge and information on the other hand. The strong ties generate a certain amount of cohesion amongst the members of that particular group, but it also results in a limited amount of knowledge which could have been received from other groups (Granovetter, 1973). The ability to obtain knowledge through weak ties is explained through the concept of "structural holes" by Burt (2004). Burt states that network participants focus on the activities and information within their group. This creates holes of information between different groups in the network. These holes of information are the structural holes. His research shows that if network participants are located in such a way that they bridge structural holes there is a better flow of information and more new insights are generated (Burt, 2004). Figure 2 illustrates this. It shows that in situation one two groups with strong ties are sharing information within their group, but not between groups. Situation two shows that a weak tie can bridge this gap to make sure that information that is shared within the groups can also be shared between the two groups. Resulting in more information and knowledge.
Figure 2. Visualization of Bridging Structural Holes with weak ties
Provan and Lemaire (2012) state that networks shouldn't be integrated into one dense set of strong ties. Instead, networks should be organized in such a way that there is a mix of strong and relatively weak ties to increase a network's effectiveness. In short, a network should contain
1
16
strong ties to improve trust and share information which is already well known, while weaker ties are needed to gain new insights and information.
Third, legitimacy is essential for the functioning of a network and should be high. Provan and Lemaire (2012) make a distinction between internal and external legitimacy. Internal legitimacy is connected to the legitimacy the network participants attribute to the network. Building internal legitimacy can be done by demonstrating the value of the network, developing trust between the network participants, resolving conflicts and creating durable network governance and communication. External legitimacy is mainly related to the legitimacy in the eyes of stakeholders outside the organization. This is can be improved by seeking new members, promoting the network and providing resources (Provan & Lemaire, 2012).
The fourth characteristic is stability. Networks have to be quite stable at the center and they need to be flexible at the periphery. When it comes down to whole, goal-directed networks flexible connections can provide great benefits. These flexible connections allow a network to adapt quickly to external changes or emergencies, and it improves access to new information. However, this flexibility also makes the network less stable (Provan & Lemaire, 2012). Provan and Milward's (1995) research showed that systematic changes that influence the network's foundations result in a less effective network. They state that this results in a dilemma since flexibility is also seen as a positive aspect of networks to gain access to new information and to be able to adapt quickly. Provan and Lemaire (2012) keep this, and the lack of research on network stability, in mind when they state that ideally, a network should be stable at the core and flexible at the periphery. With this ideal situation, a network will be able to get the benefits of both, without the negative consequences.
The final characteristic is the application of the appropriate form of network governance from the ideal types from Provan and Kenis (2007) (Provan & Lemaire, 2012; Provan et al., 2007). Provan and Kenis (2007) define three different forms of network governance and describe the characteristics a network needs to align with one of the forms of network governance. The three different forms of network governance are shared-governance networks, lead organization-governed networks and network administrative networks. The shared-governance network is a type of network where the different members within the network govern the network without a separate or leading governance organization. This type of network can be established formally and informally. Formally by organizing regular meetings with the different network members or informally through uncoordinated actions of the members. In this
17
type of network, the majority often dictates the decision-making process. Lead organization-governed networks are different in the sense that these networks are coordinated and steered by a single member of the network. This member is, therefore, acting as a lead organization. Coordination of a lead organization leads to asymmetrical power and increased centralization within the network. The lead organization delivers administration for the network and facilitates the activities of the network members in an effort to achieve the network goals. Finally the network administrative organization. With a network administrative organization, the network is complemented by a separate administrative organization that is created to govern the network and its activities. Just as with the Lead organization and shared-governance organization models, members also interact with each other. However, this form is more centralized due to the function of broker for the NAO. In contrast to the other members of the network, the Network Administrative Organization doesn't provide its own services but is strictly brought in the network with the sole purpose of network governance. The different forms of network governance are visualized in figure 3.
Figure 3: Three “ideal” types of network governance
Source: Kenis, P., & Provan, K. G. (2009). Towards an exogenous theory of public network performance. Public Administration, 87(3), 440-456.
Provan and Kenis state that the successful adaptation of a specific form of network governance can be based on four key network characteristics: trust, the number of network participants, goal consensus and the need for network-level competencies. The earlier mentioned network governance forms should be applied when the corresponding network characteristics are represented in the network. The different network governance forms with their corresponding characteristics are visualized in table 1.
Each network characteristic can be described in more detail. The concept of trust is based on an aspect of a relationship that shows "the willingness to accept vulnerability, based Shared-Governed Organization Lead Organization-Governed Organization Network Administrative Organization
18
on positive expectations about another's intentions or behaviors". When we look at network-level interaction we need to look at the distribution of trust and determine whether or not trust is reciprocal amongst network members. Provan and Kenis make a distinction based on the density of the trust. They define the density of the trust as the number of people that trust each other within the network. Secondly, the size of the network is determined by the number of participants in a network. The third characteristic, goal consensus, is mainly related to the similarity between an individual's goals and the goals of the network. If there is a broad network-level goal consensus, network participants are often more involved, committed and willing to work together in their network. This doesn't mean that the goals of the network participants should be completely similar. In contrast similarity of goals could, in some cases, lead to conflict, especially when the goals result in competitive pressures. Every form of network governance requires another level of goal consensus. The final characteristic is the need for network-level competencies. This characteristic is based on the need for organizations to form a network. Every organization within a network has its reason to become part of the network. The thing these organizations have in common is that they joined the network to gain or achieve something that they couldn't achieve on their own. The central question here is how the competencies, required for the network-level goals, can be acquired. This question can be answered by examing the nature of the task performed by the network, and the external demands and needs that are being faced by the network (Provan & Kenis, 2007).
Table 1. Key Predictors of Effectiveness of Network Governance Forms Governance
Forms
Trust Number of
Participants
Goal Consensus Need for Network-Level
Competencies
Shared Governance
High density Few High Low
Lead organization Low density, highly centralized
Moderate Moderately low Moderate
NAO Moderate density,
NAO monitored by members
Moderate to many Moderately high High
Source: Provan, K. G., & Kenis, P. (2007). Modes of network governance: Structure, management, and effectiveness. Journal of public administration research and theory, 18(2), 229-252.
19
Based on this connection between the form of network governance and the different network characteristics, we can create four assumptions. First, the effectiveness of a network governance form is influenced by the way it fits with the characteristics that are prescribed for that particular form of network governance. Second, shared network governance will be the most effective type of network governance in networks where there is a high amount of trust density, relatively small amount of network participants, network-level goal consensus is high and the need for network-level competencies is low. Third, a lead organization based network governance requires only a low density of trust, a moderate number of network participants, a moderate to low amount of goal consensus and the need for network-level competencies is also moderate. Fourth, Network Administrative Organization governance. This type of network governance is most effective when there is a moderate trust density, moderate to many network participants, the goal consensus is moderate to high and the need for network-level competencies is high (Provan & Kenis, 2007).
3.4 Roles in a network
If we want to get a proper image of the health network Holland West we have to identify the different roles of the network participants within the network. Brandes et al. (1999) state that network visualization allows the researcher to get a clear overview of the characteristics of the network, improving his communication and explanation of the network. In contrast to matrixes or tables, network visualization also provides a clear overview of the different ties between the network participants in a network. It shows, for example, which participant is connected to the most network participants, which participants can’t reach all of the participants in the network and which participants are closest to others in the network (Brandes et al., 1999, pp. 75-78). To find these relations we will use the centrality measures from Wasserman and Faust (199$). In this research, we shall determine and analyze the Degree Centrality, Closeness Centrality, and Betweenness Centrality.
Degree Centrality can tell us how central a network participant is within a network. This is based on the number of connections a participant has within the network. Degree centrality examines the number of direct connections a participant has to other organizations within the network. For the health network, this would imply that the participant which has the highest amount of connections within the network has the highest degree centrality. Due to their connectedness, we expect that organizations with a high Degree Centrality have relatively much influence in the network and perhaps fulfill a more coordinating role in the network (Provan et al, 2007, p. 484; Wasserman & Faust, 1994, p. 178).
20
Closeness centrality is determined by looking at the length of a network participant's connections to all other participants in the network. A participant with a high level of closeness centrality has relatively short ties to the different participants in the network. This makes it easier for this participant to spread information and resources towards the other participants in the network. These short ties don't only provide benefits for the participant with a high level of closeness centrality. Due to its short ties, it is also an attractive partner for others in the network to get access to information and resources. For the health network this would imply that these participants are relatively popular, well-known, and receive a lot of information from other organizations (Provan et al, 2007, p. 484; Wasserman & Faust, 1994, pp. 183-184).
The final centrality measure we will use for the analysis of the health networks is Betweenness Centrality. When a network participant has a high level of Betweenness Centrality it is considered to be a gatekeeper in the network. It is called a gatekeeper because it is connected to participants who don't have direct ties with other participants within the network. Therefore the spread of resources and information has to pass the gatekeeper before it reaches these participants. This puts the gatekeeper in an important role for those participants who don't have direct links since they rely on the gatekeeper to spread and receive information. The Betweenness Centrality is determined by looking at the mediating effect of organizations within the network. In the health network, an organization with high Degree Centrality might produce a bottleneck effect for the spread of information and it puts the gatekeeper in an important position for the spread of information throughout the network (Provan et al, 2007, p. 484; Wasserman & Faust, 1994, pp. 188-189).
21
4. Application of theory to the ABR health network Holland West
4.1 Health network Holland West
The health network Holland West encompasses the ROAZ area Leiden. This is an area that contains the working grounds of the GGD Haaglanden and the GGD Hollands Midden. Figure 4 shows a map of this specific area. Due to the overlap of the health network Holland West the GGD's have to collaborate within the Network. The Leiden University Medical Center and both of the GGD's are closely working together to create and maintain the health network. Within the network a variety of health organizations are present. The network contains all sorts of health professionals including an internist-infectiologist, a physician-microbiologist, a specialist geriatric medicine, a physician society and health, a general practitioner, an infection prevention expert, and an epidemiologist/data manager (Brugmans & Van der Zouwe, 2017; GGD HM).
Figure 4: Map of health network Holland West
Bron: Brugmans & Van der Zouwe, 2017
The health network Holland West is a mandated goal-oriented network because the different participating organizations are brought together in a network to fulfill a common goal, have to work together to accomplish this goal, and the health network is mandated by the government. Similar to the other nine health networks, the health network Holland West is mandated by the government (Schippers, 2015) and has to fulfill the tasks as specified by the enactment from Minister Bruins to be able to receive grants. Like the other health networks, the health network Holland West has to have a Stuurgroep, a regional coordination team and a network coordinator (Bruins, 2018). The health network Holland West is structured as follows. The Stuurgroep reports to the ROAZ twice a year. The regional coordination team (RCT) develops policy and transmural work agreements and advises healthcare providers and organizations. The RCT contains an internist-infectiologist, a physician-microbiologist, a
22
specialist geriatric medicine, a physician society and health, a general practitioner, an infection prevention expert, and an epidemiologist/data manager. The health network Holland West also has a Kerngroep, which consists of professionals from the different health sectors. They examine the progress, provide input based on their experiences and provide communication between the network and people from their field. The structure of the health network Holland West is illustrated in figure 5, which has been created by the GGD HM (GGD HM, 2017).
Figure 5: Structure of health network Holland West
Bron: GGD HM, 2017
Kenis and Provan (2009) distinguish between the mandated and voluntary inception of networks. Stating that different performance criteria are suitable for different network types. The criteria for voluntary networks are considered to be mainly based on internal legitimacy: network legitimacy, the ability to keep the network active and the willingness of participants to work for the good of the network as opposed to their self-interest. In contrast to voluntary networks, mandated networks are said to be more in need of external legitimacy in their starting phase due to their dependence on external stakeholders. These networks are often created and funded to fulfill a specific set of tasks but might lack internal legitimacy because these tasks have been imposed by an external actor. Kenis and Provan, therefore, propose that different effectiveness criteria, especially internal versus external legitimacy, should be applied to the different types of networks (Kenis & Provan, 2009). For the health network Holland West we can, therefore, expect to find a larger need for external legitimacy due to its dependence on government funds.
Provan and Lemaire (2012) describe a goal-oriented network as a network where the different organizations within the network have a common goal. The health network Holland West can be considered to be a goal-oriented network since the network has a common goal: to reduce the spread of antibiotics resistance through the spread of information and proper
Kwartier-maker ROAZ Stuur- groep RCT Kern-groep
23
communication (Provan & Lemaire, 2012). These goals are further specified by the enactment of Minister Bruins and incorporated into the goals of the health network Holland West (Brugmans & Van der Zouwe, 2017). Since we are looking at a goal-directed network, we should apply a network approach to examine the network of the health network Holland West as a whole.
4.2 Contextualized Hypotheses
To establish the effectiveness of the health network Holland West we have to apply the theories on network effectiveness to it. The previous part of this chapter described how the health network Holland West is organized, structured and whom the main network participants are. In this chapter, we shall generate hypotheses for this research. For each hypothesis, we shall combine a part of theory on network effectiveness with the contextual characteristics of the health network Holland West. This will result in case specific hypothesis which we can test during the research.
So far we know that the health network is a mandated goal-oriented network, but we don’t know which type of network governance is applied to the network and what the perspectives of the network participants are on the health network. We do know that the health network Holland West is not functioning optimally according to the former Kwartiermaker from the GGD HM. To determine if the network can indeed be improved, we will test the contextualized hypotheses.
We use a combination of Provan and Lemaire’s (2012) and Provan and Kenis’ (2007) network characteristics of an effective network to formulate the hypotheses. According to Provan and Lemaire, a higher amount of multilevel involvement will lead to more effectiveness in a network. The health network Holland West consists of higher-level managers in the
Stuurgroep and health professionals at the operational level RCT and Kerngroep (Brugmans &
Van der Zouwe, 2017). If both of these groups are actively participating in practice, the network will be more effective. However, we aren't sure if they are actively involved with the health network. Therefore the first hypothesis is as follows:
H1: Increased multilevel involvement will lead to more effectiveness for the health network Holland West.
Provan and Lemaire’s (2012) characteristic of network design states that a network needs both weak and strong ties to be able to generate trust between the strongly connected network participants and, at the same time, be able to gather new information through the weak ties. The
24
health network Holland West consists out of a variety of health organizations, professionals and groups. We expect to find relatively strong ties amongst those network participants who have to cooperate a lot and had to cooperate a lot before the health network was implemented. These network participants will be colleagues from the same health organization or network participants who work in the same field. Aside from the network participants which worked together before the implantation of the health network we also expect to find that the members of the RCT, Kerngroep¸and Stuurgroep are strongly connected while being less connected to network participants outside this ‘core' of the network. The network participants inside this core can be connected to a wider variety of organizations and network participants via the health network Holland West. This could provide them with an increased amount of weak ties, which allows them to access new sources of information. This assumption of a strong core with possibly few connections to organizations outside the core leads to hypothesis two:
H2: Adding weak ties to the health network Holland West will increase the network's effectiveness.
The third characteristic is legitimacy. We consider legitimacy to be subjective. Therefore we use the concept of perceived legitimacy during this research. The perceived legitimacy of the health network is connected to the opinions of the different network participants about the health network. If they think that it is beneficial and helpful, they will consider it to be legitimate. It is hard to make statements about the opinions of the network participants before the research, due to its subjective nature. However, we do know that the health network consists of health professionals. Therefore we expect that the network participants consider the network's purpose to be legitimate due to the serious threat antibiotics resistance poses to their patients. Assuming that health professionals want to solve health issues we generated the following hypothesis:
H3: The effectiveness of the health network Holland West, can't be increased through an increase in legitimacy since legitimacy is already high.
The fifth characteristic is stability. Stability entails that networks have to be stable at the center and more flexible at the periphery. In the health network, Holland West the center consists of the RCT, Stuurgroep, Kerngroep, and Kwartiermaker. This core is supposed to interact with each other to shape, improve and update the health network. The RCT connects this core to the other organizations within the network. However, we don't know if these groups are stable or if the configuration of these groups changes a lot over time. Therefore we hypothesize that:
25
H4: Increased stability can positively affect the effectiveness of the health network Holland West.
Finally the characteristic of an appropriate type of network governance from Provan and Lemaire (2012). Based on theory from Kenis and Provan (2009) and Provan and Kenis (2007) we expect that a fit between the form of network governance and the characteristics of the network will result in a more effective network. When we look at the description of the health network we find that it is coordinated by the LUMC, GGD Haaglanden, and GGD HM. This would imply that these organizations function as lead organizations within the network. However, another actor has also been added to the network, the Kwartiermaker. His or her job is to coordinate the network and facilitate the interaction processes. The role of Kwartiermaker could be linked to a network with a Network Administrative Organization with the
Kwartiermaker functioning as the Network Administrative Organization. This would imply that
the health network Holland West is a hybrid of a Lead Organization based and a Network Administrative Organization based network. When we take Provan and Kenis’ (2007) key predictors of effective network governance forms we might find that one specific type of network is better suited for the network.
The health network Holland West is designed to stimulate communication and the spread of information about antibiotics resistance between different health organizations. The participating organizations are unable to track their patients across different health organizations without sharing information. This means that the network participants need to form a network to be able to track the status of their patients and to stay up to date of any antibiotics the patients became resistant to and new knowledge on antibiotics resistance. This is described by Provan and Kenis (2007) as the need for network-level competencies. Due to the threat of antibiotic resistance poses, we assume that the purpose of the network and with it, the goals of the network, will be supported by the different network participants. This provides the network with a high degree of goal consensus amongst its participants. The health network consists of a relatively large amount of health professionals. This shows us that it is a network with a high amount of participants, but it could also mean that there is a higher amount of trust since the network consists out of professionals which operate in the same field.
In short, we expect the health network Holland West to have a high need for network-level competencies, a high goal consensus, a large number of participants and a relatively high trust density. Looking at the expected characteristics we would advise the network to utilize a
26
Network Administrative Organization based form of network governance. Therefore we hypothesize that:
H5: The health network Holland West can be more effective by implementing a Network Administrative Organization based form of network governance.
If we combine these network characteristics from Provan and Lemaire (2012) we can generate a conceptual framework. Figure 6 shows the conceptual framework for network effectiveness. This research tries to identify the relationship between the characteristics of effective networks from Provan and Lemaire (2012) and the effectiveness of the health network Holland West. Therefore the independent variables are multilevel-involvement, network design, appropriate governance form, legitimacy and stability, and the dependent variable is network effectiveness. Due to the ambiguity surrounding the concept of network design we have decided to change it to the presence of strong and weak ties. This makes the variable more specific and clear. The variable appropriate governance form is determined by the characteristics of the network. The appropriate form of governance depends on the degree of trust, the number of participants, goal-consensus and the need for network level competencies. These characteristics determine which form of network governance has the best fit in case of the health network Holland West (Provan & Kenis, 2007). The control variables which will be used are the age of the respondents, the function they have, the group they are part of in the network, and the amount of time they are already fulfilling their current position. This allows us to see if there are major differences between ages, different occupations, whether someone is part of the network’s core or not and different durations of employment.
Figure 6 : Conceptual design of network effectiveness
Multilevel-involvement
Presence of Weak and Strong Ties
Appropriate governance form
Legitimacy
Stability
27
Aside from the network effectiveness we also look at the different roles of the network participants in the health network Holland West. If we can determine the roles of the network participants we can see if it corresponds with one of the ideal types of network governance from Provan and Kenis (2007). This allows us to provide advice on whether or not the form of network governance should be changed.
The roles of the different network participants within the health network Holland West will be determined by applying the centrality measures from Wasserman and Faust (1994). When we consider the different centrality measures and apply them to the health network, we come up with a couple of assumptions. First of all, we expect the Kwartiermaker to have the highest degree centrality in the network due to its coordinating task and position between the
Stuurgroep, RCT, and Kerngroep. This allows the Kwartiermaker to the largest amount of
network participants in the health network. Therefore the eight hypotheses are as follows: H6: The Kwartiermaker from the GGD HM has the highest degree centrality in the health network Holland West.
Besides degree centrality, we also look at closeness and betweenness centrality. Looking at the health network we expect to find that the network participants which are represented in the RCT will have the highest closeness and betweenness centrality. The highest closeness centrality because they are connected in the RCT, the higher level members in the Stuurgroep, and their professional fields to which they have to communicate. The RCT members will also have the highest betweenness centrality because they connect the organizations from their respective professional fields to the other network participants and therefore act as a gatekeeper for access to their respective field. When we consider this the final hypothesis becomes:
H7: The RCT members have the highest level of closeness and betweenness centrality in the health network Holland West.
28
5. Research design and data collection
This research is created to measure the effectiveness of the health network ABR Holland West and to determine if there are points of improvement for the health network. To measure the effectiveness of the health network, the literature on network effectiveness is combined to determine the indicators of the network's effectiveness. We generated the hypotheses based on the indicators from Provan & Kenis (2007) Provan and Lemaine (2012) and Wasserman and Faust (1994). These hypotheses will be tested to determine the effectiveness of the network and whether or not there is room for improvement. In this chapter we discuss the methodology, explain why we picked the case of the health network Holland West, illustrate the operationalization of the concepts, portray the methods of data collection and analysis, and finally, we will evaluate the validity and reliability of the research methods which are applied.
5.1 Methodology Research strategy
This research is a single case study about the effectiveness of the antibiotics resistance health network Holland West. We decided to utilize a case study due to Yin's (2003) considerations for picking a case study design. He states that a case study design can be used in the research is created to answer a "how" or "why" question, the behavior of the subjects of the research can't be influenced, you want to include contextual conditions or if the boundaries between the phenomenon you want to study and the context are unclear (Yin, 2003). The case of the health network Holland West can be linked to Yin's considerations since we utilize a how question, the behavior of the network participants can't be controlled and the importance of contextual factors for the case.
This research is explanatory because we try to identify the causal mechanisms that make the health network Holland West more or less effective. According to Toshkov (2016) explanatory research seeks to identify the causal effects. He mentions that it is important to specify whether the research is prospective or retrospective, whether it is about a specific case or an attempt to make generalizations, and whether the goal is to provide a full or partial explanation. We shall shortly mention the nature of the research concerning each of these topics. In this research, we try to identify the causal relationship between the presence or absence of network characteristics of an effective network and the actual effectiveness of the health network Holland West. It is retrospective because the research is based on documents and interviews. The documents consist of minutes from past meetings of the RCT and Stuurgroep.
29
Therefore they only provide detailed information about past events. During the interviews, we ask respondents to think back in time to when the health network wasn't implemented, and we ask their opinion of the network's development. This entails that the respondents have to recall how the network used to function and to compare this to how it is currently functioning. The utilization of these two data sources and the need to compare over time makes this research retrospective. As mentioned earlier we utilize a specific case that is focused on the health network Holland West. This makes it hard to generalize the findings of this research to other health networks. However due to the similarity of the networks generalizations might be possible, but a broader comparative study is necessary before we can draw such conclusions.
Toshkov (2016) also mentions that a causal relationship can either be complete or partial. Complete explanatory theories try to identify all the different causes which led to a certain outcome, while a partial explanatory theory tries to identify a relationship between a limited amount of phenomena. Based on this definition we can conclude that in the case of the health network Holland West we utilize a partial explanatory theory. We apply a select group of indicators to find causal explanations for the effectiveness of the network. Since we focus on these indicators and leave out others, we apply partial explanatory research. In short, this research is, retrospective, specific and partial.
An explanatory research design tries to find the influence of variables that can be manipulated and controlled. If this research can find causal relationships, as opposed to associational, these can be used to change the health network This allows the researcher, or a manager to manipulate these variables to change outcomes (Toshkov, 2016). Since this research tries to find points of improvement for the health networks explanatory research is rather fitting. The results from the explanatory design might guide the managers of the health network to make their network more effective.
5.2 Case selection
As mentioned earlier, and specifically in chapter 4, this research focusses on the antibiotics resistance health network Holland West. The ABR health networks in the Netherlands are mandated by the government to stimulate health organizations to share information and to facilitate the spread of information throughout the network. The origin of these two tasks is specified in the enactment from Minister Bruins, which states the required goals, tasks and structure of the health networks. Due to the enactment and some background research, on the
30
origin of the different health networks, we consider the health networks to be similar enough to select a case on a more pragmatic basis. We came to this conclusion due to the legislative requirements for the ABR health networks and the fact that the main difference between the health networks was that some health organizations were already sharing information on antibiotics in one way or another.
The network of choice is the ABR health network Holland West. This health network already had some experience with antibiotics resistance and is located in Leiden. Aside from the fact that it's conveniently located in the neighborhood of Leiden University, it is also an attractive network because we already have some connections with organizations in this health network. Before this research, we had already established some connections at the Leiden University Medical Centre (LUMC), the coordinating hospital of the health network, and the GGD Hollands Midden, which provides the Kwartiermaker for the coordination of the network. These pre-established connections are likely to generate a higher response rate, resulting in a more accurate measurement of the networks effectiveness and higher validity of the research.
5.3 Operationalization
According to Toshkov (2016) operationalization is the translation of relatively abstract concepts and the corresponding attributes into concepts that can be measured and classified based on observations. The concept of network effectiveness is rather abstract and the indicators to measure network effectiveness are as well. Therefore we have to operationalize network effectiveness and its indicators to be able to measure and classify the data from the documents and interviews. In this study, we try to measure effectiveness based on the concepts of network effectiveness from Provan and Kenis (2007) and Provan and Lemaire (2012). We use involvement at multiple levels, network design, appropriate governance, perceived legitimacy and stability as indicators for network effectiveness. We operationalize these concepts, based on earlier research in the field of public administration. Aside from these concepts we also have to operationalize network effectiveness itself to be able to state whether or not the network is effective. The operationalization will be used in the interviews to measure and classify the information.
Multilevel involvement
Multilevel involvement entails that people at different levels in an organization are participating in the network. Provan and Lemaire (2012) acknowledge that there is a limited amount of research on this topic. However, they state that it's closely related to the concept of multiplexity.
