The path dependency theory: analytical framework to study institutional integration. The case of France

Background The literature on integration indicates the need for an enhanced theorization of institutional integration. This article proposes path dependence as an analytical framework to study the systems in which integration takes place. Purpose PRISMA proposes a model for integrating health and social care services for older adults. This model was initially tested in Quebec. The PRISMA France study gave us an opportunity to analyze institutional integration in France. Methods A qualitative approach was used. Analyses were based on semi-structured interviews with actors of all levels of decision-making, observations of advisory board meetings, and administrative documents. Results Our analyses revealed the complexity and fragmentation of institutional integration. The path dependency theory, which analyzes the change capacity of institutions by taking into account their historic structures, allows analysis of this situation. The path dependency to the Bismarckian system and the incomplete reforms of gerontological policies generate the coexistence and juxtaposition of institutional systems. In such a context, no institution has sufficient ability to determine gerontology policy and build institutional integration by itself. Conclusion Using path dependence as an analytical framework helps to understand the reasons why institutional integration is critical to organizational and clinical integration, and the complex construction of institutional integration in France.


Introduction
A large part of the scientific literature on the integration of care and services for frail older adults explores and analyses clinical and organizational interdependencies [1,2] because of the multidimensional, chronic and changing nature of gerontology problems. These authors conclude that the keys to successful clinical and organizational integration lie in the national healthcare system in which the integrated networks operate: "For an integrated system of care to function over the long-term, systemic integration is necessary. To achieve this, the organizing principles across the entire system of care must be consistent with the dynamic of local initiatives" [2, p. 46]. What is needed is to have 'policies at the healthcare system level that foster the organizational and professional development that is critical to the success of these initiatives' [3, p. 253].
These authors indicate the need to shape institutional integration, in the sense of the institutionalism theoretical framework. In this theoretical approach, institutional integration would be translated by common values, standards and rules across the authorities that regulate the healthcare and social services system. In this way, the institutionalism approach makes it possible to think simultaneously about normative integration and functional integration, two components of systemic integration [2].
From a qualitative study of the experimental integration project in France (PRISMA France), we chose to use path dependency theory as a theoretical framework to analyze the difficulties arising from institutional fragmentation. We used path dependency theory as a theoretical framework to conceptualize the political and social roots of the different clinical, organizational and institutional actors participating in the construction of a public integration policy in France. This theoretical framework seems to us useful to analyze the institutional problems with integrated networks. The main objective of this paper is to document the usefulness of this approach in the French context and the interest of using this framework to analyze the context of other countries.
First we present the organizational and institutional context in which the construction of a public policy of integration based on the model PRISMA was launched. Then we describe the analysis framework and the data studied to analyze the problem of institutional integration. After the presentation of our results, the discussion will focus on the interests and limitations of the path dependency theory in this analysis.

Background: structure of the gerontology field in France
The services involved in maintaining individual autonomy at home include medical and related services, home support services, and legal aid services. Other services include those that provide temporary facilities away from home (short-and medium-term stay hospitals, day hospitals and day centers), as well as memory clinics and leisure centers. In addition to these types of assistance and care, there are services that assess the need for these services and determine the costs to be charged to the recipients. Information, orientation, and the clinical and tactical coordination of all these services and teams are mainly under the responsibility of local information and coordination centers and health networks.

Fragmentation of services and the welfare system in France
In France, as in some other developed countries, home care services for frail older adults are quite fragmented [4][5][6]. This fragmentation of services can be seen at four levels: between the healthcare and social services sectors, between community and hospital workers, between the public, private-for-profit and private-nonprofit sectors, and between home and institutional environments. With these organizational divisions, the clinical coordination of home care services for frail older adults is difficult and requires a multidimensional and inter-sectoral approach. The concrete result is an organization of gerontological services at home in which: The gerontology sector in France can also be characterized by the diversity of the procedures regulating home services of various government authorities: National Social Action Branch in the social work sector, National Health Branch for public health, National Hospital and Healthcare Organization Branch for healthcare services, and National Social Security Branch for health insurance and old age insurance funds. Each of these national authorities has institutional prerogatives with respect to the definition, implementation and evaluation of policies for home care services for frail older adults. The result is an institutional structure that differs with each authority and is based on a sector-based and functional approach.
Another feature of the system stems from the levels of governance of gerontology sectors. The state, the region, the department 1 and the municipality are all involved in home care services for frail older adults and have their own standards and administrative rules. The overlapping of authority results in structural complexity and a variety of normative systems.
In this context, the development of a clinical and organizational coordination policy started in the 1980s. In the 2000s 'local information and coordination centers' and 'health networks' were set up, which improved coordination significantly. However, their fields of action are still fragmented, i.e. mainly social for the former, which come under the National Social Action Branch, and mainly health-related for the latter, which come under the National Hospital and Healthcare Organization Branch [7].
The overall result of these different elements is that gerontology services come under many regulatory and funding authorities in two different institutional welfare systems, namely social insurance and territorial action. The former has prerogatives in health insurance and old age insurance while the latter has jurisdiction over the territorial entities (region, 'department' and 'municipality').
Criteria for access to benefits (work-related: con-1.
tributory benefits; citizenship: universal benefits; or need: benefits depend on available resources); Nature and amount of benefits (cash or kind, fixed 2. rate or proportional); Funding method (income taxes or social contri-3. butions); Decisional, organizational and management struc-4.
tures of the organization delivering the benefits (centralized, decentralized, or delegated to social partners or to for-profit or non-profit organizations).
Based on these four dimensions, three institutional logics can be identified: A social insurance logic, where the objective is to 1.
protect against the risk of loss of income. Benefits are funded by salary contributions and are paid to contributors. This is the case with healthcare refunds in France and pensions paid to seniors. A social assistance logic, where the objective is to 2.
create solidarity between individuals and combat poverty. It consists of ensuring a minimum income, which does not necessarily cover a specific risk.
Payments are conditional on resources (and not on prior contributions). This is the case with the minimum old age benefit in France.
A universal safety net logic, which aims at covering 3.
certain categories of expenses for certain population profiles. Benefits are not conditional on contributions or resources, but are the same for a whole group of individuals. This logic includes the 'personalized autonomy benefit', which covers the costs of elder care and services based on different levels of loss of autonomy.

Description of policy development
To address the need for coordination in the gerontology field and based on international experiments, a pilot project was launched in France to implement the PRISMA (project and research of integration of services to maintain the autonomy) model [5]. It is a model theorized and developed in Quebec, Canada, using an implementation study and an impact study. This model is an evidence-based model regarding the prevention of autonomy loss, user satisfaction and empowerment, and financial considerations [12].
The PRISMA model includes six tools and mechanisms [13]: Coordination at all levels: strategic (regulatory and 1. funding agencies), tactical (service managers) and clinical (home care workers); A case management process led by a professional 2.
in charge of the needs assessment, planning and coordination of services; A single entry point;

3.
A common tool for assessing older adults' medical 4. and psychosocial needs; A standardized service planning tool; and 5.
A shared clinical file.

6.
This integration model involves the different organizations in the same territory in establishing systematic coordination systems to ensure the continuity of home care for frail older adults. To construct a public integration policy based on the PRISMA model, the participating organizations combine some of their resources, jurisdiction and prerogatives, thereby moving their respective boundaries [14,15]. How the functions of these six integration components are operationalized is determined by a development process that is both horizontal (co-construction at national, regional and 1 In France, the department is an intermediate territorial area between the region and the municipality. There are 36,000 municipalities, 100 departments, and 26 regions. International Journal of Integrated Care -Vol. 10, 30 June 2010 -ISSN 1568-4156 -http://www.ijic.org/ local committee levels) and vertical (two-way channel between the committees to ensure the tools and procedures are relevant and legal). In principle, with this approach it should be possible to implement an integrated network in different service contexts [14,15].
The adaptability and flexibility of this model are other elements explaining the choice of PRISMA for building an integrated network in France.
The French experiment was launched by the Ministry of Health and the National Solidarity Fund for Autonomy in 2006 after consulting all authorities with institutional jurisdiction (centralized and decentralized government agencies, National Solidarity Fund for Autonomy, Health Insurance and Old Age Insurance Funds, regions, departments, municipality). The pilot project was assigned to an independent project team, supported by a multidisciplinary research team. It was conducted on three sites deliberately contrasted in terms of the delivery of care and services and population density. This pilot project was funded with the aim of documenting the possibilities for the French system of healthcare and social services for frail older adults to move towards an integrated system. For the funding agencies, therefore, it was a matter of deploying a small-scale integrative policy with a view to planning a larger-scale public policy. With this aim in mind, the pilot project was deployed in three types of areas (rural, urban and mega-urban 2 ) in order to do a comparative (between-site) and overall (with the French healthcare and social services system as a whole) analysis.
The Canadian implementation study used an innovative tool: the penetration rate of the integrated model in the system [14]. The implementation study in France adopts and adapts this methodology [16]. This measurement relates to the functions of the six tools and mechanisms of the model. In the first 18 months of the experiment, this evaluation of the penetration rate showed similar progress at all three sites, between 5% and 20%. During this first period, the pre-implementation phase, the committees work on the definition of the nature, structure and goals of case management, the targeting method, the profiles and operational aspects of case management (location, supervision, etc.) and the tools for assessment, care planning and monitoring. After this period, the start of the case management process accelerated the implementation of the tools and mechanisms. In January 2009, the first site which began the experiment had three case managers in place. The implementation rate was around 50%. In the other two sites, where the experiment started 12 months later, case management was not active yet.

Problem statement
The PRISMA France pilot project provided an opportunity to analyze the institutional dimension of the integration both as a condition for the deployment of a public integration policy and as a factor that explains the difficulties encountered.

Theoretical framework
The approach we took to the analysis of institutional integration is based on 'path dependency' theory [10].
The path dependence analytical framework is part of the school of thought of the historical institutionalism [17], which considers institutions as structural variables from which stem arrangements of ideas, interests, and powers. They are the focal point of the activity of public policies, in the sense that institutions contribute to structuring them by encouraging or constraining the organizations and their actors and thus their activities.
Path dependency theory starts from the premise that organizations and actors are part of institutions that structure and channel their behavioral standards and activities along established paths. These paths are made up of institutions (with their values, standards and rules) and public policies determined by previous choices that impose constraints on institutional development processes [18]. Thus the notion of dependence in relation to the path taken highlights the historical dynamic that dictates that once a path is chosen, it is difficult to change it because the processes become institutionalized and are reinforced over time [19]. It becomes increasingly difficult to reverse past institutional choices because not following the rules and standards established by previous choices (exit option) generates 'costs' in terms of investment, learning, coordination and anticipation [20]. That is why existing institutions are usually modified and not replaced despite their less than optimal nature [20], and institutional inertia is generated [19].

Methodology
In this article we use the data collected during the implementation programme of the PRISMA France pilot project. Several types of data, collected between June 2006 and February 2009, were used: A large number of semi-directed interviews with the 1.
participants in the three pilot projects, regardless of their level of jurisdiction: national level (n=10), authorities responsible for • • defining gerontology policies, 2 The word 'mega-urban' refers to the concept of megalopolis. It involved studying a site in a megalopolis-type configuration, but without considering the entirety of the area thus defined. regional and departmental levels (n=33), orga-• • nizations in strategic and/or financial positions related to gerontology services, and local level (n=29), decision-makers responsible • • for delivering gerontology services and for managing clinical teams.
These in-depth individual interviews were based on an interview grid designed to identify the factors modulating the implementation of the PRISMA model at the three sites. Transcripts of these interviews were coded with NVivo. The essential themes and coding categories used here were derived from a principally emergent analysis of discourse. They were related to the 'impact of' or the 'need for' institutional integration ('governance of gerontology policies', 'objects of controversy', 'representations of integration', 'forward-looking dimensions').
Direct and participants' observations at coordina-2.
tion meetings at the following levels: national (n=52), in a steering committee (led by • • the project team), including the national authorities of the experimental project, regional and departmental (n=30), in a strategic • • committee (led by the project team), including the authorities with responsibility for regulating the services, and local (n=26), in a tactical committee (led by the • • project team), including all the managers in charge of the gerontology services provided in the territory.
Political/institutional documentation (legislation, 3. programming, statistical documentation). According to the recent works explaining the French welfare system in general and the elder support sector in particular, we tried to document its characteristics in the real context of implementing the pilot project.
First, the examination of the political/institutional documentation provided the data for the characterization of the French healthcare and social services system using the path dependence theoretical framework.
Here we tried to identify which elements in this theoretical framework could link the past and the present in the French healthcare and social services system to its capacity to move towards the integration.
Second, the analysis of the interviews provided an illustration of the consequences of the institutional characteristics as factors that explain the complexity of moving towards integration for the French healthcare and social services system.
More generally, the results presented here were generated by triangulating these three types of data. Thus it was from an inductive analysis of the interviews and participants' observations that the theme of institu-tional integration emerged as a necessary and central (but not sufficient) condition for organizational and clinical integration. The comments cited were selected because they illustrate what was said from the viewpoint of an overall understanding of the French system, which enabled us to do an in-depth, cross-analysis of these three types of data.

Path dependence in the gerontology field in France
Historical approach: one founding and two reforming paths The institutionalization of welfare in France between 1945 and 1970 produced a so-called 'neo-corporatist' historical compromise, that is to say, applying Beveridge's principle of universality with Bismarck's insurance-based methods [21]. More specifically, it involves honoring rights acquired through work (contributory and proportional benefits, organized by risk) stemming from the Bismarckian method (by sector of productive activity, rights are defined by tripartite negotiations between the state, employer representatives and workers' representatives), while ensuring universal coverage of services by a series of mechanisms to compensate for lost income for all types of populations. In reality, right from the start this compromise laid the foundations for a fundamental fragmentation of the French system.
This founding path of the welfare system based on work contributions depends on the economic context and exogenous shocks and went through some major changes in the 1980s, 1990s and 2000s [19][20][21][22]. Budgetary pressure resulting from the full employment crisis in the 1980s gave rise to a long series of recovery plans. In the mid-1990s, the French government introduced social security funding legislation and created new welfare benefits financed by income taxes. This led to a strengthening of the assistance aspect of the welfare system and made the insurance aspect more independent [21]. Thus a first reformative path in the 1990s can be identified: the conjunction of the previously used Bismarckian methods with the Beveridgian methods. One of the basic features of this reforming path was the introduction of a link between taxes and benefits (the 'general social contribution', for example) and the creation of assistance benefits (particularly universal health coverage). At the same time, another competing reforming path was developing, inspired by the 'new public governance' [3,9,23], to develop benefits more directly targeting frail populations. To regulate these evolutions, the government agencies implemented new contractual instruments. These instruments were deployed across  [19,21,22].
Thus, from a historical analysis using the path dependence theoretical framework, today's French welfare system is the product of the conjunction of a founding path (largely Bismarkian, although ambiguous) with two reforming paths (Beveridgian and new public governance). According to this characterization, the French system is a 'multi-path' system.

Situation today: regulation and institutional fragmentation
In parallel, laws to decentralize jurisdiction, recently acquired by government agencies, to territorial entities (region, department and municipality) reinforced the deployment of the contractual instruments of the new governance policy.
The political/institutional data examined show how the decentralization legislation resulted in a series of contracts transferring jurisdiction and budgets (gerontology programs under departmental jurisdiction, healthcare plans under regional jurisdiction). These contracts are deployed: Within each territorial entity (region, department • • and municipality); Between the territorial entities (with the regions hav-• • ing the main jurisdiction over health services and healthcare infrastructures, the departments over social services and medico-social infrastructures, and the municipalities over discretionary social assistance dependent on municipal political decisions); Between the territorial entities and decentralized gov-• • ernment agencies (agencies of the ministries having authority on health and social services in particular); Between the territorial entities and the health and • • old age insurance funds.
These, often juxtaposed, contractual procedures are not unified by framework agreements or general constitutional statements [22]. They also come under different legal codes (social action and families, public health, territorial, social security, mutuality, public service, administration, criminal and general code for regional entities). The result is a wide variety of organizational and standards and values relating to care and services for frail older adults.
In addition, these contractual instruments are expressed through a wide range of intervention logics for regulatory purposes deployed by national and territorial public authorities. These instruments are based on mechanisms that are competitive (tendering procedures) or non-competitive (authorization), coercive (accredita-tion, rate-setting, etc.) or incentivized (tendering procedures, labelling and certification in particular) [24]. The result is that many instruments [25] are used to regulate gerontology services. Resultant of an overlapping of legislative and regulatory interventions, the choice of intervention instruments was made 'program by program', using a sector-based approach and not a populationnel-based approach.
Thus the institutional fragmentation that characterizes the French 'multi-path' system resulted in an accumulation of disjointed programs and sometimes contradictory jurisdictions.

Effect of path dependence in France on a pilot project like PRISMA
The diversity of contractual and regulatory forms and procedures generates a wide range of standards and guidelines as well as potential divergences in orientation. These divergences, which are illustrated below, take into account the complexity of the construction of collective rules for the development and implementation of coherent and integrated gerontology public policies.
Home services for frail older adults come under two institutional systems: the social insurance system on the one hand and the territorial intervention system on the other, which combines the logics of universal and assistance coverage. There are institutional factors specific to each of the two welfare systems: The social insurance system provides contributory • • and universal benefits: "Home care for every retiree from the general plan […] which considers the need to deal with all acquired rights." while territorial intervention tries to adapt the program to the socio-demographic characteristics of the territories and fund them through the national tax system: "The program is positioned to keep the personal allowance for autonomy mechanism more favourable [financially for the old persons] in our department" (this comment was related to decisions concerning the application of new rules for this allowance). The social insurance system uses a sector-and • • category-based approach: "Social assistants from the National Old Age Insurance Fund deal with older people. If they are very sick, older people are also dealt with by social assistants from the Primary Health Insurance Fund because they are sick, regardless of whether they are elderly." while territorial gerontology policies take a transversal and global approach: The social insurance system is organised as a cen-• • tralized pyramid: "The National Old Age Insurance Fund is a special case in that it is both the National Fund with hierarchical and functional control over the 16 Regional Funds in France, and also the [Regional Fund] for the Paris area." while territorial gerontology policies are multilevel and come under specific territories: "The desire to be close to people, by having social assistants in charge of people over 60, in territories that cover a canton. We already had this idea that working close to them was important for this population." The social insurance system is a 'management • • administration': "I have to implement the Social Action policy and one of the objectives we have in the objectives and management agreement is that to implement it we go through the Social Action managers?" while territorial gerontology policies must be implemented through 'mission administration' and be evaluated: "Present us with a project in an area that you think is a priority, show us why it is a priority, ask us for the money for what you want to do, and we'll give you the money." In the gerontology field, the juxtaposition of welfare system logics also explains some of the institutional inertia. Institutional inertia has been evident for 30 years with respect to financial support between two options: the creation of a 5th risk-'risk of dependence'-in social security (insurance logic) or the territorial piloting of a 'dependence benefit' (universal logic). The incremental process favored departmental management to the detriment of social security agencies in 2004. However, in 2008 the government relaunched national consultations regarding the creation of this 5th risk, and debated the decentralization legislation that made departments the 'leaders' of gerontology policies: This context of contradictory logics, and efforts to find a balance between the action logics, make it hard in France to define the nature of a 'pivot' public authority of gerontology policy. No one authority has the sole prerogative for defining and regulating the implementation of a home care policy for older adults. This finding leads to the conclusion that institutional integration is an essential (if probably an insufficient) prerequisite for the deployment of a public integration policy. During the pilot project numerous advances were attributed to the initial joint commitment, weak as it was, of all the agencies with authority over the definition of public policies.

Discussion
We used a path dependence analytical framework to understand the foundations of the construction of the institutional integration inherent in public policy pilot projects for the integration of care and services for frail older adults. Path dependence provides heuristic value from three perspectives.
First, specifically in the PRISMA France pilot project, the analysis of institutional development processes led to a characterization of the French healthcare and social services system in terms of institutional fragmentations which are reinforced over time. This characterization sheds light on the conditions for implementing a service integration public policy. Compared to the Quebec experience, it provides a certain amount of information regarding the initial low level of organizational and clinical integration and the slower progress of the implementation of the integration [14,16].
Second, this first finding could lead, to the view that a 'single path' healthcare and social services system, like the universal Beveridge inspired system in Quebec, would be easier to integrate than a 'multi-path' system as is the case in France. This is in agreement with the insight of other authors [14]. Studies comparing national cases should be conducted for countries with very different welfare systems such as the United Kingdom, Germany and the Netherlands.
In fact, and this is the third proposition, an analysis of the institutional system, including from a path dependency theory's perspective, could enrich an examination of all integration projects from the standpoint of the conditions for success. This approach sheds light on the foundations of the institutional options chosen to construct and lead the change towards organizational and clinical integration. Are some integration models more adapted to Bismarckian, Beveridgian or liberal welfare systems? To 'single path' or 'multi-path' contexts? We think this opens up a broad research field.
In the path dependency theory, actors are hemmed in by existing institutions and structures that channel  [26]. Thus, to address all the aspects of normative and functional integration more extensively, it seems necessary to take other dimensions into account. We are thinking in particular of the meticulous analysis of the conditions of economic conditions [19,26] or the in-depth study of the various forms of advocacy coalition networks in the policy-making [27].

Conclusion
The path dependency theory seems an appropriate approach to analyze the systemic dimension of integration. This theory could be a new way to study national healthcare systems in both their policy and incentive dimensions as well as their administrative and coercive dimensions.
In the case of France, according to path dependency theory, our results reveal two institutional systems (the social insurance system and the territorial intervention system, with the latter combining the universal and assistance logics of welfare). This structural complexity requires a willingness among all authorities with gerontology responsibilities to work toward the integration of services. In this complex and uncertain context, it is not a question of going back to the traditional 'top down' way of constructing public policy, whose limitations in deployment in organizations and practices have been widely documented. It is a matter of framing the 'bottom-up' deployment of service integration more clearly.
When France implements an integration policy across its entire national territory (essentially through the Regional Health Agencies, and Homes for the Integration and Autonomy for people suffering from Alzheimer's or associated disorders based on PRISMA model), it is essential to bear in mind that the success of such public policies cannot be assured without the deployment of a strong and shared desire for institutional integration and cannot depend solely on local actors.